Ocular Pigmentation Protects the Rabbit Retina From Gentamicin-Induced Toxicity

Ocular Pigmentation Protects the Rabbit Retina From Gentamicin-Induced oxicity Esther Zemel* Anat Loewenstein,"f Bo Lei* Moshe Lazar,\ and Ido Perlman* Purpose. his study was designed to investigate the possibility that gentamicin-induced retinal toxicity is dependent on ocular pigmentation by comparing the effects of the drug on the functional and morphologic integrity of the retina in albino and pigmented rabbits. Methods. In each rabbit, a solution of gentamicin sulfate was injected into the vitreous of one eye, and saline was injected into the other eye. Retinal function was assessed by electroretinogram (ERG) at different time intervals after injection. Retinal structure was examined at the light microscopic level. Results. In albino and pigmented rabbits, functional retinal damage developed to a maximal level within the first week after gentamicin injection. hereafter, gradual recovery could be seen in eyes that suffered less than 80% maximal reduction in the ERG b-wave. For each dose >0.1 mg studied, retinal damage was more severe in the albino rabbits than in the pigmented ones. he degree of damage was not affected by the level of ambient illumination, nor was it reduced by the administration of N-acetylcystein, a free radical scavenger, together with gentamicin. Conclusions. Ocular pigmentation partially protects the rabbit retina from the toxic action of gentamicin. his protection probably reflects binding of the drug by the melanin, which thereby reduces the concentration of the free gentamicin. When the initial gentamicin-induced retinal damage is expressed in <80% reduction in the ERG, substantial recovery may occur in both strains of rabbits. Invest Ophthalmol Vis Sci. 1995;36:1875-1884. Oecause of their effectiveness against gram-negative and gram-positive bacteria, 1 aminoglycosides are the preferred antibiotics for the treatment and prophylaxis of ocular infections. However, the use of these antibiotics has been limited by their well-documented toxicity to the retina. Numerous studies have been conducted on different animal species to determine the safe, nontoxic dosage of gentamicin for intravitreal injection. A wide range of "safe" dose has been reported that partially could be accounted for by species differences. However, even in the same species, different results were obtained. In one study on albino rabbits, retinal damage was detected when gentamicin in a dose as low as 0.03 mg was injected with the bevel of the needle From the *he Bruce, linppnport Faculty of Medicine, echnion Israel Institute of echnology, and llie Rappaport Institute for Research in the Medical Sciences, Haifa; and the f Department of Ophthalmology, Ichilov Hospital, el-aviv, Israel. Sulnnilled for publication December 28, 1994; revised February 27, 1995; accepted April II, 1995. Proprietary interest category: N. Reprint requests: Ido Perlman, he Bruce Rappaport Faculty of Medicine, echnion-israel Institute of echnology, P.O. Box 9649, Haifa 31096, Israel. pointing to the retina, whereas a dose of 0.2 mg was not harmful when the bevel was pointed toward the anterior chamber. 2 A second study on the same animal species reported that a dose of 0.08 mg gentamicin produced functional retinal damage as evidenced by electrophysiological recordings. 3 In pigmented rabbits, a dose of 0.2 mg of gentamicin was found to be toxic based on electrophysiology and light microscopy. 4 In another study, a dose of 0.5 mg was found to cause structural damage, 5 but it was reported to be safe in a third study.' 1 Others found damage only when a dose of at least 0.8 mg was used. 78 Research conducted in monkeys found 0.4 mg gentamicin to be safe to the retina, as judged by ophthalmoscopy, electrophysiological testing, and light and electron microscopic criteria, 9 whereas others found doses of gentamicin from 1.0 to 10.0 mg to be toxic. 10 " 12 he wide diversity of results in these animal studies did not allow accurate determination of a safe dose for intravitreal injection of gentamicin in human patients. Moreover, species differences may cause a certain dose to be safe in one animal but toxic in another. Investigative Ophthalmology & Visual Science, August 1995, Vol. 36, No. 9 Copyright Association for Research in Vision and Ophthalmology 1875

1876 Investigative Ophthalmology & Visual Science, August 1995, Vol. 36, No. 9 In fact, many incidences of retinal toxicity in humans after the injection of supposedly safe doses are still being reported. 13 ""' It is now obvious that although most patients can tolerate an intra vitreous dose of 0.1 mg of gentamicin sulfate, there may be specific, yet unidentified, characteristics of the eyes of some patients that make them susceptible to injury at this low dose level. 16 he cause of varying thresholds of gentamicin toxicity emerging from animal studies is most probably multifactorial. he facile explanation of species differences will not suffice because wide ranges of allegedly safe doses have been reported in studies of the same animal species. 3 " 8 hus, other variables must be considered. he clinical factors involve the surgical state of the eye, 17 the presence and severity of infection, 8 and the position of the needle bevel. 2 In the animal studies, most of these factors are well controlled so that only those physiological variables that may either augment or reduce gentamicin-induced toxicity need to be considered. he physiological factors include the degree of ocular pigmentation, the level of ambient illumination, room temperature, and stress. Exposure to light, 1819 high temperature, 20 and stress 21 are known to induce retinal damage both by themselves and by augmenting the effects of other retinopathic agents. hey, therefore, also may play a role in gentamicininduced retinal damage. Melanin in the pigment epithelium may play a dual protective role in drug-induced retinopathy. First, it reduces the exposure of the retina to light by absorbing the photons that bypass the photoreceptors. herefore, albino animals are rendered more susceptible to light-induced retinal damage than are pigmented animals. 22 Additionally, melanin is known as a scavenger of free radicals. 23 If gentamicin-induced toxicity is mediated by free radicals, the melanin of the pigment epithelium may provide some protection. he aim of this study was to compare quantitatively the development of gentamicin-induced retinal toxicity between albino and pigmented rabbits to determine the role of ocular pigmentation in this pathologic process. MEHODS Animals Fifty-two albino and 30 pigmented rabbits weighing 1.5 to 3.2 kg each were used in this study. Each experiment was performed in parallel on a similar number of albino and pigmented rabbits of similar weights to minimize the variability in gentamicin toxicity unrelated to ocular pigmentation. he animals were housed in separate cages under 12-hour light-12-hour dark cycles in the Animal Facility of the Faculty of Medicine, echnion-israel Institute of echnology (Haifa, Israel). After the intravitreal injections, most of the animals were returned to the same room on a 12-hour light-12-hour dark cycle. One experimental group of eight albino rabbits was placed in total darkness for 1 week after gentamicin injection to test for a possible interaction between ambient illumination and gentamicin toxicity. All experimental procedures conformed to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Procedure he animals were anesthetized by an intramuscular injection (0.5 ml/kg body weight) of a "cocktail" prepared in the following proportions: 1 ml Ketalar (Parke-Davis, Gwent, UK), 0.2 ml acepromazine maleate, and 0.3 ml xylazine. he pupils were fully dilated with cyclopentolate hydrochloride 1%. opical anesthesia (benoxinate HCL 0.4%) was administered before carrying out electroretinogram (ERG) recordings and before intravitreal injection. he ERG was used to estimate the functional integrity of the retina. he test was performed at different time intervals after intravitreal injection of the drug. After the last elecrophysiological recording session, eyes were enucleated and prepared for histology, and the rabbits were killed by an overdose of sodium pentobarbital. Drugs Gentamicin sulphate (eva, Jerusalem, Israel) at a concentration of 40 mg/ml was used. It was diluted as necessary with physiological saline to produce the desired concentrations. he most concentrated gentamicin solution used in this study (10 mg/ml) had an osmolarity of 255 mosm and a ph of 4.3. N-acetylcystein (Sigma, St. Louis, MO) was used as an antioxidant to test for possible involvement of free radicals in gentamicin toxicity. 24 ' 25 Intravitreal Injection All intravitreal injections were administered as described 26 with a 25-gauge needle using a 1.0-ml tuberculine syringe. he needle was introduced 3 mm posterior to the limbus under visual control by means of an indirect ophthalmoscope and was advanced toward the region of the optic disk as close to the retina as possible. In every rabbit, a volume of 0.1 ml of drug solution was injected into the experimental eye while a similar volume of saline was injected into the fellow, control eye. he injection was carried out with the bevel of the needle pointing away from the retina to prevent any damage to the retina. Electroretinogram he ERG responses were recorded simultaneously from the experimental and control eyes with corneal

Gentamicin-Induced Retinal oxicity 1877 electrodes (Medical Workshop, Groningen, Holland) as described. 2 ' 1 he reference and ground electrodes were inserted into the ears. he ERG signals were amplified by a factor of 20,000 and filtered (0.3 to 300 Hz) by differential amplifiers (P511; Grass, Quincy, MA). he output of each amplifier was digitized at a rate of 2 khz (Lab-Master; Scientific Solutions, Encino, CA) and stored in a 386 computer for off-line analysis. Light stimuli were obtained from one of two sources. A xenon photostimulator (PS22; Grass) was used to elicit the ERG responses in the light-adapted state (background illumination of 11 foot lamberts). he intensity of this light source was controlled by the instrument settings (II to 116). he dark-adapted ERG responses were evoked by an electronic camera flash that could be attenuated by a set of neutral-density filters covering a range of 5 log units. Electroretinographic analysis was based on amplitude measurements of the b-waves. It was measured from the trough of the a-wave to the peak of the b- wave. For each rabbit in each ERG recording session, the b-waves of the experimental eye were divided by the corresponding values recorded from the control eyes. he mean b-wave ratio (experimental eye/control eye) was used as an index for retinal function in the experimental eye. 2t> Histology he eye was soaked for 10 minutes in 2% paraformaldehyde and 2.5% glutaraldehyde in 0.1 M phosphate buffer (ph 7.4). It was slit 2 mm posterior to the limbus to facilitate fixation. After 72 hours, the anterior segment of the eye was removed by a circumferential incision 2 mm posterior to the limbus. After removal of the vitreous, the posterior eyecup was bisected at the level of the optic disk. One half of each eyecup was rinsed in water and then dehydrated twice in 70% alcohol for 3 hours and twice in 96% alcohol for another 3 hours. Embedding was performed in a JB-4 resin (Bio-Rad, Watford, England). issue sections were cut by a Reichert Jung microtome at a thickness of 2 /Ltm and mounted onto slides. For light microscopy, the sections were stained with Richardson's stain. In a previous study, 27 we showed that small volumes of drugs injected into the vitreous might produce only localized structural effects. herefore, two loci were always inspected for each rabbit one close to the site of injection and the other approximately 15 mm more peripherally. RESULS he effects of 1.0 mg gentamicin on representative ERG responses of one albino and one pigmented rab- Albino Rabbit A (A) Light-adapted >A 116 *Vi A.V.-ZH-'-V J J V K H.: -3.0 \ -1.0 i \ l (B) Dark-adapted Pigmented Rabbit *\s*^ v»k FIGURE l. he effects of 1.0 mg gentamicin on the light- and dark-adapted electroretinographic responses as measured 1 week after intravitreal injection in one albino and one pigmented rabbit. Each pair of traces represents the responses recorded from the left, control eye (upper trace) and the right, experimental eye (lower trace). Stimulus intensity is given as the setting of the Grass photostimulator for the light-adapted responses and in stimulus attenuation (density of neutral filter) for the dark-adapted responses. Calibration bars: vertical, 100 /xv; horizontal, 100 msec. bit are compared in Figure 1. he responses were recorded 1 week after injection of the drug to the experimental right eye (lower traces) and of saline to the control left eye (upper traces). he ERG recordings demonstrate a severe functional deficit of the experimental retina in the light- and dark-adapted states in both animals. However, the reduction of the ERG amplitude is more pronounced in the albino than in the pigmented rabbit. his is most evident in the responses elicited in the dark-adapted state by the brightest light stimuli (bottom part of the figure). o assess the functional damage to the retina in a quantitative manner, the b-wave ratio was defined. 2 ' 1 th i

1878 Investigative Ophthalmology 8c Visual Science, August 1995, Vol. 36, No. 9 A FIGURE 2. Gentamicin-induced structural damage to the albino rabbit retina. he eyes were prepared for histology 2 weeks after injection of 1.0 mg gentamicin. Light micrographs from the experimental retina (C,D) are compared to corresponding areas from the control retina injected with saline solution (A,B). he micrographs were taken from the site of injection (A,C) and from a remote (15 mm from the optic disk) site (B,D). Calibration bar: 50 /im. For each stimulus intensity, the amplitude of the bwave recorded from the experimental eye was divided by the corresponding value recorded from the control eye. hese ratios were averaged to give die light- and dark-adapted b-wave ratios. For the two rabbits whose ERG responses are shown in Figure 1, the mean (± SD) of the light- and dark-adapted b-wave ratios were 0.16 ± 0.09 and 0.12 ± 0.08 for the albino rabbit and 0.33 ± 0.12 and 0.35 ± 0.15 for the pigmented rabbit. hus, the effect of 1.0 mg gentamicin was more pronounced in the albino rabbit retina. he functional loss induced by a dose of 1.0 mg gentamicin was supported by histologic observations as demonstrated for an albino rabbit in Figure 2. he eyes of this rabbit were enucleated 2 weeks after the injection of gentamicin, and both retinas were prepared for light microscopy. Micrographs of die experimental and control retinas, obtained from similar loci relative to the site of injection, were compared. he retina of the control eye injected with saline solution appeared unharmed in both loci; one was adjacent to die site of injection (Fig. 2A) and the other was more peripheral (15 mm from die optic disk) (Fig. 2B). he micrographs that had been obtained from the experimental retina exhibited severe damage at the site of injection (Fig. 2C) and more peripherally (15 mm away from the optic disk) (Fig. 2D). According to these micrographs, gentamicin-induced structural damage seemed to be restricted to the outer retina. It was manifested by an almost complete loss of inner and outer segments of the photoreceptors, whereas the outer nuclear layer and die inner layers of the retina appeared normal. he mean (± SD) light- and dark-adapted b-wave ratios of diis rabbit were 0.20 ± 0.05 and 0.10 ± 0.05, respectively. o estimate the safe dose of gentamicin, specifically the maximal dose that would not produce functional or structural retinal damage, or both, to the albino and pigmented rabbits, solutions containing different concentrations of die drug were injected intravitreally. heir toxic action was assessed from the b-wave ratios of the ERG responses 1 week after injec-

Gentamicin-Induced Retinal oxicity 1879 1.2 1.0 0.6 0-4 I D Albino 0 Pigmcnlcd 0.1 0.3 0.6 1 GENAMICIN DOSE (mg) K D Albino 0 Pigmcnlcd 0.1 0.3 0.6 I I GENAMICIN DOSK (mg) z^**w '/S'i'K FIGURE 3. he effects of different gentamicin doses on the electroretinogram (ERG) responses of albino and pigmented rabbits as determined 1 week after intravitreal injection. he mean (± SD) of the light-adapted (upper part) and dark-adapted (lower part) b-wave ratios are given for each dose. he size of each experimental group is indicated in the bars. Each ERG ratio that was significandy (P < 0.05) different between the two groups of rabbits is denoted by an asterisk. tion. hese data are summarized in Figure 3. he mean (± SD) values of the light-adapted b-wave ratios (upper part, Fig. 3) and of the dark-adapted b-wave ratios (lower part, Fig. 3) are given for four different doses of gentamicin sulfate: 0.1 mg, 0.3 mg, 0.6 mg, and 1.0 mg. In preliminary experiments, the b-wave ratios (right eye:left eye) were calculated for 25 untreated albino rabbits. Mean (± SD) values of 1.1 ± 0.24 and 0.99 ±0.16 were obtained for the light- and dark-adapted states, respectively. hese control data describe the interocular variability of the b-wave amplitudes and are independent on ocular pigmentation. hey depend only on the normal variability between the two eyes and on technical differences such as electrode placement, lighting conditions, and pupil size. he data of each experimental group were compared to the control group using a Student's West. he b- wave ratios of the albino and pigmented rabbits injected with 0.1 mg gentamicin did not differ from the normal (P > 0.05). In albino rabbits, both the lightand dark-adapted b-wave ratios were significantly (P < 0.05) reduced compared to the normal ratios for all other doses studied (0.3, 0.6, and 1.0 mg). For the pigmented rabbits, only doses of 0.6 and 1.0 mg gentamicin produced significant (P< 0.05) reduction of the light- and dark- adapted b-wave ratios. he data in Figure 3 indicate that the b-wave ratio was higher in the pigmented rabbits than in the albino rabbits for every dose of gentamicin studied. he differences were tested for statistical significance, and a significant (P < 0.05) difference between the b-wave ratios of albino versus pigmented rabbits was obtained for two doses, 0.3 mg and 1.0 mg (denoted by the asterisks above the bars, Fig. 3). he dose of 0.1 mg gentamicin did not cause any apparent ERG deficit in either albino or pigmented rabbits, whereas the damage caused by 0.6 mg gentamicin was more severe in albino than in pigmented rabbits. his difference, however, was not statistically significant. hese data indicate that the safe dose of gentamicin is higher for pigmented rabbits than for albino rabbits as determined electroretinographically 1 week after injection. It was demonstrated that synthetic melanin markedly reduced the activity of aminoglycosides by binding of the antibiotic. 28 hus, injected gentamicin may remain longer in the pigmented eye than in the albino eye. If so, gentamicin-induced toxicity to the pigmented rabbit retina, even though lower initially, may develop slowly over a longer period of time to reach a level similar or even greater than that seen in the albino eye. o investigate this possibility, we performed eight consecutive ERG recordings at different time intervals after the injection of 0.6 mg gentamicin in four albino and four pigmented rabbits. he first ERG recording was performed 1 hour after injection, and the last was performed 84 days later. he changes in the mean (± SD) light- and dark-adapted b-wave ratios with time are shown in Figure 4 (open circles, albino rabbits; filled circles, pigmented rabbits). Maximal reduction in retinal function, assessed from the ERG b-wave ratio, was seen 1 to 2 weeks after injection in both groups of rabbits. hereafter, the b-wave ratios in the pigmented rabbits gradually recovered and almost reached the preinjection values within 84 days. he albino rabbits exhibited litde recovery of the darkadapted b-wave ratios and virtually none of the lightadapted b-wave ratios within the follow-up period (84 days). he mean b-wave ratios representing the pigmented rabbits (filled circles) were significantly (P < 0.05) larger than the corresponding values for the albino rabbits (open circles) for every time interval longer than 7 days after injection (Fig. 4, asterisks). he data in Figure 4 indicate that the rabbit retina may recover from the deleterious effects of gentamicin even if severe damage initially is seen. o assess poten-

1880 Investigative Ophthalmology & Visual Science, August 1995, Vol. 36, No. 9 tial recovery and its relationship to initial damage, the maximal dark-adapted b-wave ratios, measured 84 days after gentamicin injection, were compared to the minimal values determined 1 week after injection. hese data are summarized in Figure 5 for 10 albino and 10 pigmented rabbits (open and filled circles, respectively) injected with either 0.3, 0.6, or 1.0 mg and tested at least twice (1 and 12 weeks) after injection. If there had been no functional recovery of the retina, all the data points in Figure 5 would have been expected to cluster around the identity line (continuous line). his was not the case. Only seven albino and three pigmented rabbits exhibited permanent severe ERG deficit that did not change appreciably from the measurement conducted 1 week after injection to that measured at the end of the follow-up period (12 weeks after injection). Substantial recovery was observed in seven pigmented and three albino rabbits, and normal b-wave ratios were measured in some of them at the end of the follow-up period. he 10 rabbits that exhibited substantial ERG recovery were characterized by minimal b-wave ratios larger than 0.2 (Fig. 5, dashed vertical line). hese data indicate that the rabbit retina, regardless of ocular pigmentation, may recover from the toxic action of gentamicin provided that the initial effect of the drug measured 1 week after injection is expressed by an ERG reduction <80%. he development of structural damage is illustrated in Figure 6, in which light micrographs of retinas from two albino rabbits injected with 0.6 mg gentamicin are compared. One rabbit was killed 3 weeks after injection (Figs. 6A, 6B), and the other was killed 12 weeks after injection (Figs. 6C, 6D). Micrographs were taken from the site of injection (Figs. 6A, 6C) 40 60 and more peripherally (15 mm from the optic disk) ime (Days) (Figs. 6B, 6D). In both rabbits, structural damage was similar across the retina and did not vary as a function of the distance from the site of injection. he retina that was prepared 3 weeks after injection was very thin but maintained its organization into alternate layers of cell bodies and synaptic interactions. Few cells were apparent in the outer nuclear layer, and there were no inner or outer segments. his retina was characterized by light- and dark-adapted b-wave ratios of 0.37 ±0.10 and 0.39 ± 0.07, respectively. he retina prepared 12 weeks after injection exhibited more severe anatomic destruction. It had lost its organized structure, and only few unidentified retinal cells could be 40 60 ime (Days) seen, scattered randomly within the retina. he b-wave FIGURE 4. he time course of gentamicin (0.6 mg)-induced ratios of this retina were 0.13 ± 0.08 and 0.09 ± 0.06 retinal damage in four albino (open circles) and four pig-fomented rabbits {filled circles). he mean (± SD) b-wave ratios the light- and dark-adapted states, respectively. he data described in Figures 1, 3, and 4 indicate for light-adapted (upper part) and dark-adapted (lower part) states are compared between the two groups. Significant (P < 0.05) differences are denoted by asterisks. 0.2 0.4 0.6 MINIMAL B-WAVE RAIO (7 days) Pigmented O Albino FIGURE 5. he dependency of functional recovery (maximal b-wave ratio) measured 84 days after injection on the initial, maximal damage (minimal b-wave ratio) determined 1 week after injection. Data for 10 albino (open circles) and 10 pigmented rabbits (filled circles) injected intravitreally with different doses of gentamicin are compared. Error bars denote the standard deviation of the b-wave ratio for each rabbit at each electroretinogram (ERG) recording session. he continuous line is the identity line. he dashed vertical line indicates an initial damage of 80% reduction in the ERG responses. 0.8

1881 Gentamicin-Induced Retinal oxicity Structural retinal damage in two albino rabbits that underwent permanent electre retinogram deficit after the injection of 0.6 mg gentamicin. he retina of one rabbit was prepared for light microscopy at 3 weeks (A,B), and the retina of the other rabbit was prepared at 12 weeks (C,D) after injection. Light micrographs were obtained from the site of injection (kc)and a more peripheral (15mm from the optic disk) site (C,D).Calibration bar: 50 pm. FlCURE6. that the melanin in the pigment epithelium provides the rabbit retina with some degree of protection against the toxic effects of gentamicin. Melanin may influence gentamicin-induced retinopathy by at least three mechanisms. It absorbs stray light bypassing the photoreceptors thereby reducing the degree of light exposure;22 it is a scavenger of free radicals and may act as an antioxidant;'%d it binds the drug and reduces the concentration of free gentarnicin.'" hese alternatives were tested by comparing the effects of 0.6 mg gentamicin on the ERG in 25 albino rabbits that were divided into three groups, as shown in Figure 7. One group was maintained in a 12-hour light12-hour dark cycle. he second group was kept in total darkness after injection, and the animals in the third group were injected with a mixture of gentamicin and the antioxidant N-acetylcystein. he mean ( 2 SD) light- and dark-adapted b-wave ratios of these groups did not differ statistically. he data shown in Figure 7 suggest that neither the lighting conditions nor the addition of an antioxidant provided any pro- tection to the albino rabbit retina against gentamicininduced damage. It should be mentioned that in preliminary experiments, we found that neither 1 week of darkness nor N-acetylcystein produced any effect on the b-wave ratios (right eye:left eye) of albino rabbits. DISCUSSION hese findings support those of previous reports2-* on the toxic action of gentamicin on the rabbit retina. Functional retinal damage, assessed from the ERG responses, could be detected as early as 3 hours after intravitreal injection of the drug. he degree of functional damage slowly developed with time, reaching a maximal level within 1 to 2 weeks of injection. hereafter, it either stabilized or exhibited slow but progressive recovery. In some of the rabbits, the ERG responses completely recovered to the normal preinjection level within 12 weeks (Fig. 4). here seems to be a clear threshold that separates retinas that can recover from those that suffer from permanent damage (Fig.

1882 Investigative Ophthalmology 8c Visual Science, August 1995, Vol. 36, No. 9? 0.8 8 0.7 1 0.6 I 0.5 0.4 P 0.3 * 0.2 < 0.1 m 0.0 D Light adapted 0 Dark adapted (N=9) (N=8) (N=8) Light/Dark Dark Light/Dark with NAC FIGURE 7. he effect of 0.6 mg gentamicin on the mean (± SD) light- and dark-adapted b-wave ratios of three groups of albino rabbits. One group of rabbits was injected with gentamicin alone and was kept in a 12-hour light-12-hour dark cycle. Rabbits in the second group were injected with a mixture of gentamicin and N-acetylcystein and kept in a 12-hour light-12-hour dark cycle. Rabbits in the third group were injected with gentamicin alone and were kept in complete darkness. he number of animals in each experimental group is denoted in the figure. 5). Eyes that were characterized by more than 80% reduction in the ERG responses (b-wave ratios <0.2) exhibited little recovery, whereas those suffering from less initial damage (b-wave ratio >0.2) exhibited substantial recovery and could attain the preinjection b- wave ratio. he retinas that recovered from the toxic action of gentamicin appeared unharmed at the light microscopic level (data not shown). he observation that the rabbit retina may recover from gentamicininduced damage raises the possibility that some of the variability in the safe dose of gentamicin, determined in different studies, can be accounted for by differences in the experimental protocols used. hus, a dose of gentamicin that leads to a 50% ERG reduction when measured 1 week after injection and that is therefore reported as highly toxic may appear safe if the ERG is measured several weeks later. Morphologic observations indicated that the initial site of gentamicin toxic action was the outer retina. 78 wo weeks after injection, when the physiological effect of gentamicin reached its peak, the inner retina and the outer nuclear layer seemed intact, whereas the inner and outer segments of the photoreceptors were missing (Fig. 2). his type of structural damage is reminiscent of the one induced by hemicholinium-3, a drug that is a competitive inhibitor of choline uptake and that interferes with the outer segment renewal process. 29 Gentamicin also may exert its toxic action by interfering with the synthesis of new outer segment disks. wo cellular processes needed for the production of new outer segment disks, protein synthesis and lipid metabolism, may be the site of action of gentamicin. he aminoglycoside antibiotics are known to act on the bacterial ribosomes, leading to the production of faulty proteins and eventually to death of the microorganism. 8031 Alternatively, aminoglycoside-induced ototoxicity has been attributed to the interactions of the drug with polyphosphoinositoles. 32 ' 33 Phosphatydilinositol is abundant in the rod outer segment and is needed for synthesis of outer segment disks. 34 ' 35 herefore, interaction of gentamicin with this phospholipid may interfere with the outer segment renewal process. At prolonged periods of time after gentamicin injection, the rabbit retina may exhibit either substantial recovery or permanent functional and structural damage, depending on the degree of initial damage (Fig. 5). When the initial damage is severe (b-wave ratio <0.2), the synthesis of new outer segment disks fails to restore the outer segments, which leads to permanent damage. In these retinas, structural damage, first seen in the outer segments, spread with time to the entire retina (Fig. 6). In retinas experiencing only mild to moderate initial damage (b-wave ratio >0.2), the rate of synthesis of new outer segment disks is probably sufficient to restore the integrity of the photoreceptors. In that case, the retina appears structurally and functionally normal at longer time intervals after injection. We found that the degree of gentamicin-induced functional damage strongly depended on ocular pigmentation. A dose of 0.1 mg did not cause any noticeable functional (Fig. 3) and structural damage (not shown here) either in albino or in pigmented eyes and could, therefore, be considered safe for both. his dose can be converted to a vitreal concentration of 68 //g/ml, assuming a vitreous volume of 1.47 ml for rabbits weighing 1.8 to 2.7 kg 36 and complete mixing of the drug in the vitreous. his concentration is similar to the levels used clinically. However, for every higher dose tested (0.3, 0.6, and 1.0 mg), the maximal degree of damage, assessed 1 week after injection from the b-wave ratio, was larger in albino than in pigmented rabbits (Fig. 3). herefore, permanent damage was considerably more apparent in albino than in pigmented rabbits. Approximately 3 months after injection of gentamicin, 7 out of 10 albino rabbits exhibited severe ERG deficit whereas only 3 out of 10 pigmented rabbits exhibited similar degrees of permanent functional damage (Fig. 5). hus, the safe dose of gentamicin is expected to be higher in pigmented eyes than in albino eyes. Similar differential susceptibility to gentamicin of albino and pigmented animals has been demonstrated in the cochlea of guinea pigs. 37 ' 38 he protection that melanin provided the rabbit retina from gentamicin toxicity could arise from the absorption of stray light, 22 from the absorption of free

Gentamicin-Induced Retinal oxicity 1883 radicals, 523 or from the binding of free gentamicin. 28 Our data suggested against the first two alternatives. Reducing the level of ambient illumination by keeping the rabbits in darkness or mixing gentamicin with a scavenger of free radicals did not protect the retina of albino rabbits from gentamicin-induced damage (Fig. 7). herefore, binding of free gentamicin by melanin was probably the mechanism by which ocular pigmentation partially protected the retina from gentamicin toxicity. hese observations indicate that gentamicin toxicity to the retina is mediated by the free, unbound drug only and that bound gentamicin does not affect the retina. Key Words electroretinography, gentamicin, melanin, rabbit, retina References 1. Doft BH, Barza M. Ceftazidime or amikacin: Choice of intravitreal antimicrobials in the treatment of postoperative endophthalmitis. Arch Ophthalmol. 1994; 112:17-18. 2. Peyman GA, Paque J, Meisels HI, Bennet O. Postoperative endophthalmitis: A comparison of methods for treatment and prophylaxis with gentamicin. Ophthalmic Surg. 1975; 6:45-55. 3. Mochizuki K, orisaki K, Kawasaki Y, et al. Retinal toxicity of antibiotics: Evaluation by electroretinogram. Doc Ophthalmol. 1988; 69:195-202. 4. Zachari IG, Forster RK. Experimental intravitreal gentamicin. Am J Ophthalmol. 1976;82:604-611. 5. Atkins WS, McPherson SD Jr. Intravitreal injection of ampicillin, cephaloridine, gentamicin, and tobramycin. Surg Forum. 1975; 26:542-544. 6. Peyman GA, May DR, Ericson ES, Apple D. Intraocular injection of gentamycin: oxic effects of clearance. Arch Ophthalmol. 1974; 92:42-47. 7. D'Amico DJ, Libert J, Kenyon KR, Hanninen LA, Caspers-Velu L. Retinal toxicity of intravitreal gentamicin: An electron microscopic study. Invest Ophthalmol VisSci. 1984; 25:564-572. 8. D'Amico DJ, Caspers-Velu L, Libert J, Shanks EMS, Hanninen LA, Kenyon KR. Comparative toxicity of intravitreal aminoglycoside antibiotics. Am J Ophthalmol. 1985; 100:264-275. 9. Ling CH, Peyman GA, Raichard M. Electron microscopic study of toxicity of intravitreal injections of gentamicin in primates. Can J Ophthalmol. 1985;20:l79-183. 10. Conway BP, abatabay CA, Campochiaro PA, D'Amico DJ, Hanninen LA, Kenyon KR. Gentamicin toxicity in the primate retina. Arch Ophthalmol. 1989; 107:107-112. 11. Bennet O, Peyman GA. oxicity of intravitreal aminoglycosides in primates. Can J Ophthalmol. 1974;9:475-478. 12. Brown GC, Eagle RC, Shakin EP, Gruber M, Arbizio W. Retinal toxicity of intravitreal gentamicin. Arch Ophthalmol. 1990; 108:1740-1744. 13. McDonald HR, Schatz H, Allen AW, et al. Retinal toxicity secondary to intraocular gentamicin injection. Ophthalmology. 1986; 93:871-877. 14. Snider JD, Cohen HB, Chenoweth RG. Acute ischemic retinopathy secondary to intraocular gentamicin. In: Ryan SJ, Dawson AK, Little HL, eds. Retinal Diseases. New York: Grune & Stratton; 1985:227-238. 15. Conway BP, Campochiaro PA. Macular infarction after endophthalmius treated with vitrectomy and intravitreal gentamicin. Arch Ophthalmol. 1986; 104:367-371. 16. Campochiaro PA, Lim JI. Aminoglycoside toxicity in the treatment of endophthalmitis: he Aminoglycosides oxicity Study Group. Arch Ophthalmol. 1994; 112:48-53. 17. alamo JH, D'Amico DJ, Hanninen LA, Kenyon KR, Shanks E. he influence of aphakia and vittectomy on experimental retinal toxicity of aminoglycoside antibiotics. Am J Ophthalmol. 1985; 100:840-847. 18. so MOM. Retinal photic injury in normal and scorbutic monkeys. rans Am Ophthalmol Soc. 1987; 85:498-556. 19. Noell WK. Possible mechanisms of photoreceptor damage by light in mammalian eyes. Vision Res. 1979;20:1163-117l. 20. de Lint PJ, van Norren D, oebosch AM. Effect of body temperature on threshold for retinal light damage. Invest Ophthalmol Vis Sci. 1992;33:2382-2387. 21. O'Steen WK, Sweatt AJ, Brodish A. Effects of acute and chronic stress on the neural retina of young, midage, and aged Fischer-344 rats. Brain Res. 1987; 426:37-46. 22. Rapp LM, Williams P. he role of ocular pigmentation in protecting against retinal light damage. Vision Res. 1980;20:1127-1131. 23. Riley PA. Radicals in melanin biochemistry. Ann NY AcadSci. 1988;551:119-120. 24. Knight KR, MacPhadyen K, Lepore DA, Kuwata N, Eadie PA, O'Brien BM. Enhancement of ischaemic rabbit skin flap survival with the antioxidant and free radical scavenger N-acetylcysteine. Clin Sci (Colch). 1991;81:31-36. 25. Sala R, Moriggi E, <Jorvasce G, Morelli D. Protection by N-acetylcysteine against pulmonary endothelial cell damage induced by oxidant injury. Euro Respir J. 1993; 6:440-446. 26. Loewenstein A, Zemel E, Lazar M, Perlman I. Druginduced retinal toxicity in albino rabbits: he effects of imipenem and aztreonam. Invest Ophthalmol Vis Sci. 1993; 34:3466-3476. 27. Zemel E, Loewenstein A, Lazar M, Perlman I. he effects of myristyl gamma-picolinium chloride on the rabbit retina: Morphologic observations. Invest Ophthalmol Vis Sci. 1993; 34:2360-2366. 28. Barza M, Baum J, Kane A. Inhibition of antibiotic activity in vitro by synthetic melanin. Antimicrob Agents Chemother. 1976; 10:569-570. 29. Pu GA, Masland RH. Biochemical interruption of membrane phospholipid renewal in retinal photoreceptor cells. J Neurosd. 1984; 4:1559-1576.

1884 Investigative Ophthalmology & Visual Science, August 1995, Vol. 36, No. 9 30. Brewer NS. Antimicrobial agents: II: he aminoglycosides: Streptomycin, kanamycin, gentamycin, tobramycin, amikacin, neomycin. Mayo Clin Proc. 1977; 52:675-679. 31. Moellering RC. In vitro antibacterial activity of the aminoglycoside antibiotics. Rev Infect Dis. 1983;5: S212-S222. 32. Wang BM, Weiner ND, akada A, Schacht J. Characterization of aminoglycoside-lipid interactions and development of a refined model for ototoxicity testing. Biochem Pharmacol. 1984; 33:3257-3262. 33. Schacht J. Isolation of an aminoglycoside receptor from guinea pig inner ear tissues and kidney. Arch Otorhinolaryngol. 1979; 224:129-134. 34. Anderson RE, Maude MB, Kelleher PA. Metabolism of phosphatidylinositol in the frog retina. Biochim Biophys Ada. 1980; 620:236-46. 35. Wetzel MG, O'Brien PJ. urnover of palmitate, arachidonate and glycerol in phosopholipids of rat rod outer segments. Exp Eye Res. 1986; 43:941-954. 36. Green H, Sawyer JL, Leopold IH. Elaboration of bicarbonate ion in intraocular fluids: II: Vitreous humor, normal Values. AMA Arch Ophthalmol. 1957;57:85-89. 37. Conlee JW, Gill SS, McCandless P, Creel DJ. Differential susceptibility to gentamicin ototoxicity between albino and pigmented guinea pigs. Hear Res. 1989;41:43-52. 38. Conlee JW, Jensen RP, Parks N, Creel DJ. urn-specific and pigment dependent differences in the stria vascularis of normal and gentamicin-treated albino and pigmented guinea pigs. Hear Res. 1991; 55:57-69.