Efficacy of Antifungal Agents in the Cornea

/. A Comparative Study

Denis M. O'Day, Richard Robinson, and W. Steven Head

A standardized model of Candida albicans keratitis was developed in pigmented rabbits using a quantitative mycologic technique to evaluate the disease at intervals throughout the course. In this model, using two different infecting strains, the efficacy of five antifungal agents was compared. Amphotericin B, in concentrations of 0.5% to 0.075%, was superior to all other agents tested. Na- tamycin 5% ranked next, followed by 1% flucytosine, and 1% miconazole. Ketoconazole 1% was ineffective. Invest Ophthalmol Vis Sci 24:1098-1102, 1983

The selection of appropriate therapy for fungal cor- (minimal inhibitory concentration [MIC] > 50 /xg/ neal infection remains an unsettled question. Al- ml), whereas strain F357 is resistant to flucytosine though there is considerable clinical evidence to sup- (MIC > 50 iig/m\) but susceptible to ketoconazole port the use of certain agents,1"6 objective experi- (MIC = 0.10 Mg/ml). The discrepancy between the mental data is still lacking because of the difficulties inhibitory and fungicidal values with these agents is associated with the development of a suitable animal indicative of a fungistatic rather than a fungicidal model of fungal disease. In this paper we describe a mechanism of action. In contrast, the levels with am- model of Candida albicans infection in the rabbit eye, photericin B suggest a fungicidal mechanism in vitro. using quantitative mycologic techniques, and its ap- Animals: Pigmented outbred male rabbits, 1.5-3 plication to a comparison of efficacy of a variety of kg were used in all experiments. antifungal agents. Infection protocol: Rabbits were anesthetised with either intravenous pentobarbital (Nembutal) or in- Materials and Methods tramuscular Ketamine and xylazine hydrochloride. A retrobulbar injection of 1% xylocaine was given Development of the Animal Model and topical anesthesia was achieved with Alcaine Inoculum: Two-day-old cultures of C. albicans 0.5%. Using an operating microscope, eight wells, grown on trypticase soy agar with 5% sheep blood one-half stromal thickness in depth, were trephined (BBL) were used to inoculate the rabbit cornea. A in the cornea of each eye in a regular pattern. A glass suspension of the inoculum in normal saline in a trephine prepared from a microhematocrit tube 1 concentration of 5 X 109 per ml was prepared and mm in diameter was used. A second trephine loaded stored at 4 C over night. In preliminary studies only with C. albicans suspension was used to inoculate two of the five strains tested produced significant dis- each well. The trephine was placed in each well, ro- ease and these were used in subsequent studies (Strain tated three to four times and then removed. Each well LV and Strain F357). The susceptibility of these two was inoculated a second time with a new trephine strains to five antifungal agents, using a modification loaded with fresh inoculum. If inadvertent corneal (personal communication, Dan B. Jones, 1980) of the perforation occurred the animal was removed from method of Shadomy,7 is shown in Table 1. In vitro, the study. the LV strain is susceptible to flucytosine at a con- Isolate recovery: At predetermined intervals the centration of 0.30 ng/m\ but resistant to ketoconazole animals were sacrificed with a commercially prepared euthanasia solution T-61 (Taylor Pharmacal, Deca- tur, IL). The whole cornea was removed by excision From the Department of Ophthalmology, Vanderbilt University at the limbus, placed in a sterile petri dish, and cut School of Medicine, Nashville, Tennessee. into small pieces. These were then ground in a tissue Supported by NIH grant EY 01621 and an unrestricted grant grinder (Ultraturex Model SDT) for three 10-sec in- from Research to Prevent Blindness, Inc. Submitted for publication: November 18, 1982. tervals in 3 ml of sterile saline. Ten and 100 n\ ali- Reprint requests: Denis M. O'Day, MD, Room D-5217, Van- quots of this corneal suspension were plated in trip- derbilt Medical Center North, Nashville, TN 37232. licate on trypticase soy agar with 5% sheep blood.

0146-0404/8 3/0800/1098/$ 1.05 © Association for Research in Vision and Ophthalmology

1098

Downloaded from iovs.arvojournals.org on 09/24/2021 No. 8 EFFICACY OF ANTIFUNGAL AGENTS IN THE CORNEA / O'Doy er ol. 1099

Table 1. In vitro susceptibility (/tig/ml) Amphotericin B Natamycin Ketoconazole Flucytosine Miconazole

Strain MIC MFC MIC MFC MIC MFC MIC MFC MIC MFC

LV 0.2 0.39 6.25 >50 >50 >50 0.39 >50 12.5 >50 F357 0.1 1.56 3.12 25 0.10 >50 >50 >50 12.5 >50

MIC = minimal inhibitory concentration. MFC = minimal fungicidal concentration.

After 48 hrs incubation at 25 C the colony forming The following experiments were performed: units (CFU) were counted, and the number of CFU a. Candida albicans strain LV: a comparison of per whole cornea was calculated based on a 3-ml total all five agents (natamycin 5%, amphotericin B volume. 0.5%, miconazole 1%, ketoconazole 1% and flucyto- sine 1%). Evaluation of Topical Antifungal Agents in the Model b. Candida albicans strain LV: a comparison of different dilutions of amphotericin B in distilled wa- Agents evaluated: Polyenes: Amphotericin B 0.5%, ter, 0.5%, 0.15%, 0.075%. 0.15%, and 0.075% in distilled water (Fungizone®, c. Candida albicans strain F357: a comparison of E. R. Squibb & Son, Princeton, NJ) and Natamycin 1% flucytosine, 1% ketoconazole and 0.15% ampho- 5% suspension (Alcon Laboratories, Fort Worth, tericin B. TX). Imidazoles: Miconazole 1% in Cremophor (Janssen Pharmaceuticals, New Brunswick, NJ) and Results Ketoconazole 1% in polyethylene glycol 400 (Jans- sen). Pyrimidine: Flucytosine 1% solution in normal Animal Model saline (Roche Laboratories, Nutley, NJ). Within 24 hours of inoculation a ring of infiltrate Treatment protocol: Treatment was begun 30 min developed around the base of each trephination. This after inoculation. Animals were assigned randomly area of infiltrate grew larger over the next 36 hrs and to a treated or untreated control group. Six rabbit then slowly diminished as the infection resolved clin- eyes were exposed to each treatment regimen with ically over a 5-day period. With both strains the both eyes of each animal receiving either the test drug course of the disease was similar and paralleled the or no treatment. Drops were administered 10 times quantitative isolate recovery rate at the end of each per day at hourly intervals. Eighteen hours after the 24-hr period. During the first 48 hrs the isolate re- conclusion of the treatment period both treated and covery rate was high (Fig. 1), but by 72 hrs it had control animals were killed and the corneas pro- fallen to 10% of the 24-hr value and by the fifth day cessed. it was less than 1%.

CANDIDA ALBICANS - NORMAL DISEASE

Strain LV m Strain F357

Fig. 1. The normal course (r of disease following inocu- O lation with two different o strains of Candida albicans as measured by the number of CFU/cornea. Each data point represents six eyes + (mean and range).

Inoculation 24 48 72 96 120 TIME POST INOCULATION (hours)

Downloaded from iovs.arvojournals.org on 09/24/2021 1100 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / August 1983 Vol. 24

CANDIDA ALBICANS-STRAIN LV

Control i*S] Treated

AM "B"O,5% AM "B"O,I5% AM "B" 0.075% Fig. 2. The response of an infection to differing con- centrations of amphotericin P

24 HOURS 48 HOURS 24 HOURS 48 HOURS 24 HOURS 48 HOURS

DURATION OF TREATMENT

Although the range at each sampling point was Drug Studies quite broad, the course of the disease was consistent with each isolate. Because the most active infection Due to a multiplicative error structure, the re- in the untreated model occurred in the first 48 hrs, sponse variable analyzed was loge (1 + CFU). Prelim- evaluation of the response to drug therapy was re- inary analysis showed a highly significant correlation stricted to this period. between right and left eye results (P < 0.0001) in both treated and untreated groups. Hence, a nested anal- ysis of variance was used to test differences between CANDIDA ALBICANS -STRAIN LV treatments. The error term was between rabbit mean 9 ^Control HJTreoted squares. In the experiment comparing the efficacy of five NATAMYCIN 5% FLUCYTOSINE 1% antifungal agents against the LV strain, there was a marked variation in the response to treatment (Figs.

P

P= 0.2471 centrations of 0.15% and 0.075%. As can be seen in P =0.0370 P=0.7369 P =0.0056 Figure 2, these were also highly effective. After 48 Ifl hours of treatment infection was eliminated in most ifl eyes. The final experiment was designed to study the relationship between in vitro derived susceptibility data and the in vivo therapeutic response. The re- m 24 HOURS 48 HOURS 24 HOURS 48 HOURS sponse to treatment of strain F357 (resistant to in vitro flucytosine but susceptible to ketoconazole) was DURATION OF TREATMENT compared to that of the LV strain (resistant to ke- Fig. 3. The response of Candida albicans infection (strain LV) toconazole, susceptible to flucytosine). Treatment to various antifungal agents as measured by the number of CFU/ cornea isolated after 24 and 48 hrs of treatment. Each data point with amphotericin B was used as a positive control represents six eyes (mean and range). (Fig. 4). Ketoconazole was no more effective against

Downloaded from iovs.arvojournals.org on 09/24/2021 No. 8 EFFICACY OF ANTIFUNGAL AGENTS IN THE CORNEA / O'Doy er ol. 1101

CANDIDA ALBICANS- STRAIN F357

^Control p| Treated

AMPHOTERICIN "B" 0.15% FLUCYTOSINE 1% KETOCONAZOLE 1% Fig. 4. The response of Candida albicans (strain F357) to amphotericin B, = 0.6064 P = 0.4957 P-0.0768 P = O.45I8 flucytosine and ketocona- _ P<0.000l PO.00I0 zole. This strain is suscep- tible in vitro to ketoconazole ffl but resistant to flucytosine. Each data point represents six eyes (mean and range).

24 HOURS 48 HOURS 24 HOURS 48 HOURS 24 HOURS 48 HOURS

DURATION OF TREATMENT

the LV strain than against F357. However, there was present model, using these techniques we have de- a difference in the response of the two strains to flu- fined the natural course of the disease with two sep- cytosine that paralleled in vitro susceptibility. arate isolates. With each, the infection tends to re- solve over a period of 5 days, but during the first 48 hrs there is sufficient disease to allow therapeutic Discussion manipulation. The fact that only two of five isolates The principal topical antifungal agents currently produced adequate disease emphasizes the difficulties available for the treatment of corneal disease have associated with this type of experimental infection. been derived, with the exception of natamycin, from For the efficacy studies, we selected five agents that systemically administered compounds. They are used have been advocated for the treatment of Candida in concentrations that are empirical and appear lim- infections using currently recommended concentra- ited only by toxicity and solubility. Over the years tions, the purpose being to determine the relative ef- there have been numerous attempts to examine the ficacy of the agents. The clear superiority of the poly- efficacy of these antifungal agents in the external eye enes in this external ocular model is remarkable (Figs. but these efforts have been largely frustrated by the 2, 3). Flucytosine, the next most effective agent, pro- lack of good models of disease, as well as a means of duced a smaller but significant reduction in CFU. quantifying the response.9"15 As a result, we continue Most disappointing was the inferior efficacy of mi- to rely on accumulated data derived largely from un- conazole and the lack of any discernible therapeutic controlled clinical studies.1"6 effect with ketoconazole despite its promise as a sys- 18 The development of appropriate models of fungal temic therapeutic agent for Candida infection. corneal disease that incorporate an accurate measure While it must be emphasized that an experiment that of the response would offer an opportunity to ex- demonstrates a cure of a fungal corneal infection amine therapy of fungal corneal disease in a way not within two days does not parallel human experience, hitherto possible. Such models should parallel human the fact that under the same conditions all other disease in terms of infecting organisms and clinical agents failed to match this performance must be con- manifestations and should not require alteration of sidered of some significance. the host response. The disease process should be of Since the LV strain showed in vitro resistance to sufficient duration to permit therapeutic manipula- ketoconazole, it could be argued that ketoconazole tion but not of a degree of severity that would defeat was ineffective for this reason. The strain was also therapeutic intervention. Finally, an objective method resistant to miconazole. The availability of another of evaluating response to therapy must be used. The strain (F357) susceptible in vitro to ketoconazole but application of quantitative techniques to mycologic resistant to flucytosine afforded an opportunity to studies in the cornea offers a more precise way to compare in vitro and in vivo antifungal activity for achieve objective and comparable data.1617 In the these two drugs. Our animal studies demonstrated a

Downloaded from iovs.arvojournals.org on 09/24/2021 1102 INVESTIGATIVE OPHTHALMOLOGY G VISUAL SCIENCE / Augusr 1983 Vol. 24

lack of correlation with the in vitro data for keto- References conazole, with no therapeutic response observable 1. Jones BR: Principles in the management of oculomycosis. against either strain. However, with flucytosine the Trans Am Acad Ophthalmol Otolaryngol 79:OP15, 1975. in vitro susceptibility tests appeared to predict to 2. Jones DB, Forster RK, and Rebell G: Fusarium solani keratitis some degree the in vivo response although it was still treated with natamycin (pimaricin); eighteen consecutive markedly inferior to either of the polyenes. There are cases. Arch Ophthalmol 88:147, 1972. 3. Romano A, Segal E, Eylan E, and Stein R: Treatment of ex- obviously many reasons why a topical drug may fail ternal ocular Candida infections with 5-fluorocytosine. to act, and this study has not examined other factors Ophthalmologica 172:282, 1976. that may affect bioavailability. Such studies are 4. Forster RK, Rebell G, Wilson LA: Dematiaceous fungal ker- needed before the relationship between in vitro and atitis; clinical isolates and management. Br J Ophthalmol in vivo data can be fully evaluated. 59:372, 1975. 5. Wood TO and Williford W: Treatment of keratomycosis with Although treatment with 0.5% amphotericin B amphotericin B 0.15%. Am J Ophthalmol 81:847, 1976. eradicated the infection in 48 hrs, animal and human 6. Foster CS: Miconazole therapy for keratomycosis. Am J experience would indicate that at this concentration Ophthalmol 91:622, 1981. toxicity can still be a problem with prolonged use. 7. Shadomy S and Espinel-Ingroff A: Susceptibility testing with For this reason, we studied the efficacy of the drug antifungal drugs. In Manual of Clinical Microbiology, 3rd ed., Lennette EH, Balows A, Hausler WJ Jr, and Truant JP, editors. at reduced concentrations. Even at a concentration Washington, DC, American Society for Microbiology, 1980, of 0.075%, amphotericin B given topically is superior pp. 647-653. to natamycin, the next most effective drug. In view 8. Snedecor GW and Cochran WG: Statistical Methods, 6th ed., of the overwhelming superiority of dilute amphoter- Ames, Iowa, Iowa State University Press, 1967. icin B in this model and the apparent lack of toxicity 9. Richards AB, Clayton YM, and Jones BR: Antifungal drugs (O'Day, unpublished data) we believe that a re-eval- for oculomycosis: IV. The evaluation of antifungal drugs in the living animal cornea. Trans Ophthalmol Soc UK 89:847, uation of its potential as a topical preparation may 517 1969. be worthwhile. 10. Ishibashi Y: Experimental fungal keratitis due to Fusarium: Although this model has, thus far, offered a more Studies on animal model and inoculation technique. Conci- precise way of examining some of the problems as- lium Ophthalmologicum, 23rd, 1978, Kyoto. 2:1705, 1979. 11. Oji EO: Development of quantitative methods of measuring sociated with antifungal therapy in the cornea, it does antifungal drug effects in the rabbit cornea. Br J Ophthalmol have some limitations. Suitable strains for study are 65:89, 1981. difficult to find so that, thus far, data has been derived 12. Burda CD and Fisher E Jr: The use of cortisone in establishing from only two strains. While in this model of super- experimental fungal keratitis in rats. Am J Ophthalmol 48:330, ficial infection drug penetration is probably not a sig- 1959. 13. Ellison AC, Newmark E, and Kaufman HE: Chemotherapy nificant variable, with deep infection it may influence of experimental keratomycosis. Am J Ophthalmol 68:812, efficacy. Finally, it is not yet possible to evaluate the 1969. relevance of these findings to the problem of fila- 14. Forster RK and Rebell G: Animal model of Fusarium solani mentous fungal keratitis. For these reasons, the over- keratitis. Am J Ophthalmol 79:510, 1975. all applicability of these findings will not be clear until 15. Newmark E, Ellison AC, and Kaufman HE: Combined pi- maricin and dexamethasone therapy of keratomycosis. Am J further experiments have been performed. We do Ophthalmol 71:718, 1971. believe, however, that it is now possible to begin to 16. Osato MS and Jones DB: Effect of topical flurbiprofen on examine some of the questions associated with the experimental fungal keratitis. ARVO Abstracts. Invest therapy of fungal infections in the cornea that have Ophthalmol Vis Sci 20(Suppl):108, 1981. only recently been addressed for bacterial infections. 17. Stern GA, Okumoto M, and Smolin G: Combined ampho- tericin B and rifampin treatment of experimental Candida albicans keratitis. Arch Ophthalmol 97:721, 1979. Key words: Candida albicans, fungal corneal infection, an- 18. Heil RG: Systemic Candidosis and Candidemia. In Ketocon- tifungal agents, amphotericin B, natamycin, miconazole, azole in the Management of Fungal Disease, Levine HB, ed- ketoconazole, flucytosine itor. Sydney, Australia, Adis Press, 1982.

Downloaded from iovs.arvojournals.org on 09/24/2021