Cytotoxicity of Indocyanine Green on Retinal Pigment Epithelium: Implications for Macular Hole Surgery

Objective  To evaluate the potential cytotoxic effects of indocyanine green (ICG) on cultured human retinal pigment epithelium (RPE) and the resultant implications for macular hole surgery.

Methods  Human RPE cells were exposed to ICG in concentrations from 0.001 to 5 mg/mL. The exposure duration ranged from 5 minutes to 3 hours. Light microscopy, MTS viability assay, and calcein AM–ethidium homodimer 1 staining were used to evaluate the cytotoxic effects of ICG.

Results  The RPE cells incubated with up to 5 mg/mL of ICG for 5 minutes or less exhibited no morphologic change and no significant decrease in dehydrogenase activity. When RPE cells were exposed to 5 mg/mL of ICG for 10 minutes, 1 mg/mL of ICG for 20 minutes, or 0.01 mg/mL of ICG for 3 hours, cell morphologic features were altered, mitochondrial dehydrogenase activity decreased, and some cells were necrotic.

Conclusions  Indocyanine green caused cytotoxicity in cultured human RPE in a dose- and time-dependent manner. Cell death occurred by necrosis.

Clinical Relevance  Exposure of RPE cells to ICG concentrations up to 5 mg/mL for 5 minutes or less was not injurious; prolonged exposure to a low ICG concentration was toxic. Since ICG may be retained in the vitreous cavity for a lengthy period, thorough washout of ICG during macular hole surgery is required.

RECENTLY, SEVERAL investigators have reported that peeling of the internal limiting membrane (ILM) to eliminate tangential traction force improves the anatomic closure rate and functional outcomes, ^1-4 although other investigators hold a different opinion.^5-7 However, if ILM removal is attempted during macular hole surgery, its visualization may be difficult. The staining of the ILM with indocyanine green (ICG) to enhance its visibility has recently been proposed.^8-12 The procedure involves direct application of ICG to the inner surface of the retina in the macular area. No adverse effects of ICG were reported in these studies. The concentration of ICG used in these reports ranged from 0.6 to 5 mg/mL, and ICG was left in the vitreous cavity from 30 seconds to 5 minutes.

Conversely, Engelbrecht et al¹³ observed a high incidence of unusual atrophic retinal pigment epithelium (RPE) changes that occurred at the site of a previous macular hole and its surrounding subretinal fluid after ICG-assisted ILM peeling. These changes were not consistent with those caused by light toxicity. Engelbrecht and colleagues used an ICG concentration of 1 mg/mL, and the ICG was left in the eye for periods ranging from 0.5 to 2.5 minutes. The median preoperative best-corrected visual acuity was 20/200, whereas the median postoperative best-corrected visual acuity was 20/400. The macular hole was closed in 86% of eyes. Gandorfer et al¹⁴ held a similar opinion. They used 5 mg/mL of ICG, and the ICG was drained immediately after injection. After a review of the medical records of their patients who had undergone macular hole operations involving removal of ILM viewed with or without the aid of ICG staining, these investigators found a tendency toward less favorable visual outcome in those receiving the ICG-assisted operation than in those receiving the identical operation but without the use of ICG. They assumed that ILM staining with ICG might be responsible for the less favorable functional result, because this was the only surgical step that had been changed.

In an in vitro study, ¹⁵ a significant reduction in enzymatic activity of mitochondrial dehydrogenase was observed in cultured RPE cells exposed to 1 mg/mL of ICG for 20 minutes. That study reported no histologic or ultrastructural differences between the treated and control RPE cultures. To investigate whether ICG is cytotoxic, we evaluated the effects of ICG on cultured human RPE cells. In this study, when RPE cells were exposed to 5 mg/mL of ICG for 10 minutes, 1 mg/mL of ICG for 20 minutes, or 0.01 mg/mL of ICG for 3 hours, cell morphologic features were altered and mitochondrial dehydrogenase activity decreased. Some cells were necrotic as demonstrated by calcein AM and ethidium homodimer 1 and acridine orange–ethidium bromide staining.

Human RPE cell (ARPE-19) was obtained from American Type Culture Collection(ATCC, Manassas, Va). This cell line is not transformed and has structural and functional properties characteristic of RPE cells in vivo.¹⁶ The RPE cells were cultured in Dulbecco's modified Eagle medium and F12 medium(1:1) containing 10% fetal bovine serum (GIBCO; Invitrogen Corporation, Grand Island, NY). The following substances were added: transferrin, 0.01 g/L; insulin, 0.01 g/L; sodium bicarbonate, 0.912 g/L; penicillin, 100 U/mL; streptomycin, 0.1 mg/mL; gentamicin, 5 mg/mL; HEPES, 3.5745 g/L; and d-glucose, 1.749 g/L. The cells were cultured at 37°C in 5% carbon dioxide.

The ICG was prepared by completely dissolving 25 mg of sterile ICG powder(Daiichi Pharmaceutical Co, Tokyo, Japan) in 0.5 mL of sterile distilled water. A total of 4.5 mL of balanced salt solution (BSS Plus; Alcon Laboratories Inc, Fort Worth, Tex) was added to achieve a final ICG concentration of 5 mg/mL (275 m Osm/kg). The other ICG concentrations were made by diluting this 5-mg/mL ICG solution with an appropriate amount of BSS. The ICG concentrations included 5, 1, 0.1, 0.01, and 0.001 mg/mL (275, 299, 304.4, 304.9, and 305m Osm/kg, respectively). Exposure times ranged from 5 minutes to 3 hours. The RPE cells were kept in the dark during the ICG exposure period and the evaluation period. Distilled water was added to BSS (305 m Osm/kg) to prepare diluted BSS with osmolarities identical to the tested concentrations of ICG solutions(ie, 275, 299, 304.4, and 304.9 m Osm/kg). The RPE cells were exposed for 3 hours to the diluted BSSs to investigate whether the injury caused by ICG was due to alterations in osmolarity.

Phase-contrast microscopy was used to observe the effects of ICG on RPE cell morphologic features. Cell viability was assessed by MTS colorimetric assay (3-[4, 5-dimethylthiazol-2-yl]-5-[3-carboxymethoxyphenyl]-2-[4-sulfophenyl]-2H-tetrazolium, inner salt) (Promega Corp, Madison, Wis). This quantitative assay detects living but not dead cells.¹⁷ The absorbance at 490 nm (test wavelength) and at 650 nm (reference wavelength) was measured using an enzyme-linked immunosorbent assay (ELISA) microplate reader (VERSAmax; Molecular Devices, Sunnyvale, Calif); wells containing culture medium but no cells served as blanks. In our experiments, 10⁴ cells in 100µL of culture medium were seeded into each well of a 96-well plate. After culturing the cells for 48 hours and achieving 80% to 90% confluence, the culture medium was removed, the cells were rinsed with phosphate-buffered saline, and 100 µL of either various concentrations of ICG solutions or BSS alone as control was added to the wells. After incubation of a predetermined period (5, 10, 20, or 30 minutes, 1, 2, or 3 hours), the well was washed to remove the reagents. The MTS assay was performed by adding 100 µL of the culture medium and 20 µL of MTS to each well. After a 3-hour incubation at 37°C, the absorbance at 490 nm was recorded. Five wells were evaluated for each ICG concentration at each time point. The experiments were performed at least 3 times (for a total of at least 4 times). Statistical significance was calculated by comparing results by the 2-tailed, unpaired t test. P<.05 was considered statistically significant.

The RPE cells were cultured in chamber slides for the calcein AM–ethidium homodimer 1 assay. Each chamber was seeded with 0.7 mL of RPE cells (10⁵ cells/mL). After 48 hours, the appropriate ICG concentration or BSS alone as control was added. The samples were then incubated with a solution containing 2-µM calcein AM and 4-µM ethidium homodimer 1 (Molecular Probes, Eugene, Ore) for 45 minutes. At least 4 wells were evaluated for each ICG concentration at each time point. In each well, at least 250 cells were evaluated for viability vs death. Two of the authors (S.-N.C. and H.-C.C.), who were blinded to the ICG treatment or control status of the samples, evaluated the RPE cells.

Cultured RPE cells exhibited no morphologic changes after they were exposed to 5 mg/mL of ICG for up to 5 minutes or 1 mg/mL of ICG for up to 10 minutes. Increasing the duration of ICG exposure induced progressive morphologic changes. After a 10-minute treatment with 5 mg/mL of ICG, some RPE cells swelled, and the cytoplasm was finely vacuolated; other cells appeared to be more shrunken. Similar observations were noted when RPE cells were exposed to 1 mg/mL for 20 minutes (Figure 1A and B). No morphologic change was noted when RPE cells were incubated in BSS for 20 minutes(Figure 1C). For the same exposure duration, the degree of morphologic change was greater as the ICG concentration increased. The morphologic change also increased with ICG exposure time. Membranous and intracytoplasmic retention of fine ICG granules was noted in most cells(even after unbound ICG was washed away). Moreover, with increased exposure duration, ICG began to stain some cells green (including swollen cells and cells with a shrunken appearance) (Figure 1D). Finally, many cells underwent lysis, and the cellular outline became poorly defined with only the nucleus discernible after 3 hours of ICG(1 mg/mL) exposure (Figure 1E). Control RPE cells exposed to BSS for the same period appeared unchanged (Figure 1F). No morphologic changes were evident in RPE cells exposed to diluted BSS (275 m Osm/kg), with an osmolarity identical to that of the 5 mg/mL ICG solution, for as long as 3 hours (data not shown). Therefore, the injury caused by incubation of the cells with the ICG solution could not be attributed to alterations in osmolarity.

Enzymatic integrity in cultured human RPE cells was evaluated with an MTS colorimetric assay. The MTS tetrazolium compound is reduced to a colored formazan product by a nicotinamide adenine dinucleotide phosphate– or nicotinamide adenine dinucleotide–dependent dehydrogenase in metabolically active cells.¹⁸ The formazan produced was quantitated with an ELISA microplate reader at 490 nm. Exposure of RPE cells to BSS for 3 hours did not alter the dehydrogenase activity. When RPE cells were exposed to 5 mg/mL of ICG for 10 minutes, dehydrogenase activity was significantly reduced (Figure 2). This threshold concentration of ICG decreased to 1 mg/mL when the exposure time was 20 minutes and to 0.01 mg/mL when the exposure time was extended to 3 hours. As the exposure time increased, the threshold ICG concentration that caused a significant reduction in dehydrogenase activity decreased. Indocyanine green induced time- and dose-dependent reductions in RPE cell viability (Figure 2). Exposure of RPE cells to diluted BSS (275 m Osm/kg) for as long as 3 hours did not induce a significant decrease in enzymatic activity(97.5% ± 8.9% of control value); therefore, the observed effects of ICG on enzymatic activity were not caused by variations in osmolarity.

To investigate ICG cytotoxicity, RPE cells were stained with 2-µM calcein AM and 4-µM ethidium homodimer 1. Live cells are identified by the presence of ubiquitous intracellular esterase activity, which converts the virtually nonfluorescent, cell-permeant calcein AM to green-fluorescent calcein, which is retained within live cells. Ethidium homodimer 1 enters the cells with compromised membranes and, upon binding to nucleic acid, provides a bright red fluorescence in dead cells.¹⁹

In RPE cells incubated for 5 minutes in 1 mg/mL of ICG, the cytoplasm exhibited intense green fluorescence, and the nuclei were unstained; these cells remained viable (Figure 3A). Some of the RPE cells incubated with 1 mg/mL of ICG for 20 minutes had red-fluorescent nuclei and a small amount of cytoplasmic esterase activity (green fluorescence), indicating that the cells were dead. The intense green fluorescence in the cytoplasm of other cells signified their viability (Figure 3B). Structures of most red fluorescent nuclei were normal. There were no signs of apoptosis, such as condensed and fragmented nuclei or margination of the chromatin to the nuclear membrane. At this concentration, ICG apparently caused cell death by a necrotic mechanism. The number of cells with red fluorescent nuclei and little cytoplasmic green fluorescence increased with longer ICG incubation period (Figure 3C). Almost all cells showed red fluorescent nuclei after 3 hours of exposure to 1 mg/mL of ICG (Figure 3D). The percentage of dead cells (with red-fluorescent nuclei), as observed by means of calcein AM–ethidium homodimer 1 staining, is plotted in Figure 4. The ICG toxicity in RPE cells was dose and time dependent, according to results from the MTS colorimetric assay and the fluorescence viability assays.

Exposure of RPE cells to diluted BSS (275 m Osm/kg; identical to osmolarity of 5 mg/mL of ICG solution) for as long as 3 hours caused no cell morphologic changes. Calcein AM–ethidium homodimer 1 staining showed that these cells remained viable (Figure 5). The MTS colorimetric viability assay revealed no reduction in dehydrogenase activity (97.5% ± 8.9% of control).

Prolonged ICG exposure induced necrotic cell death in cultured RPE cells as evidenced by the appearance of morphologic indicators of necrosis, such as cellular lysis and lack of nuclear fragmentation and chromatin condensation. Calcein AM–ethidium homodimer 1 staining also supported this conclusion. In addition, we performed acridine orange–ethidium bromide staining, ^20,21 which confirmed the results obtained with calcein AM–ethidium homodimer 1 staining. During macular hole surgery, ICG is applied for periods ranging from 30 seconds to 5 minutes. Although ICG is removed by washing after the surgery, ICG may persist for a long time(even 8 months after surgery).^22-24 Therefore, the exposure times used in this study (ranging from 5 minutes to 3 hours) had clinical relevance. The ICG cytotoxicity was not caused by osmotic variations; exposure of RPE cells to diluted BSSs (with osmolarities comparable to the ICG solutions) did not induce morphologic or functional changes in the cells.

Although ICG is a commonly used dye with a long history of safety and low toxicity, ^25-28 it is most often administered intravenously in humans. After intravenous injection, ICG is taken up exclusively by hepatic parenchymal cells and is rapidly cleared from the circulation via bile secretion. In earlier publications, the plasma half-life of ICG was estimated to be only approximately 2 to 4 minutes.^29,30 More recent spectrophotometric studies³¹ suggest the clearance of ICG in blood is biphasic, with a rapid initial phase (half-life of 3 to 4 minutes) and a secondary phase(half-life of more than 1 hour) at low concentrations. The pharmacokinetics of intravenous ICG most likely differ markedly from those of ICG injected into the vitreous cavity. Owing to the slow turnover rate in the vitreous cavity, ICG must have a longer half-life in the vitreous cavity compared with its plasma half-life. The reports describing cases with persistent ICG fluorescence after intraocular ICG administration (even as long as 8 months) lend support to our speculation.^22,23

Removal of the ILM has been reported to improve both anatomic and visual results.^1-4 However, difficulty in visualizing the ILM may lead to an increase in surgical time and the risk of phototoxicity. Staining with ICG may improve visualization of the membrane. Some studies^8-12 reported no adverse effects associated with the use of ICG in macular hole surgery. The concentration of ICG used in these reports ranged from 0.6 to 5 mg/mL, and ICG was left in the vitreous cavity for 30 seconds to 5 minutes. However, toxicity may be associated with ICG use. Gandorfer et al¹⁴ reported that ICG-assisted ILM peeling potentially caused retinal damage and that visual outcomes were poorer when ICG was used during surgery. They used a concentration of 5 mg/mL of ICG, and the ICG was drained immediately after injection. They suggested that the use of ICG was responsible for the less favorable outcome. Engelbrecht et al¹³ reported that RPE changes occurred after macular hole surgery when ICG was used as an adjuvant in ILM peeling. They used an ICG concentration of 1 mg/mL, and the ICG was left in the eye for periods ranging from 0.5 to 2.5 minutes.

The cause of the postoperative changes in RPE cells and the poorer visual outcome in these reports is not yet clear. These observations could possibly be attributed to cytotoxic effects of ICG on RPE cells in vivo, similar to those effects observed in RPE cells in vitro. Indocyanine green induced dose- and time-dependent changes in cell morphologic features and cell lysis, decreased enzymatic activity, and compromised plasma membrane integrity. In this study, the threshold durations that induced significant reductions in mitochondrial dehydrogenase activity were 10 minutes for 5 mg/mL of ICG, 20 minutes for 1 mg/mL of ICG, and 3 hours for 0.01 mg/mL of ICG. Based on these results, the ICG concentrations and exposure times commonly applied in macular hole surgery (0.6-5.0 mg/mL and 30 seconds to 5 minutes) may not be injurious to RPE cells during the ICG incubation period. Although the initial ICG incubation is brief (30 seconds to 5 minutes), some residual ICG may remain in the eye for a prolonged period even after washout.^22,23 The toxic effects of prolonged ICG exposure (even in a very low concentration) on RPE cells, as demonstrated in this study, are clinically significant. Assuming that ICG injected into the vitreous cavity (volume of 4 mL in phakic eyes) is uniformly distributed, the intravitreal ICG concentration would be 0.25 to 1.25 mg/mL.^8,9,13,22 If 99% of the ICG is washed out after the incubation and 1% of the ICG remains in the eye, the residual intravitreal ICG could reach a concentration of 0.0025 to 0.0125 mg/mL (1% × 0.25-1.25 mg/mL). In the presence of a macular hole, ICG has access to the subretinal space and can be in direct contact with the RPE monolayer. In this study, exposure of RPE cells to 0.01 mg/m Lof ICG for 3 hours caused a significant reduction in dehydrogenase activity and changes in staining patterns. The ICG cytotoxicity may be responsible for the RPE damage at the macular hole and the surrounding subretinal fluid area observed by Engelbrecht et al.¹³

Indocyanine green has been used to stain the anterior capsule of the lens during circular continuous capsulorrhexis; this use of ICG was reported to be safe.³² However, direct contact between ICG and rabbit corneal endothelial cells led to cytoplasmic edema and swelling in corneal endothelial cells.³³ This finding is compatible with our observations. The use of ICG may be safer in the anterior chamber than in the vitreous cavity, because the barrier surrounding the vitreous compartment is tighter than that surrounding the anterior chamber.³⁴ The elimination of ICG must be slower from the vitreous cavity than from the anterior chamber. The RPE cytotoxicity is more severe with longer ICG exposures, and ICG may be retained in the vitreous cavity for a prolonged period. Therefore, ICG should be used cautiously in the vitreous cavity, and a thorough washout of ICG following vitreal surgery is crucial.

Phototoxicity or photodynamic toxicity may also induce RPE changes after ICG-assisted ILM peeling. In this study, we demonstrated that ICG induced RPE toxicity after prolonged exposure, even at low ICG concentrations. We kept the RPE cells in the dark during and after the ICG exposure period. Therefore, the RPE cytotoxicity demonstrated in this study was due to the toxic effects of ICG itself. However, we did not exclude the possibility that photodynamic toxicity contributed to the RPE changes after ICG-assisted ILM peeling clinically, because ICG is also a photosensitizing agent.^35,36 Indocyanine green staining of the ILM may facilitate the peeling procedure and reduce surgical time. However, ICG use may decrease the safety time for macular light exposure. If photodynamic toxicity contributes to RPE toxicity in the clinical situation, reducing the intensity of the light and maintaining a sufficient distance between the light pipe and the retina will be helpful in preventing phototoxicity.³⁷ Again, a thorough washout of the ICG would also be advantageous.

Indocyanine green staining of the ILM facilitates its identification and removal during macular hole surgery. Indocyanine green staining can be a valuable tool; however, no standardized procedure that specifies concentration, volume, incubation time, and dilution exists for the intravitreal use of ICG. In this study, ICG caused detrimental effects on cultured human RPE cells in a dose- and time-dependent manner. Further in vivo studies are necessary to establish the ideal parameters for intravitreal use of ICG and to achieve safer staining and efficient ILM dissection.

Corresponding author and reprints: Ray Jui-Fang Tsai, MD, Department of Ophthalmology, Chang Gung Memorial Hospital, 5 Fu-Hsin St, Kuei-Shan, Taoyuan, Taiwan (e-mail: raytsai@ms4.hinet.net).

Submitted for publication September 19, 2002; final revision received March 12, 2003; accepted April 10, 2003.

DrHo had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.