Evaluation of the Retinal Toxicity and Pharmacokinetics of Dexamethasone After Intravitreal Injection Hyung Woo Kwak, MD, Donald J. D'Amico, MD
\s=b\ The intravitreal injection of steroids may be potentially useful in the treatment of endophthalmitis and other ocular inflammatory diseases. The retinal toxicity and intraocular turnover of aqueous solutions of dexamethasone sodium phosphate in doses ranging from 440 to 4000 \g=m\g were evaluated in the rabbit; evaluation was also performed for a 0.1-mL injection of a commercially available preparation (dexamethasone phosphate [Decadron] injection, 4 mg/mL). After the 440-\g=m\g dose, a transient increase in staining of the M\l=u"\llercells was observed, which normalized after 2 days. Progressively higher doses resulted in an increasing spectrum of disorganization in M\l=u"\llerand other retinal cells. The half-life of the intravitreally injected drug was 3.48 hours. These findings suggest a primary interference in M\l=u"\llercell function, possibly through dexamethasoneinduced alterations in retinal glutamate or glucose metabolism.
(Arch Ophthalmol. 1992;110:259-266) ""1 he intravitreal injection of steroid has been evaluated for the treat¬ ment of endophthalmitis1"4 and
ative
prolifer¬
and may po¬ tentially be useful in the treatment of other ocular diseases. In the treatment of endophthalmitis, the intravitreal in¬ jection of steroids has been beneficial in certain animal models,1-4 and its suc¬ cessful clinical use in association with intravitreal antibiotic injections and vi¬ trectomy has been reported.2,3 The ra¬ tionale for intravitreal steroid adminis¬ tration in endophthalmitis is based on the need to preserve the retina and other intraocular tissues from the po¬ tentially devastating and rapidly pro¬ gressive effects of inflammation associated with infection. These con¬ siderations suggest that intravit¬ real steroid injection may be a reason¬ able adjunct to initial intravitreal anti-
vitreoretinopathy5"7
Accepted for publication August 16,
1991. From the Retina Service (Dr D'Amico) and the Howe Laboratory of Ophthalmology (Drs Kwak and D'Amico), Massachusetts Eye and Ear Infirmary, Department of Ophthalmology, Harvard Medical School, Boston. Reprint requests to Massachusetts Eye and Ear Infirmary, 243 Charles St, Boston, MA 02114
(Dr D'Amico).
biotic
therapy. In the management of experimental proliferative vitreoreti¬ nopathy, intravitreal steroid therapy has demonstrated
a
beneficial effect in
reducing the severity of intraocular cellular proliferation and subsequent retinal detachment.
Dexamethasone, a synthetic steroid,
is widely used in the treatment of ocular inflammatory diseases. Despite considerable clinical interest in intra¬ vitreal dexamethasone therapy for en¬ dophthalmitis and other ocular inflam¬ matory disorders, the ocular toxic re¬ action and drug turnover following intravitreal injection of dexametha¬ sone and related compounds have re¬ ceived limited study.1 Dexamethasone alone is poorly suited to intravitreal use because it has limited solubility in aqueous media, and this property would require the injection of poten¬ tially toxic solvent vehicles into the eye. However, dexamethasone sodium phosphate has a high degree of solu¬ bility and stability in aqueous solution; furthermore, commercial preparations are readily available. Herein, we de¬ scribe dose-related effects of intra¬ vitreal dexamethasone sodium phos¬ phate on the retina of rabbits and report the half-life of this drug after injection into the vitreous. MATERIALS AND METHODS Intraocular Toxic Reaction All studies were performed using Dutch Belted pigmented rabbits (weight range, 2.5 to 3.5 kg). They were housed in sepa¬ rate cages with a 12-hour light-dark cycle. All animal handling and procedures con¬ formed to the guidelines of the Association for Research in Vision and Ophthalmology. The toxicities of (1) pure dexamethasone sodium phosphate in aqueous solution; (2) dexamethasone phosphate (Decadron) injection, a commercially available prepara¬ tion (Merck Sharp & Dohme, West Point, Pa); and (3) Decadron phosphate injection vehicle were evaluated after intravitreal injection. The following doses were studied: (1) in the dexamethasone sodium phosphate study group, 440, 800, 1200, 1600, 2000, and 4000 µg (a total of 56 eyes); (2) in the Decadron phosphate injection study group, a dose of 0.1 mL of commercially available 4 mg/mL preparation (this dose contains 440 µg of dexamethasone sodium phosphate after correction for the appropriate molecu-
Downloaded From: http://archopht.jamanetwork.com/ by a Western University User on 06/11/2015
lar weight of the salt form; a total of 10 eyes included in this group); and (3) in the Decadron phosphate injection vehicle group, 0.1 mL (six eyes). Two additional control groups were evaluated, and these consisted of 10 saline-injected and 10 uninjected eyes. The rabbits were anesthetized with a 1.5-mL mixture of equal parts of xylazine hydrochloride (20 mg/mL) and ke¬ tamine hydrochloride (100 mg/mL) by an intramuscular injection, and their pupils were dilated with 0.5% topical tropicamide and 5% phenylephrine hydrochloride drops. Before intravitreal injections, an anterior chamber paracentesis was performed. The drugs were injected into the vitreous using a 1-mL syringe with a 30-gauge needle inserted 2 to 3 mm posterior to the limbus. All drug injections were performed in one eye of each animal; the other eye served as one of the controls and was either left uninjected or injected with 0.1 mL of saline solution as described above. All solutions were passed through a 0.2-µ filter before were
injection.
Rabbits were periodically examined by indirect ophthalmoscopy and photographed after pupillary dilation. They were killed at intervals after the injection ranging from 3 hours to 3 weeks. The eyes were hemisected at the equator, the vitreous was removed, and the posterior portion of the eyes was fixed in 2.5% glutaraldehyde for 24 hours. For light and transmission elec¬ tron microscopy, the tissue was dissected into 2-mm2 fragments, postfixed in 1% os¬ mium tetroxide, dehydrated in graded alco¬ hols, and embedded in epoxy resin. Semithin sections were stained with méthy¬ lène blue-azure II for examination by light microscopy, and ultrathin sections were stained with uranyl acetate-lead citrate. For scanning electron microscopy, the reti¬ na was divided into sensory retina and retinal pigment epithelium (RPE). Speci¬ mens were then postfixed in buffered 1% osmium tetroxide and dehydrated through graded ethanols. The specimens were criti¬ cal point dried with carbon dioxide, mount¬ ed on stubs, and coated with gold-
palladium. For light-microscopic immunocytochemistry for glial fibrillary acidic protein (GFAP), the eyes were hemisected at the equator and the vitreous removed with
filter paper. The posterior portion of the globe was fixed in 4% paraformaldehyde in 0.1 mol/L of sodium phosphate buffer (pH 7.4) for 1 hour at room temperature and for an additional 12 hours at 4°C. After fixa¬ tion, specimens were incubated in 5% su¬ crose in sodium phosphate buffer for 1 hour and then immersed in 30% sucrose for 12 hours. Tissues were embedded in OCT
Fig 1 Fundus photographs after the intravitreal injection of 4000 µg of dexamethasone sodium phosphate. Left, Three days after injection, retinal edema and serous detachment are shown. Right, One week after injection, the retinal edema —
.
has resolved.
Fig 2.—Light micrograph after intravitreal dexamethasone sodium phosphate injection. Top left, Six hours after injection of 440 µg of dexamethasone sodium phosphate, conspicuous staining of Müller cells is observed (arrows). Top right, Two days after injection of 440 µ of dexamethasone sodium phosphate, Müller cell staining has normalized. Bottom left, One week after injection of 2000 µg of dexamethasone sodium phosphate, a vacuole is noted at the border of the outer plexiform and outer nuclear layers (arrow). Bottom right, Three days after injection of 4000 µg of dexamethasone sodium phosphate, marked perinuclear edema of photoreceptor and other neuronal cells is shown. BM indicates Bruch's membrane (méthylène blue-azure II stain, original magnification 1800). (Miles Scientific, Naperville, 111), and 5-µ cryostat sections were prepared and main¬
tained at -20°C. The sections were covered by a drop of normal goat serum (Cappel, Organon Tecknika Corp, West Chester, Pa) diluted 1:50 with phosphate-buffered saline
(PBS) for 20 minutes. The normal goat blotted from the sections, which
serum was
then incubated 45 minutes with mouse monoclonal anti-GFAP (Sigma Chemical Co, St Louis, Mo). The anti-GFAP was diluted 1:250 with PBS and contained 1%
were
Downloaded From: http://archopht.jamanetwork.com/ by a Western University User on 06/11/2015
bovine serum albumin. After a 10-minute rinse in PBS, the sections were incubated with biotinylated goat anti-mouse IgG (1:1000 preparation; Jackson Immunoresearch Labs Ine, Avondale, Pa) for 45 min¬ utes. The slides were subsequently rinsed
Fig
3.—Transmission electron
micrograph
1 week after intravitreal
injection of
2000 µg of dexamethasone sodium
phosphate. A large round vacuole (asterisk) appears between the outer plexiform and outer nuclear layers (OPL and ONL).
The vacuole was located inside the synaptic vesicle at the receptor site of the nuclear layer (uranyl acetate-lead citrate, original magnification 3300). in PBS for 10 minutes and incubated with
peroxidase-conjugated streptavidin (1:1000 preparation; Jackson Immunoresearch Labs) for 45 minutes. After rinsing for 10 minutes in PBS, incubating with 3-amino-9ethylcarbazole peroxidase substrate solu¬ tion (Polysciences Ine, Washington, Pa) for 10 minutes, and rinsing in distilled water, a coverglass was applied to the sections with gelvatol.
Intraocular Clearance Before intravitreal injections, an anterior paracentesis was performed. All injections were performed slowly with the bevel of the needle facing anteriorly. In one group, dexamethasone sodium phosphate (440 µg in 0.1 mL of preservative-free sterile wa¬ ter) was injected through the pars plana, 2 mm posterior to the limbus, using a 30-gauge needle. In a second group, the major component of Decadron phosphate
injection vehicle, paraben (150 µg), was similarly injected. The eyes were enucleat¬ ed immediately following death at 1.5, 3, 6, 12, 24, 48, and 72 hours after injection. Aqueous specimens were collected with a
photoreceptor terminal. INL indicates inner
25-gauge needle. The eyes were subse¬ quently immersed in liquid nitrogen, and the frozen vitreous
was
removed. The
re¬
sulting samples were analyzed by highperformance liquid chromatography (HPLC).8 After homogenization, the speci¬ mens were applied to a Sep-Pak C-18 car¬ tridge (Waters Associates, Milford, Mass) and eluted. The eluate was evaporated to dryness by a speed vacuum concentration. Each residue was then resuspended in 1.0 mL of methanol and vortexed. Chroma¬
tography was performed on a Waters µ Bondapak C18 column (Waters Associates, Milford, Mass) run isocratically at 1.0 mL/min using 50:50 methanol-water (vol/vol)
as a
solvent. Graded amounts of
prednisone were added to each specimen to
as a recovery marker and HPLC standard. Similarly, HPLC was used to determine paraben concentrations by mea¬ suring recovery and location of added para¬ ben. The limits of detection for the total vitreous content by HPLC were 5.2 ±1.7 µg for dexamethasone sodium phosphate and 8.7 ±2.8 µg for paraben. serve
Downloaded From: http://archopht.jamanetwork.com/ by a Western University User on 06/11/2015
RESULTS Intraocular Toxic Reaction
The cornea and lens of all experi¬ mental and control eyes appeared nor¬ mal by slit-lamp examination. The vit¬ reous and retina appeared normal by indirect ophthalmoscopy in vehicle-in¬ jected and 0.1 mL Decadron phospha¬ te-injected eyes. The dexamethasone sodium phosphate-injected eyes ap¬ peared normal in doses up to 2000 µg; however, after the 4000^g dexameth¬ sodium phosphate injection, asone there was vitreous clouding, retinal edema, and serous detachment within 24 hours (Fig 1). These findings re¬ solved within 2 weeks, and all experi¬ mental eyes appeared normal to oph¬ thalmoscopy at that time (Fig 1). Light microscopy and ultrastructural evalua¬ tion showed no evidence of retinal ab¬ normality in the vehicle-injected, sa¬ line-injected, or uninjected eyes.
Three days after intravitreal injec¬ tion of 4000 µg of dexamethasone sodi¬ um phosphate, the photoreceptor cells in the ONL and neuronal cells in the inner nuclear layer showed prominent swelling (Fig 2, bottom right; Fig 4) in addition to the amorphous granules on the surface of the RPE. Photorecep¬ tors also were degenerated, with disin¬ tegration of the regular lamellar pat¬ tern of their outer segment discs in association with necrotic debris (Fig 5). The Müller cells displayed vacuolar and degenerative changes in the cyto¬ plasm. After 1 week, the RPE cells showed marked melanin granules (Fig 6, top) in association with the amor¬ phous granules on the RPE surface
(Fig 6, bottom). By light microscopic immunohisto¬ chemistry, anti-GFAP labeling was positive 1 week after injection of 2000 µg of dexamethasone sodium phos¬ phate compared with uninjected con¬ trols (Fig 7). The reaction product appeared as strands extending from the internal limiting membrane to the OPL, and there was a specific increase in labeling of the Müller cells. Intraocular Clearance
The ocular clearance of intravitreal
injections of 440 µg of dexamethasone sodium phosphate or 0.1 mL of para¬ ben vehicle are shown in Figs 8 and 9;
Fig 4.—Transmission electron micrograph 3 days after intravitreal injection of 4000 µg of dexamethasone sodium phosphate. Photoreceptors and other neuronal cells display promi¬ nent swelling. INL indicates inner nuclear layer; OPL, outer plexiform layer; and ONL, outer nuclear layer (uranyl acetate-lead citrate, original magnification 6900). After the 0.1-mL Decadron phos¬ or the 440^g dexamethasone sodium phosphate injection, micro¬ scopie examination revealed marked staining in the Müller cells compared with the other cells (Fig 2, top left). This staining was prominent at 6 hours after injection and then decreased and normalized at 2 days (Fig 2, top right). Apart from the increased staining de¬ scribed, no structural abnormalities were noted at any time with light or transmission electron microscopy after this dosage level. Minimal changes were seen after intravitreal injection of 800 µg of dexa¬ methasone sodium phosphate. After 3 days, small vacuoles appeared be¬ tween the outer plexiform layer (OPL) and the outer nuclear layer (ONL); these vacuoles disappeared at 1 week. Intravitreal injection of 1200 µg of dexamethasone sodium phosphate pro¬ duced more distinct vacuoles in the same sites. In addition, an accumula¬ tion of amorphous granules ranging
phate
from 2 to 10 µ appeared on the surface of the RPE and in the interphotoreceptor matrix after 1 day. These granules appeared to be mem¬ brane limited and were still evident at 3 weeks. Disintegration of the photore¬ ceptor outer segments was noted, and photoreceptor cells in the ONL were edematous. After 3 weeks, the edema of the photoreceptor cells subsided, but occasional necrotic photoreceptor cells were observed. Large round vacuoles appeared be¬ tween OPL and ONL 3 days after the intravitreal injection of 1600 to 4000 µg of dexamethasone sodium phosphate (Fig 2, bottom left), and the size of the vacuoles did not change during 3 weeks. Ultrastructurally, vacuoles were located inside the synaptic vesi¬ cle primarily at the receptor site of the photoreceptor terminal (Fig 3). In some cases, the vacuoles appeared to be fused. By scanning electron micros¬ copy, the RPE showed pleomorphism and irregularity of apical cell border.
Downloaded From: http://archopht.jamanetwork.com/ by a Western University User on 06/11/2015
the values represent the total amount of dexamethasone sodium phosphate or paraben present in the entire aque¬ ous or vitreous body. Dexamethasone sodium phosphate in the aqueous reached its highest level (8.2 ±0.7 µg) in 1.5 hours after injection but was not detected in the aqueous at 12 hours. Vitreous dexamethasone sodium phos¬ phate content at initial injection was 387.5 ±16.9 (ig; at 3 hours, 188.6 ±6.3 µg; at 6 hours, 126.1 ±39.8 µg; and at 24 hours, 5.7 ±2.2 µg. At 72 hours, dexamethasone sodium phosphate was not detected in the vitreous; the calcu¬ lated half-life of the intravitreal dexa¬ methasone sodium phosphate was 3.48 hours. The paraben was not detected in the aqueous after intravitreal injec¬ tion. The initial vitreous paraben con¬ tent was 84.3 ± 16.2 µg, and at 3 hours the paraben was not detectable in the vitreous; the calculated half-life of in¬ travitreal paraben was 0.42 hours. COMMENT
This study demonstrates the lack of retinal toxic reaction following an in¬ travitreal injection of 440 µg of dexa¬ methasone sodium phosphate or the equivalent dose of a commercially available, soluble dexamethasone sodi¬ um
phosphate preparation (Decadron
ry site of intravitreal dexamethasone toxic reaction is the Müller cell. It is well established that the structural and nutritional support systems of the retina are supplied predominantly by Müller cells. Müller cells also play a prominent role in maintaining an envi¬ ronment required for neuronal func¬ tion. In this regard, the documented functions of the Müller cells are exten¬ sive and include the clearance of potas¬ sium released by neuronal activity or
by pathologic conditions,9 the uptake
and catabolism of neurotransmitters axon terminals,10 the control of water movement between intracellular and extracellular com¬ partments and tissue pH,1112 and, most importantly, the provision of glucose.13 Consequently, the Müller cell is sus¬ ceptible to abnormality by many fac¬ tors, including mechanical disruption, metabolic abnormalities, and drugs. The mechanism of the observed ef¬ fects of intravitreal dexamethasone so¬ dium phosphate on the Müller cell is unknown. However, steroids are known to influence at least two specific pathways present in Müller cells, and these interactions suggest possible mechanisms for the toxic reaction ob¬ served in the present study. The first interaction centers on the observation that dexamethasone treat¬ ment produces an increase in the level of glutamine synthetase in a variety of cells and specifically in Müller cells. In the retina of the late chick embryo, a rapid and striking increase in gluta¬ mine synthetase has been reported fol¬ lowing systemic elevation of adrenal corticoid levels.14 Furthermore, in the retina and other cell systems, it has been specifically established that dexa¬ methasone enhances the level of gluta¬ mine synthetase messenger RNA.15,16 In the retina, this enzyme has been shown to be exclusively confined to Müller cells." The pathways for this enzyme have been partially character¬ ized in the central nervous system.18 Glutamine synthetase is confined to glial cells and functions to convert glu¬ tamate (a putative excitatory neuro¬ transmitter and potential excitotoxin to neural cells) to glutamine (a neurophysiologically inactive substance). As a result of the action of glutamine synthetase, glutamine produced in gli¬ al cells is transported into neuronal cells, where glutamate is again pro¬ duced and sequestered in glutaminergic synaptic vesicles through the action of glutaminase. It is, therefore, possible that intra¬ vitreal adrenocorticoids activate gluta¬ mine synthetase in Müller cells with a resulting increase in levels of neuronal
discharged by
Fig 5. —Transmission electron micrograph 3 days after injection of 4000 µg of dexamethasone sodium phosphate. Photoreceptors display degenerative changes. BM indicates Bruch's membrane (uranyl acetate-lead citrate, original magnification 4670).
phosphate injection) that includes par¬ aben in its vehicle. Although in our experimental study the Müller cells showed increased méthylène blueazure II staining with this dose com¬ pared with other retinal cells, this finding was reversible and is sugges¬ tive of reactive rather than degenera¬ tive change. However, with 800^g and higher doses of the primary com¬ ponent dexamethasone sodium phos¬ phate, the Müller cells displayed an increasing toxic reaction manifested by edema and degeneration. In addition, an increasing spectrum of changes was
also noted in other retinal cells. When doses ranging from 800 to 4000 µg of
dexamethasone sodium phosphate injected into the vitreous, a doserelated increase in the number of vacu¬ oles between OPL and ONL was ob¬ served. The vacuoles were more prominent after injection of 1600 to 4000 µg and were located inside the neuronal synapses. In addition, after intravitreal injection of dexametha¬ sone sodium phosphate at 1200-µg and higher doses, the photoreceptor outer segments showed edema and disinte¬ gration, and there was accumulation of amorphous granules on the surface of the RPE or in the interphotoreceptor matrix. These findings suggest that a prima-
were
Downloaded From: http://archopht.jamanetwork.com/ by a Western University User on 06/11/2015
Fig 6. —Top, Transmission electron micrograph 1 week after injection of 4000 µg of dexameth¬ sodium phosphate. An abnormal accumulation of amorphous granules (arrows) ranging from 2 to 10 µ are observed on the surface of the retinal pigment epithelium (RPE) or in the interphotoreceptor matrix. These granules appeared to be membrane limited and did not disappear at 3 weeks. The RPE cells show prominent melanin pigment. BM indicates Bruch's membrane (uranyl acetate-lead citrate, original magnification 5100). Bottom, Scanning electron micrograph of the RPE surface 1 week after injection of 4000 µg of dexamethasone sodium phosphate. Amorphous granules (arrows) are observed in association with heteroge¬ neity of the RPE (gold-palladium coated, original magnification 2040). asone
glutamine and subsequently neuronal glutamate; the resulting increase may produce cellular dysfunction and toxic
reaction. Potential toxic effects that appear to be mediated by glutamate in the retinal neurons include an increase in neuronal permeability to sodium with secondary ionic imbalances re¬ sulting in cellular edema, dysfunction, and necrosis.19 The neuronal toxicity of
oral glutamate has been examined in the retina and has been shown to pro¬ duce severe damage to the inner reti¬ nal layers, necrosis of ganglion cells, and extrusion of photoreceptor nuclei toward the RPE.2" Although this mechanism of intravitreal dexametha¬ sone toxic reaction remains specula¬ tive, it is supported by the observa¬ tions that (1) the exclusive locus of
Downloaded From: http://archopht.jamanetwork.com/ by a Western University User on 06/11/2015
glutamine synthetase in the retina is the Müller cell and (2) levels of this enzyme have been shown to be in¬ creased after dexamethasone adminis¬ tration in other retinal experiments. The second possible mechanism of retinal toxic reaction relates to the demonstrated ability of glucosteroids to inhibit glucose oxidation, glucose uptake, glucose transport, and glucose-6-phosphate formation in many cells.21'25 It is well established that Müller cells have a high glycogen con¬ tent and are an important source of glucose for other retinal neurons.13 Consequently, possible impairment of glucose transport and metabolism in the Müller cell by dexamethasone might result in dysfunction of other retinal cells. The clearance of intraocular dexa¬ methasone sodium phosphate has been measured using tritium labeling and was found to be approximately 3 hours.1 Tritium labeling provides high sensitivity and is very specific; howev¬ er, it requires radioactive prepara¬ tions, which have limited use in a clinical setting. This study employed HPLC to identify and quantitate dexa¬ methasone sodium phosphate in sam¬ ples of vitreous; this method has been shown to be sensitive and specific for steroid detection and quantification.8 The half-life of dexamethasone sodium phosphate in the rabbit vitreous was 3.48 hours, with disappearance of de¬ tectable drug by 3 days. This result compares favorably with the 3-hour half-life determined by studies employ¬ ing tritium labeling. The dexametha¬ sone sodium phosphate in the anterior chamber attained its highest concen¬ tration at 1.5 hours and subsequently measured approximately 2.3% of that in the vitreous body at any time be¬ yond the first few hours after injec¬ tion. This finding, in accordance with mathematical modeling of intraocular drug turnover as presented by Mau¬ rice,26 suggests that only a small pro¬ portion of dexamethasone sodium phosphate is cleared from the vitreous via the anterior route. The half-life of paraben in the vitreous was 0.42 hours, and no pathologic changes were observed with morphologic study. In a previous study of the toxicity of an intravitreal injection of 400 µg of dexamethasone sodium phosphate in rabbits, no abnormalities were de¬ tected by ophthalmoscopy, electroreti¬ nography, or light microscopy up to 2 months after injection; in addition, the intraocular pressure remained nor¬ mal.1 A second study examined intra¬ vitreal injection of a variety of vehicles in commercial steroid preparations in
12
3
4
Time, h Fig 8.—Graphic representation of total aque¬ ous dexamethasone sodium phosphate con¬ tent at different intervals following intravitreal injection of 440 µg of dexamethasone sodium phosphate. Aqueous dexamethasone sodi¬ um phosphate reached its highest level 1.5 hours after injection and was undetectable at 12 hours.
4009 300 200 j 100 0
L
10
20
30
Time, h
Fig 9.—Clearance of 440 µg of dexametha¬ sodium phosphate (open figures) and 150 µg of paraben (closed figures) from the entire vitreous body after separate intravitreal injection in separate eyes. The half-life of the dexamethasone sodium phosphate was 3.48 sone
hours, and it was undetectable at 72 hours. The half-life of the paraben was 0.42 hours, and it was undetectable at 3 hours.
the rabbit and found no toxic reaction after a 0.1-mL injection of Decadron phosphate vehicle; however, a doublestrength preparation of vehicle caused a localized area of retinal edema and degeneration, and the authors consid¬ ered both the osmolarity and the pre¬ servatives as possible causes of the observed changes.27 Intravitreal injec¬ tion of LO mg of triamcinolone aceto¬ nide in rabbits produced no abnormali¬ ties on ophthalmoscopy, light and transmission electron microscopy, or electroretinography at 3 months.28 An
electroretinographic study following an intravitreal injection of 5 mg of
Fig 7.—Light microscopic anti-glial fibrillary acidic protein (anti-GFAP) immunohistochemistry. Top, Normal retina with nonimmune IgG control. There is no staining. Center, Normal retina with anti-GFAP. There is faint staining of the retina. Bottom, Specimen 1 week after intravitreal injection of 2000 µg of dexamethasone sodium phosphate; heavy labeling of the Müller cells in the inner retina is present. The staining extends from the vitreous cavity to the outer plexiform layer. ILM indicates internal limiting membrane; BM, Bruch's membrane (immunoperoxidase staining, original magnification 1800).
Downloaded From: http://archopht.jamanetwork.com/ by a Western University User on 06/11/2015
betamethasone acetate in the rabbit (with or without vitrectomy) showed a transient supernormality of the bwave, which recovered to normal with¬ in 20 days or less.29 Clinical cases of inadvertent intra¬ ocular injection of steroids have been reported. In an early series of six cases, retinal detachment developed in
three eyes following the injection, with visual results of 20/200 or worse, and in one eye injected with triamcinolone extensive pigmentary retinal changes developed in association with pre¬ retinal fibrosis and loss of light percep¬ tion. However, two eyes injected with methylprednisolone acetate (20 and 40 mg) had gradual résorption of the drug without evident toxic effects.30 A second report following the accidental intravitreal injection of methylpredni¬ solone acetate (removed with vitrec¬ tomy) documented only transient re¬ duction in the b-wave with no other adverse effects noted.31 In another re¬ port, an eye with retinal detachment and accidental intravitreal corticoste¬ roid injection was managed successful¬ ly with vitrectomy and without docu¬ mented steroid toxic effect.32 The inadvertent injection of corticosteroid into the choroidal vasculature has also been reported, with full recovery of visual acuity and without evident reti¬ nal abnormalities.*3 Resolution of cys¬ toid macular edema with improvement of visual acuity was noted by one au¬ thor following an accidental injection of steroid into the eye of a patient with cystoid macular edema.34 The author later performed intentional intravit¬ real steroid injections in nine addition¬ al eyes and noted similar improvement in eight eyes. The use of intravitreal steroid injec¬ tions for endophthalmitis has been evaluated in experimental and clinical studies. The potential advantages of this adjunctive therapy include reduc¬ tion of the harmful effects of inflamma¬ tion within the eye to preserve the integrity of the retina and other tis¬ sues; the potential disadvantages in¬ clude the possibility of suppressing mechanisms for infection control and the risk of drug toxic reaction from steroids, either alone or in combination with antibiotics. In an experimental model of Pseudomonas aeruginosa en¬ dophthalmitis in the rabbit, treatment with intravitreal dexamethasone (360 µg) and gentamicin was superior to treatment with gentamicin alone.1 Similarly, in a rabbit model of Staphylococcus epidermidis endophthalmitis, the use of an intravitreal injection of 400 µg of dexamethasone sodium phos¬ phate in association with intravitreal antibiotics produced superior results compared with antibiotic therapy alone.4 Although clinical information on the role of intravitreal steroid therapy in the treatment of endophthalmitis in patients is sparse, its successful use has been reported.23 The precise role of intravitreal steroid injection in the
endophthalmitis in pa¬ tients is unknown, and specific recom¬ mendations must await additional ex¬ perimental and clinical information. This study establishes the lack of retinal toxic effect of a single intra¬ vitreal injection of 440 µg of dexameth¬ asone sodium phosphate, either pre¬ pared as a pure solution or in a com¬ mercially supplied preparation, but documents increasing retinal toxic re¬ action with higher doses. The structur¬ al features suggest a primary Müller cell site of action for this drug. Further work on the mechanism of retinal toxic reaction of dexamethasone is required to clarify the role of glutamine and/or glucose oxidation alterations in the ob¬ served structural changes. These ob¬ servations may assist the investigation of the role of intravitreal dexametha¬ sone therapy in the treatment of en¬ dophthalmitis and other ocular inflam¬ matory diseases. treatment of
The authors have no proprietary interest in dexamethasone or its various derivatives dis¬ cussed in this article.
1. Graham RO, Peyman GA. Intravitreal injection of dexamethasone: treatment of experimentally induced endophthalmitis. Arch Ophthalmol.
1974;92:149-154. 2. Peyman GA, Herbst R. Bacterial endophthalmitis: treatment with intraocular injection of genta1974;91:416-418.
3. Diamond JG. Intraocular
dophthalmitis:
1987;120:1179-1183. 17. Riepe RE, Norenberg MD. M\l=u"\llercell locali-
sation of glutamine synthetase in rat retina. Nature. 1977;268:654-655. 18. Robinson MB, Coyle JT. Glutamate and related excitatory neurotransmitters: from basic science to clinical application. FASEB J. 1987;1:446\x=req-\ 455. 19. Price MT, Olney J, Samson L, Labruyere J. Calcium influx accompanies but does not cause excitotoxin-induced neuronal necrosis in the retina. Brain Res Bull. 1985;14:369-376. 20. Lucas DR, Newhouse JP. The toxic effect of sodium L-glutamate on the inner layers of the retina. Arch Ophthalmol. 1957;58:193-201. 21. Livingston JN, Lockwood DH. Effect of glucocorticoids on the glucose transport system of isolated fat cells. J Biol Chem. 1975;250:8353-8360. 22. Olefsky JM. Effect of dexamethasone on insulin binding, glucose transport, and glucose oxidation of isolated rat adipocytes. J Clin Invest.
1975;56:1499-1508.
References
micin and dexamethasone. Arch
13. Kuwabara T, Cogan DG. Retinal glycogen. Arch Ophthalmol. 1961;66:680-688. 14. Piddington R. Hormonal effects on the development of glutamine synthetase in the embryonic chick retina. Dev Biol. 1967;16:186-188. 15. Patejunas G, Young AP. Constitutive and glucocorticoid-mediated activation of glutamine synthetase gene expression in the developing chick retina. J Biol Chem. 1990;265:15280-15285. 16. Max SR, Thomas JW, Banner C, Vitkovic L, Konagaya M, Konagaya Y. Glucocorticoid receptor-mediated induction of glutamine synthetase in skeletal muscle cells in vitro. Endocrinology.
Ophthalmol.
management of en-
systematic approach. Arch Ophthalmol. 1981;99:96-99. 4. Meredith TA, Aguilar E, Miller MJ, Gardner SK, Trabelsi A, Wilson LA. Comparative treatment of experimental Staphylococcus epidermidis endophthalmitis. Arch Ophthalmol. 1990;108:857\x=req-\ a
860. 5. Tano Y, Sugita G, Abrams G, Machemer R. Inhibition of intraocular proliferations with intravitreal corticosteroids. Am J Ophthalmol. 1980; 89:131-136. 6. Tano Y, Chandler D, Machemer R. Treatment of intraocular proliferation with intravitreal triamcinolone acetonide. Am J Ophthalmol.
1980;90:810-816.
7. Chandler DB, Rozakis G, deJuan E, Machemer R. The effect of triamcinolone acetonide on a refined experimental model of proliferative vitreoretinopathy. Am J Ophthalmol. 1985;99:686\x=req-\ 690. 8. Kwak HW, Oum BS, Baker PA, Wolfe JK, D'Amico DJ. Intravitreal dexamethasone sodium phosphate: a new method for evaluation of drug clearance by HPLC. Invest Ophthalmol Vis Sci.
1989;30(suppl):510.
9. Newman EA. Distribution of potassium conductance in mammalian M\l=u"\ller(glial) cells: a comparative study. J Neurosci. 1987;7:2423-2432. 10. Ehinger B. Glial and neuronal uptake of GABA, glutamic acid, glutamine, and glutathione in the rabbit retina. Exp Eye Res. 1977;25:221-234. 11. Gerschenfeld HM, Wald F, Zadunaisky JA, DeRobertis EDP. Function of astroglia in the water-ion metabolism of the central nervous system.
Neurology. 1959;9:412-425. 12. Sarthy PV, Lam DMK. Biochemical studies of isolated glial (M\l=u"\ller)cells from the turtle retina.
J Cell Biol. 1978;78:675-684.
Downloaded From: http://archopht.jamanetwork.com/ by a Western University User on 06/11/2015
23. Garvey WT, Huecksteadt TP, Monzon R, Marshall S. Dexamethasone regulates the glucose transport system in primary cultured adipocytes: different mechanisms of insulin resistance after acute and chronic exposure. Endocrinology.
1989;124:2063-2074. 24. Carter-Su C, Okamoto K. Inhibition of hexose transport in adipocytes by dexamethasone: role of protein synthesis. Am J Physiol. 1985;248:E215\x=req-\
E223. 25. man
Haynes RC. Adrenocortical steroids.
In: Gil-
AG, Rall TW, Nies AS, Taylor P, eds. Good-
and Gilman's the Pharmacological Basis of 8th ed. Elmsford, NY: Pergamon Press Inc; 1990:1436-1444. 26. Maurice DM. Injection of drugs into the vitreous body. In: Leopold IH, Burns RP, eds. Symposium on Ocular Therapy. New York, NY: John Wiley & Sons Inc; 1976:59-71. 27. Hida T, Chandler D, Arena JE, Machemer R. Experimental and clinical observations of the intraocular toxicity of commercial corticosteroid preparations. Am J Ophthalmol. 1986;101:190-195. 28. McCuen BW, Bressler M, Tano Y, Chandler D, Machemer R. The lack of toxicity of intravitreally administered triamcinolone acetonide. Am J man
Therapeutics.
Ophthalmol. 1981;91:785-788.
29. Koizumi K, Honda Y. ERG changes after an injection of corticosteroid into the vitreous. Metab Ophthalmol. 1985;8:13-16. 30. Schlaegel TF, Wilson FM. Accidental intraocular injection of depot corticosteroids. Trans Am Acad Ophthalmol Otolaryngol. 1974;78:847-855. 31. Andrew NC, Gregor ZJ. Intraocular injection of Depomedrone. Br J Ophthalmol. 1986;
70:298-300. 32. Zinn KM. Iatrogenic intraocular injection of depot corticosteroids and its surgical removal using the pars plana approach. Ophthalmology. 1981; 88:13-17. 33. McLean EB. Inadvertent injection of corticosteroid into the choroidal vasculature. Am J
Ophthalmol. 1975;80:835-837.
34. Michael MJ. Intraocular Kenalog injections therapeutic in some CME cases. Ocular Surg News. November 1,1984:50. seen
\s=b\ The intravitreal injection of steroids may be potentially useful in the treatment of endophthalmitis and other ocular inflammatory diseases. The retinal toxicity and intraocular turnover of aqueous solutions of dexamethasone sodium phosphate in doses ranging from 440 to 4000 \g=m\g were evaluated in the rabbit; evaluation was also performed for a 0.1-mL injection of a commercially available preparation (dexamethasone phosphate [Decadron] injection, 4 mg/mL). After the 440-\g=m\g dose, a transient increase in staining of the M\l=u"\llercells was observed, which normalized after 2 days. Progressively higher doses resulted in an increasing spectrum of disorganization in M\l=u"\llerand other retinal cells. The half-life of the intravitreally injected drug was 3.48 hours. These findings suggest a primary interference in M\l=u"\llercell function, possibly through dexamethasoneinduced alterations in retinal glutamate or glucose metabolism.
(Arch Ophthalmol. 1992;110:259-266) ""1 he intravitreal injection of steroid has been evaluated for the treat¬ ment of endophthalmitis1"4 and
ative
prolifer¬
and may po¬ tentially be useful in the treatment of other ocular diseases. In the treatment of endophthalmitis, the intravitreal in¬ jection of steroids has been beneficial in certain animal models,1-4 and its suc¬ cessful clinical use in association with intravitreal antibiotic injections and vi¬ trectomy has been reported.2,3 The ra¬ tionale for intravitreal steroid adminis¬ tration in endophthalmitis is based on the need to preserve the retina and other intraocular tissues from the po¬ tentially devastating and rapidly pro¬ gressive effects of inflammation associated with infection. These con¬ siderations suggest that intravit¬ real steroid injection may be a reason¬ able adjunct to initial intravitreal anti-
vitreoretinopathy5"7
Accepted for publication August 16,
1991. From the Retina Service (Dr D'Amico) and the Howe Laboratory of Ophthalmology (Drs Kwak and D'Amico), Massachusetts Eye and Ear Infirmary, Department of Ophthalmology, Harvard Medical School, Boston. Reprint requests to Massachusetts Eye and Ear Infirmary, 243 Charles St, Boston, MA 02114
(Dr D'Amico).
biotic
therapy. In the management of experimental proliferative vitreoreti¬ nopathy, intravitreal steroid therapy has demonstrated
a
beneficial effect in
reducing the severity of intraocular cellular proliferation and subsequent retinal detachment.
Dexamethasone, a synthetic steroid,
is widely used in the treatment of ocular inflammatory diseases. Despite considerable clinical interest in intra¬ vitreal dexamethasone therapy for en¬ dophthalmitis and other ocular inflam¬ matory disorders, the ocular toxic re¬ action and drug turnover following intravitreal injection of dexametha¬ sone and related compounds have re¬ ceived limited study.1 Dexamethasone alone is poorly suited to intravitreal use because it has limited solubility in aqueous media, and this property would require the injection of poten¬ tially toxic solvent vehicles into the eye. However, dexamethasone sodium phosphate has a high degree of solu¬ bility and stability in aqueous solution; furthermore, commercial preparations are readily available. Herein, we de¬ scribe dose-related effects of intra¬ vitreal dexamethasone sodium phos¬ phate on the retina of rabbits and report the half-life of this drug after injection into the vitreous. MATERIALS AND METHODS Intraocular Toxic Reaction All studies were performed using Dutch Belted pigmented rabbits (weight range, 2.5 to 3.5 kg). They were housed in sepa¬ rate cages with a 12-hour light-dark cycle. All animal handling and procedures con¬ formed to the guidelines of the Association for Research in Vision and Ophthalmology. The toxicities of (1) pure dexamethasone sodium phosphate in aqueous solution; (2) dexamethasone phosphate (Decadron) injection, a commercially available prepara¬ tion (Merck Sharp & Dohme, West Point, Pa); and (3) Decadron phosphate injection vehicle were evaluated after intravitreal injection. The following doses were studied: (1) in the dexamethasone sodium phosphate study group, 440, 800, 1200, 1600, 2000, and 4000 µg (a total of 56 eyes); (2) in the Decadron phosphate injection study group, a dose of 0.1 mL of commercially available 4 mg/mL preparation (this dose contains 440 µg of dexamethasone sodium phosphate after correction for the appropriate molecu-
Downloaded From: http://archopht.jamanetwork.com/ by a Western University User on 06/11/2015
lar weight of the salt form; a total of 10 eyes included in this group); and (3) in the Decadron phosphate injection vehicle group, 0.1 mL (six eyes). Two additional control groups were evaluated, and these consisted of 10 saline-injected and 10 uninjected eyes. The rabbits were anesthetized with a 1.5-mL mixture of equal parts of xylazine hydrochloride (20 mg/mL) and ke¬ tamine hydrochloride (100 mg/mL) by an intramuscular injection, and their pupils were dilated with 0.5% topical tropicamide and 5% phenylephrine hydrochloride drops. Before intravitreal injections, an anterior chamber paracentesis was performed. The drugs were injected into the vitreous using a 1-mL syringe with a 30-gauge needle inserted 2 to 3 mm posterior to the limbus. All drug injections were performed in one eye of each animal; the other eye served as one of the controls and was either left uninjected or injected with 0.1 mL of saline solution as described above. All solutions were passed through a 0.2-µ filter before were
injection.
Rabbits were periodically examined by indirect ophthalmoscopy and photographed after pupillary dilation. They were killed at intervals after the injection ranging from 3 hours to 3 weeks. The eyes were hemisected at the equator, the vitreous was removed, and the posterior portion of the eyes was fixed in 2.5% glutaraldehyde for 24 hours. For light and transmission elec¬ tron microscopy, the tissue was dissected into 2-mm2 fragments, postfixed in 1% os¬ mium tetroxide, dehydrated in graded alco¬ hols, and embedded in epoxy resin. Semithin sections were stained with méthy¬ lène blue-azure II for examination by light microscopy, and ultrathin sections were stained with uranyl acetate-lead citrate. For scanning electron microscopy, the reti¬ na was divided into sensory retina and retinal pigment epithelium (RPE). Speci¬ mens were then postfixed in buffered 1% osmium tetroxide and dehydrated through graded ethanols. The specimens were criti¬ cal point dried with carbon dioxide, mount¬ ed on stubs, and coated with gold-
palladium. For light-microscopic immunocytochemistry for glial fibrillary acidic protein (GFAP), the eyes were hemisected at the equator and the vitreous removed with
filter paper. The posterior portion of the globe was fixed in 4% paraformaldehyde in 0.1 mol/L of sodium phosphate buffer (pH 7.4) for 1 hour at room temperature and for an additional 12 hours at 4°C. After fixa¬ tion, specimens were incubated in 5% su¬ crose in sodium phosphate buffer for 1 hour and then immersed in 30% sucrose for 12 hours. Tissues were embedded in OCT
Fig 1 Fundus photographs after the intravitreal injection of 4000 µg of dexamethasone sodium phosphate. Left, Three days after injection, retinal edema and serous detachment are shown. Right, One week after injection, the retinal edema —
.
has resolved.
Fig 2.—Light micrograph after intravitreal dexamethasone sodium phosphate injection. Top left, Six hours after injection of 440 µg of dexamethasone sodium phosphate, conspicuous staining of Müller cells is observed (arrows). Top right, Two days after injection of 440 µ of dexamethasone sodium phosphate, Müller cell staining has normalized. Bottom left, One week after injection of 2000 µg of dexamethasone sodium phosphate, a vacuole is noted at the border of the outer plexiform and outer nuclear layers (arrow). Bottom right, Three days after injection of 4000 µg of dexamethasone sodium phosphate, marked perinuclear edema of photoreceptor and other neuronal cells is shown. BM indicates Bruch's membrane (méthylène blue-azure II stain, original magnification 1800). (Miles Scientific, Naperville, 111), and 5-µ cryostat sections were prepared and main¬
tained at -20°C. The sections were covered by a drop of normal goat serum (Cappel, Organon Tecknika Corp, West Chester, Pa) diluted 1:50 with phosphate-buffered saline
(PBS) for 20 minutes. The normal goat blotted from the sections, which
serum was
then incubated 45 minutes with mouse monoclonal anti-GFAP (Sigma Chemical Co, St Louis, Mo). The anti-GFAP was diluted 1:250 with PBS and contained 1%
were
Downloaded From: http://archopht.jamanetwork.com/ by a Western University User on 06/11/2015
bovine serum albumin. After a 10-minute rinse in PBS, the sections were incubated with biotinylated goat anti-mouse IgG (1:1000 preparation; Jackson Immunoresearch Labs Ine, Avondale, Pa) for 45 min¬ utes. The slides were subsequently rinsed
Fig
3.—Transmission electron
micrograph
1 week after intravitreal
injection of
2000 µg of dexamethasone sodium
phosphate. A large round vacuole (asterisk) appears between the outer plexiform and outer nuclear layers (OPL and ONL).
The vacuole was located inside the synaptic vesicle at the receptor site of the nuclear layer (uranyl acetate-lead citrate, original magnification 3300). in PBS for 10 minutes and incubated with
peroxidase-conjugated streptavidin (1:1000 preparation; Jackson Immunoresearch Labs) for 45 minutes. After rinsing for 10 minutes in PBS, incubating with 3-amino-9ethylcarbazole peroxidase substrate solu¬ tion (Polysciences Ine, Washington, Pa) for 10 minutes, and rinsing in distilled water, a coverglass was applied to the sections with gelvatol.
Intraocular Clearance Before intravitreal injections, an anterior paracentesis was performed. All injections were performed slowly with the bevel of the needle facing anteriorly. In one group, dexamethasone sodium phosphate (440 µg in 0.1 mL of preservative-free sterile wa¬ ter) was injected through the pars plana, 2 mm posterior to the limbus, using a 30-gauge needle. In a second group, the major component of Decadron phosphate
injection vehicle, paraben (150 µg), was similarly injected. The eyes were enucleat¬ ed immediately following death at 1.5, 3, 6, 12, 24, 48, and 72 hours after injection. Aqueous specimens were collected with a
photoreceptor terminal. INL indicates inner
25-gauge needle. The eyes were subse¬ quently immersed in liquid nitrogen, and the frozen vitreous
was
removed. The
re¬
sulting samples were analyzed by highperformance liquid chromatography (HPLC).8 After homogenization, the speci¬ mens were applied to a Sep-Pak C-18 car¬ tridge (Waters Associates, Milford, Mass) and eluted. The eluate was evaporated to dryness by a speed vacuum concentration. Each residue was then resuspended in 1.0 mL of methanol and vortexed. Chroma¬
tography was performed on a Waters µ Bondapak C18 column (Waters Associates, Milford, Mass) run isocratically at 1.0 mL/min using 50:50 methanol-water (vol/vol)
as a
solvent. Graded amounts of
prednisone were added to each specimen to
as a recovery marker and HPLC standard. Similarly, HPLC was used to determine paraben concentrations by mea¬ suring recovery and location of added para¬ ben. The limits of detection for the total vitreous content by HPLC were 5.2 ±1.7 µg for dexamethasone sodium phosphate and 8.7 ±2.8 µg for paraben. serve
Downloaded From: http://archopht.jamanetwork.com/ by a Western University User on 06/11/2015
RESULTS Intraocular Toxic Reaction
The cornea and lens of all experi¬ mental and control eyes appeared nor¬ mal by slit-lamp examination. The vit¬ reous and retina appeared normal by indirect ophthalmoscopy in vehicle-in¬ jected and 0.1 mL Decadron phospha¬ te-injected eyes. The dexamethasone sodium phosphate-injected eyes ap¬ peared normal in doses up to 2000 µg; however, after the 4000^g dexameth¬ sodium phosphate injection, asone there was vitreous clouding, retinal edema, and serous detachment within 24 hours (Fig 1). These findings re¬ solved within 2 weeks, and all experi¬ mental eyes appeared normal to oph¬ thalmoscopy at that time (Fig 1). Light microscopy and ultrastructural evalua¬ tion showed no evidence of retinal ab¬ normality in the vehicle-injected, sa¬ line-injected, or uninjected eyes.
Three days after intravitreal injec¬ tion of 4000 µg of dexamethasone sodi¬ um phosphate, the photoreceptor cells in the ONL and neuronal cells in the inner nuclear layer showed prominent swelling (Fig 2, bottom right; Fig 4) in addition to the amorphous granules on the surface of the RPE. Photorecep¬ tors also were degenerated, with disin¬ tegration of the regular lamellar pat¬ tern of their outer segment discs in association with necrotic debris (Fig 5). The Müller cells displayed vacuolar and degenerative changes in the cyto¬ plasm. After 1 week, the RPE cells showed marked melanin granules (Fig 6, top) in association with the amor¬ phous granules on the RPE surface
(Fig 6, bottom). By light microscopic immunohisto¬ chemistry, anti-GFAP labeling was positive 1 week after injection of 2000 µg of dexamethasone sodium phos¬ phate compared with uninjected con¬ trols (Fig 7). The reaction product appeared as strands extending from the internal limiting membrane to the OPL, and there was a specific increase in labeling of the Müller cells. Intraocular Clearance
The ocular clearance of intravitreal
injections of 440 µg of dexamethasone sodium phosphate or 0.1 mL of para¬ ben vehicle are shown in Figs 8 and 9;
Fig 4.—Transmission electron micrograph 3 days after intravitreal injection of 4000 µg of dexamethasone sodium phosphate. Photoreceptors and other neuronal cells display promi¬ nent swelling. INL indicates inner nuclear layer; OPL, outer plexiform layer; and ONL, outer nuclear layer (uranyl acetate-lead citrate, original magnification 6900). After the 0.1-mL Decadron phos¬ or the 440^g dexamethasone sodium phosphate injection, micro¬ scopie examination revealed marked staining in the Müller cells compared with the other cells (Fig 2, top left). This staining was prominent at 6 hours after injection and then decreased and normalized at 2 days (Fig 2, top right). Apart from the increased staining de¬ scribed, no structural abnormalities were noted at any time with light or transmission electron microscopy after this dosage level. Minimal changes were seen after intravitreal injection of 800 µg of dexa¬ methasone sodium phosphate. After 3 days, small vacuoles appeared be¬ tween the outer plexiform layer (OPL) and the outer nuclear layer (ONL); these vacuoles disappeared at 1 week. Intravitreal injection of 1200 µg of dexamethasone sodium phosphate pro¬ duced more distinct vacuoles in the same sites. In addition, an accumula¬ tion of amorphous granules ranging
phate
from 2 to 10 µ appeared on the surface of the RPE and in the interphotoreceptor matrix after 1 day. These granules appeared to be mem¬ brane limited and were still evident at 3 weeks. Disintegration of the photore¬ ceptor outer segments was noted, and photoreceptor cells in the ONL were edematous. After 3 weeks, the edema of the photoreceptor cells subsided, but occasional necrotic photoreceptor cells were observed. Large round vacuoles appeared be¬ tween OPL and ONL 3 days after the intravitreal injection of 1600 to 4000 µg of dexamethasone sodium phosphate (Fig 2, bottom left), and the size of the vacuoles did not change during 3 weeks. Ultrastructurally, vacuoles were located inside the synaptic vesi¬ cle primarily at the receptor site of the photoreceptor terminal (Fig 3). In some cases, the vacuoles appeared to be fused. By scanning electron micros¬ copy, the RPE showed pleomorphism and irregularity of apical cell border.
Downloaded From: http://archopht.jamanetwork.com/ by a Western University User on 06/11/2015
the values represent the total amount of dexamethasone sodium phosphate or paraben present in the entire aque¬ ous or vitreous body. Dexamethasone sodium phosphate in the aqueous reached its highest level (8.2 ±0.7 µg) in 1.5 hours after injection but was not detected in the aqueous at 12 hours. Vitreous dexamethasone sodium phos¬ phate content at initial injection was 387.5 ±16.9 (ig; at 3 hours, 188.6 ±6.3 µg; at 6 hours, 126.1 ±39.8 µg; and at 24 hours, 5.7 ±2.2 µg. At 72 hours, dexamethasone sodium phosphate was not detected in the vitreous; the calcu¬ lated half-life of the intravitreal dexa¬ methasone sodium phosphate was 3.48 hours. The paraben was not detected in the aqueous after intravitreal injec¬ tion. The initial vitreous paraben con¬ tent was 84.3 ± 16.2 µg, and at 3 hours the paraben was not detectable in the vitreous; the calculated half-life of in¬ travitreal paraben was 0.42 hours. COMMENT
This study demonstrates the lack of retinal toxic reaction following an in¬ travitreal injection of 440 µg of dexa¬ methasone sodium phosphate or the equivalent dose of a commercially available, soluble dexamethasone sodi¬ um
phosphate preparation (Decadron
ry site of intravitreal dexamethasone toxic reaction is the Müller cell. It is well established that the structural and nutritional support systems of the retina are supplied predominantly by Müller cells. Müller cells also play a prominent role in maintaining an envi¬ ronment required for neuronal func¬ tion. In this regard, the documented functions of the Müller cells are exten¬ sive and include the clearance of potas¬ sium released by neuronal activity or
by pathologic conditions,9 the uptake
and catabolism of neurotransmitters axon terminals,10 the control of water movement between intracellular and extracellular com¬ partments and tissue pH,1112 and, most importantly, the provision of glucose.13 Consequently, the Müller cell is sus¬ ceptible to abnormality by many fac¬ tors, including mechanical disruption, metabolic abnormalities, and drugs. The mechanism of the observed ef¬ fects of intravitreal dexamethasone so¬ dium phosphate on the Müller cell is unknown. However, steroids are known to influence at least two specific pathways present in Müller cells, and these interactions suggest possible mechanisms for the toxic reaction ob¬ served in the present study. The first interaction centers on the observation that dexamethasone treat¬ ment produces an increase in the level of glutamine synthetase in a variety of cells and specifically in Müller cells. In the retina of the late chick embryo, a rapid and striking increase in gluta¬ mine synthetase has been reported fol¬ lowing systemic elevation of adrenal corticoid levels.14 Furthermore, in the retina and other cell systems, it has been specifically established that dexa¬ methasone enhances the level of gluta¬ mine synthetase messenger RNA.15,16 In the retina, this enzyme has been shown to be exclusively confined to Müller cells." The pathways for this enzyme have been partially character¬ ized in the central nervous system.18 Glutamine synthetase is confined to glial cells and functions to convert glu¬ tamate (a putative excitatory neuro¬ transmitter and potential excitotoxin to neural cells) to glutamine (a neurophysiologically inactive substance). As a result of the action of glutamine synthetase, glutamine produced in gli¬ al cells is transported into neuronal cells, where glutamate is again pro¬ duced and sequestered in glutaminergic synaptic vesicles through the action of glutaminase. It is, therefore, possible that intra¬ vitreal adrenocorticoids activate gluta¬ mine synthetase in Müller cells with a resulting increase in levels of neuronal
discharged by
Fig 5. —Transmission electron micrograph 3 days after injection of 4000 µg of dexamethasone sodium phosphate. Photoreceptors display degenerative changes. BM indicates Bruch's membrane (uranyl acetate-lead citrate, original magnification 4670).
phosphate injection) that includes par¬ aben in its vehicle. Although in our experimental study the Müller cells showed increased méthylène blueazure II staining with this dose com¬ pared with other retinal cells, this finding was reversible and is sugges¬ tive of reactive rather than degenera¬ tive change. However, with 800^g and higher doses of the primary com¬ ponent dexamethasone sodium phos¬ phate, the Müller cells displayed an increasing toxic reaction manifested by edema and degeneration. In addition, an increasing spectrum of changes was
also noted in other retinal cells. When doses ranging from 800 to 4000 µg of
dexamethasone sodium phosphate injected into the vitreous, a doserelated increase in the number of vacu¬ oles between OPL and ONL was ob¬ served. The vacuoles were more prominent after injection of 1600 to 4000 µg and were located inside the neuronal synapses. In addition, after intravitreal injection of dexametha¬ sone sodium phosphate at 1200-µg and higher doses, the photoreceptor outer segments showed edema and disinte¬ gration, and there was accumulation of amorphous granules on the surface of the RPE or in the interphotoreceptor matrix. These findings suggest that a prima-
were
Downloaded From: http://archopht.jamanetwork.com/ by a Western University User on 06/11/2015
Fig 6. —Top, Transmission electron micrograph 1 week after injection of 4000 µg of dexameth¬ sodium phosphate. An abnormal accumulation of amorphous granules (arrows) ranging from 2 to 10 µ are observed on the surface of the retinal pigment epithelium (RPE) or in the interphotoreceptor matrix. These granules appeared to be membrane limited and did not disappear at 3 weeks. The RPE cells show prominent melanin pigment. BM indicates Bruch's membrane (uranyl acetate-lead citrate, original magnification 5100). Bottom, Scanning electron micrograph of the RPE surface 1 week after injection of 4000 µg of dexamethasone sodium phosphate. Amorphous granules (arrows) are observed in association with heteroge¬ neity of the RPE (gold-palladium coated, original magnification 2040). asone
glutamine and subsequently neuronal glutamate; the resulting increase may produce cellular dysfunction and toxic
reaction. Potential toxic effects that appear to be mediated by glutamate in the retinal neurons include an increase in neuronal permeability to sodium with secondary ionic imbalances re¬ sulting in cellular edema, dysfunction, and necrosis.19 The neuronal toxicity of
oral glutamate has been examined in the retina and has been shown to pro¬ duce severe damage to the inner reti¬ nal layers, necrosis of ganglion cells, and extrusion of photoreceptor nuclei toward the RPE.2" Although this mechanism of intravitreal dexametha¬ sone toxic reaction remains specula¬ tive, it is supported by the observa¬ tions that (1) the exclusive locus of
Downloaded From: http://archopht.jamanetwork.com/ by a Western University User on 06/11/2015
glutamine synthetase in the retina is the Müller cell and (2) levels of this enzyme have been shown to be in¬ creased after dexamethasone adminis¬ tration in other retinal experiments. The second possible mechanism of retinal toxic reaction relates to the demonstrated ability of glucosteroids to inhibit glucose oxidation, glucose uptake, glucose transport, and glucose-6-phosphate formation in many cells.21'25 It is well established that Müller cells have a high glycogen con¬ tent and are an important source of glucose for other retinal neurons.13 Consequently, possible impairment of glucose transport and metabolism in the Müller cell by dexamethasone might result in dysfunction of other retinal cells. The clearance of intraocular dexa¬ methasone sodium phosphate has been measured using tritium labeling and was found to be approximately 3 hours.1 Tritium labeling provides high sensitivity and is very specific; howev¬ er, it requires radioactive prepara¬ tions, which have limited use in a clinical setting. This study employed HPLC to identify and quantitate dexa¬ methasone sodium phosphate in sam¬ ples of vitreous; this method has been shown to be sensitive and specific for steroid detection and quantification.8 The half-life of dexamethasone sodium phosphate in the rabbit vitreous was 3.48 hours, with disappearance of de¬ tectable drug by 3 days. This result compares favorably with the 3-hour half-life determined by studies employ¬ ing tritium labeling. The dexametha¬ sone sodium phosphate in the anterior chamber attained its highest concen¬ tration at 1.5 hours and subsequently measured approximately 2.3% of that in the vitreous body at any time be¬ yond the first few hours after injec¬ tion. This finding, in accordance with mathematical modeling of intraocular drug turnover as presented by Mau¬ rice,26 suggests that only a small pro¬ portion of dexamethasone sodium phosphate is cleared from the vitreous via the anterior route. The half-life of paraben in the vitreous was 0.42 hours, and no pathologic changes were observed with morphologic study. In a previous study of the toxicity of an intravitreal injection of 400 µg of dexamethasone sodium phosphate in rabbits, no abnormalities were de¬ tected by ophthalmoscopy, electroreti¬ nography, or light microscopy up to 2 months after injection; in addition, the intraocular pressure remained nor¬ mal.1 A second study examined intra¬ vitreal injection of a variety of vehicles in commercial steroid preparations in
12
3
4
Time, h Fig 8.—Graphic representation of total aque¬ ous dexamethasone sodium phosphate con¬ tent at different intervals following intravitreal injection of 440 µg of dexamethasone sodium phosphate. Aqueous dexamethasone sodi¬ um phosphate reached its highest level 1.5 hours after injection and was undetectable at 12 hours.
4009 300 200 j 100 0
L
10
20
30
Time, h
Fig 9.—Clearance of 440 µg of dexametha¬ sodium phosphate (open figures) and 150 µg of paraben (closed figures) from the entire vitreous body after separate intravitreal injection in separate eyes. The half-life of the dexamethasone sodium phosphate was 3.48 sone
hours, and it was undetectable at 72 hours. The half-life of the paraben was 0.42 hours, and it was undetectable at 3 hours.
the rabbit and found no toxic reaction after a 0.1-mL injection of Decadron phosphate vehicle; however, a doublestrength preparation of vehicle caused a localized area of retinal edema and degeneration, and the authors consid¬ ered both the osmolarity and the pre¬ servatives as possible causes of the observed changes.27 Intravitreal injec¬ tion of LO mg of triamcinolone aceto¬ nide in rabbits produced no abnormali¬ ties on ophthalmoscopy, light and transmission electron microscopy, or electroretinography at 3 months.28 An
electroretinographic study following an intravitreal injection of 5 mg of
Fig 7.—Light microscopic anti-glial fibrillary acidic protein (anti-GFAP) immunohistochemistry. Top, Normal retina with nonimmune IgG control. There is no staining. Center, Normal retina with anti-GFAP. There is faint staining of the retina. Bottom, Specimen 1 week after intravitreal injection of 2000 µg of dexamethasone sodium phosphate; heavy labeling of the Müller cells in the inner retina is present. The staining extends from the vitreous cavity to the outer plexiform layer. ILM indicates internal limiting membrane; BM, Bruch's membrane (immunoperoxidase staining, original magnification 1800).
Downloaded From: http://archopht.jamanetwork.com/ by a Western University User on 06/11/2015
betamethasone acetate in the rabbit (with or without vitrectomy) showed a transient supernormality of the bwave, which recovered to normal with¬ in 20 days or less.29 Clinical cases of inadvertent intra¬ ocular injection of steroids have been reported. In an early series of six cases, retinal detachment developed in
three eyes following the injection, with visual results of 20/200 or worse, and in one eye injected with triamcinolone extensive pigmentary retinal changes developed in association with pre¬ retinal fibrosis and loss of light percep¬ tion. However, two eyes injected with methylprednisolone acetate (20 and 40 mg) had gradual résorption of the drug without evident toxic effects.30 A second report following the accidental intravitreal injection of methylpredni¬ solone acetate (removed with vitrec¬ tomy) documented only transient re¬ duction in the b-wave with no other adverse effects noted.31 In another re¬ port, an eye with retinal detachment and accidental intravitreal corticoste¬ roid injection was managed successful¬ ly with vitrectomy and without docu¬ mented steroid toxic effect.32 The inadvertent injection of corticosteroid into the choroidal vasculature has also been reported, with full recovery of visual acuity and without evident reti¬ nal abnormalities.*3 Resolution of cys¬ toid macular edema with improvement of visual acuity was noted by one au¬ thor following an accidental injection of steroid into the eye of a patient with cystoid macular edema.34 The author later performed intentional intravit¬ real steroid injections in nine addition¬ al eyes and noted similar improvement in eight eyes. The use of intravitreal steroid injec¬ tions for endophthalmitis has been evaluated in experimental and clinical studies. The potential advantages of this adjunctive therapy include reduc¬ tion of the harmful effects of inflamma¬ tion within the eye to preserve the integrity of the retina and other tis¬ sues; the potential disadvantages in¬ clude the possibility of suppressing mechanisms for infection control and the risk of drug toxic reaction from steroids, either alone or in combination with antibiotics. In an experimental model of Pseudomonas aeruginosa en¬ dophthalmitis in the rabbit, treatment with intravitreal dexamethasone (360 µg) and gentamicin was superior to treatment with gentamicin alone.1 Similarly, in a rabbit model of Staphylococcus epidermidis endophthalmitis, the use of an intravitreal injection of 400 µg of dexamethasone sodium phos¬ phate in association with intravitreal antibiotics produced superior results compared with antibiotic therapy alone.4 Although clinical information on the role of intravitreal steroid therapy in the treatment of endophthalmitis in patients is sparse, its successful use has been reported.23 The precise role of intravitreal steroid injection in the
endophthalmitis in pa¬ tients is unknown, and specific recom¬ mendations must await additional ex¬ perimental and clinical information. This study establishes the lack of retinal toxic effect of a single intra¬ vitreal injection of 440 µg of dexameth¬ asone sodium phosphate, either pre¬ pared as a pure solution or in a com¬ mercially supplied preparation, but documents increasing retinal toxic re¬ action with higher doses. The structur¬ al features suggest a primary Müller cell site of action for this drug. Further work on the mechanism of retinal toxic reaction of dexamethasone is required to clarify the role of glutamine and/or glucose oxidation alterations in the ob¬ served structural changes. These ob¬ servations may assist the investigation of the role of intravitreal dexametha¬ sone therapy in the treatment of en¬ dophthalmitis and other ocular inflam¬ matory diseases. treatment of
The authors have no proprietary interest in dexamethasone or its various derivatives dis¬ cussed in this article.
1. Graham RO, Peyman GA. Intravitreal injection of dexamethasone: treatment of experimentally induced endophthalmitis. Arch Ophthalmol.
1974;92:149-154. 2. Peyman GA, Herbst R. Bacterial endophthalmitis: treatment with intraocular injection of genta1974;91:416-418.
3. Diamond JG. Intraocular
dophthalmitis:
1987;120:1179-1183. 17. Riepe RE, Norenberg MD. M\l=u"\llercell locali-
sation of glutamine synthetase in rat retina. Nature. 1977;268:654-655. 18. Robinson MB, Coyle JT. Glutamate and related excitatory neurotransmitters: from basic science to clinical application. FASEB J. 1987;1:446\x=req-\ 455. 19. Price MT, Olney J, Samson L, Labruyere J. Calcium influx accompanies but does not cause excitotoxin-induced neuronal necrosis in the retina. Brain Res Bull. 1985;14:369-376. 20. Lucas DR, Newhouse JP. The toxic effect of sodium L-glutamate on the inner layers of the retina. Arch Ophthalmol. 1957;58:193-201. 21. Livingston JN, Lockwood DH. Effect of glucocorticoids on the glucose transport system of isolated fat cells. J Biol Chem. 1975;250:8353-8360. 22. Olefsky JM. Effect of dexamethasone on insulin binding, glucose transport, and glucose oxidation of isolated rat adipocytes. J Clin Invest.
1975;56:1499-1508.
References
micin and dexamethasone. Arch
13. Kuwabara T, Cogan DG. Retinal glycogen. Arch Ophthalmol. 1961;66:680-688. 14. Piddington R. Hormonal effects on the development of glutamine synthetase in the embryonic chick retina. Dev Biol. 1967;16:186-188. 15. Patejunas G, Young AP. Constitutive and glucocorticoid-mediated activation of glutamine synthetase gene expression in the developing chick retina. J Biol Chem. 1990;265:15280-15285. 16. Max SR, Thomas JW, Banner C, Vitkovic L, Konagaya M, Konagaya Y. Glucocorticoid receptor-mediated induction of glutamine synthetase in skeletal muscle cells in vitro. Endocrinology.
Ophthalmol.
management of en-
systematic approach. Arch Ophthalmol. 1981;99:96-99. 4. Meredith TA, Aguilar E, Miller MJ, Gardner SK, Trabelsi A, Wilson LA. Comparative treatment of experimental Staphylococcus epidermidis endophthalmitis. Arch Ophthalmol. 1990;108:857\x=req-\ a
860. 5. Tano Y, Sugita G, Abrams G, Machemer R. Inhibition of intraocular proliferations with intravitreal corticosteroids. Am J Ophthalmol. 1980; 89:131-136. 6. Tano Y, Chandler D, Machemer R. Treatment of intraocular proliferation with intravitreal triamcinolone acetonide. Am J Ophthalmol.
1980;90:810-816.
7. Chandler DB, Rozakis G, deJuan E, Machemer R. The effect of triamcinolone acetonide on a refined experimental model of proliferative vitreoretinopathy. Am J Ophthalmol. 1985;99:686\x=req-\ 690. 8. Kwak HW, Oum BS, Baker PA, Wolfe JK, D'Amico DJ. Intravitreal dexamethasone sodium phosphate: a new method for evaluation of drug clearance by HPLC. Invest Ophthalmol Vis Sci.
1989;30(suppl):510.
9. Newman EA. Distribution of potassium conductance in mammalian M\l=u"\ller(glial) cells: a comparative study. J Neurosci. 1987;7:2423-2432. 10. Ehinger B. Glial and neuronal uptake of GABA, glutamic acid, glutamine, and glutathione in the rabbit retina. Exp Eye Res. 1977;25:221-234. 11. Gerschenfeld HM, Wald F, Zadunaisky JA, DeRobertis EDP. Function of astroglia in the water-ion metabolism of the central nervous system.
Neurology. 1959;9:412-425. 12. Sarthy PV, Lam DMK. Biochemical studies of isolated glial (M\l=u"\ller)cells from the turtle retina.
J Cell Biol. 1978;78:675-684.
Downloaded From: http://archopht.jamanetwork.com/ by a Western University User on 06/11/2015
23. Garvey WT, Huecksteadt TP, Monzon R, Marshall S. Dexamethasone regulates the glucose transport system in primary cultured adipocytes: different mechanisms of insulin resistance after acute and chronic exposure. Endocrinology.
1989;124:2063-2074. 24. Carter-Su C, Okamoto K. Inhibition of hexose transport in adipocytes by dexamethasone: role of protein synthesis. Am J Physiol. 1985;248:E215\x=req-\
E223. 25. man
Haynes RC. Adrenocortical steroids.
In: Gil-
AG, Rall TW, Nies AS, Taylor P, eds. Good-
and Gilman's the Pharmacological Basis of 8th ed. Elmsford, NY: Pergamon Press Inc; 1990:1436-1444. 26. Maurice DM. Injection of drugs into the vitreous body. In: Leopold IH, Burns RP, eds. Symposium on Ocular Therapy. New York, NY: John Wiley & Sons Inc; 1976:59-71. 27. Hida T, Chandler D, Arena JE, Machemer R. Experimental and clinical observations of the intraocular toxicity of commercial corticosteroid preparations. Am J Ophthalmol. 1986;101:190-195. 28. McCuen BW, Bressler M, Tano Y, Chandler D, Machemer R. The lack of toxicity of intravitreally administered triamcinolone acetonide. Am J man
Therapeutics.
Ophthalmol. 1981;91:785-788.
29. Koizumi K, Honda Y. ERG changes after an injection of corticosteroid into the vitreous. Metab Ophthalmol. 1985;8:13-16. 30. Schlaegel TF, Wilson FM. Accidental intraocular injection of depot corticosteroids. Trans Am Acad Ophthalmol Otolaryngol. 1974;78:847-855. 31. Andrew NC, Gregor ZJ. Intraocular injection of Depomedrone. Br J Ophthalmol. 1986;
70:298-300. 32. Zinn KM. Iatrogenic intraocular injection of depot corticosteroids and its surgical removal using the pars plana approach. Ophthalmology. 1981; 88:13-17. 33. McLean EB. Inadvertent injection of corticosteroid into the choroidal vasculature. Am J
Ophthalmol. 1975;80:835-837.
34. Michael MJ. Intraocular Kenalog injections therapeutic in some CME cases. Ocular Surg News. November 1,1984:50. seen
