Safety Testing of Infracyanine Green Using Retinal Pigment Epithelium and Glial Cell Cultures

Abstract

purpose. To undertake safety testing of infracyanine green (IFCG) in a cell culture model.

methods. Experiments were undertaken in a cell culture model used previously to perform safety testing of indocyanine green (ICG). Human retinal pigment epithelium (RPE) and Müller cells were exposed to IFCG for 5 minutes, over a range of concentrations up to 0.5%. Experiments were repeated, using double-staining with trypan blue. Cell viability was measured at days 1, 5, and 15 using a mitochondrial dehydrogenase assay and a fluorescent live–dead probe containing calcein and ethidium homodimer-1. Viability was measured after exposure to 0.5% IFCG and 5 minutes of illumination with a vitrectomy endolight powered by a xenon light source.

results. RPE viability was not reduced over the range of concentrations and follow-up intervals. RPE cells exposed to IFCG and illumination had reduced viability relative to the negative control (cells exposed to saline), but not relative to those exposed to saline and illumination. Glial cells showed reduced viability at days 1 and 5, but not day 15. Illumination did not further reduce viability.

conclusions. IFCG has been advocated as a safer macular vital stain than ICG. These results suggest that it is less likely to produce phototoxicity, but despite being nearly iso-osmolar, IFCG also produces damage in cultured glial cells.

Indocyanine green (ICG) is now well established as an intraoperative macular vital stain. It is most commonly used to assist removal of the internal limiting membrane (ILM), as part of surgery to repair full-thickness macular holes. ^{1 2 3 4} Identifying the thin, optically clear ILM can be difficult, but after exposure to ICG, it assumes a green hue, making it much easier to delineate and remove. Although ICG is undoubtedly useful, concerns have been raised that it may damage the retinal pigment epithelium (RPE) and neuroretina. ⁵  

ICG is usually dissolved in distilled water that is then mixed with a balanced saline solution (BSS). This results in an approximately 275-mOsM solution that is hypo-osmotic relative to ocular tissue. Some investigators believe that it is the combination of low osmolarity and ICG exposure that damages cells. ⁶ Others have suggested that it is the interaction of surgical endoillumination and ICG that produces a phototoxic effect on the RPE and neuroretina. ^{7 8 9}  

More recently, researchers have advocated infracyanine green (IFCG) as an alternative vital stain. ^{6 10 11} IFCG differs from ICG, in that it contains no iodine and can be dissolved directly into a 5% glucose solution, avoiding a hypo-osmotic preparation. It is therefore postulated that IFCG is less likely to produce cell damage than ICG, and a brief experimental report suggests that this is the case. ⁶  

There are theoretical reasons why IFCG is less likely to produce retinal phototoxicity than ICG, as absorption has been found to be lower between 600 and 700 nm. ¹² It is therefore postulated that IFCG is less likely to produce phototoxic damage when used in conjunction with surgical endoillumination. Despite a potentially safer absorption spectrum, a brief report by Haritoglou et al. ¹³ showed retinal damage in cadaveric eyes exposed to IFCG and light. 

The experiments presented in this article were designed to test the safety of IFCG and to compare findings with those in experiments using ICG. To do this, we adapted a cell culture model used to test the safety of ICG. ¹⁴ If IFCG is to be used clinically, then further safety testing is certainly worthwhile. 

Methods

Dye Preparation

IFCG (Laboratoires SERB, Paris, France) was prepared by dissolving 25 mg in 5 mL of the dextrose monohydrate diluent provided by the manufacturer, producing a 0.5% (5 mg/mL) solution of 309 mOsM ⁶ ; 0.5% was selected on the basis of clinical reports. ¹⁰ Serial dilutions (0.5%, 0.25%, 0.125%, 0.0625%, and 0.03125%) were also prepared to look for dose-related effects and to simulate the lower concentrations that would occur if IFCG were injected into a fluid-filled vitreous cavity, as is often done with ICG. It is also possible that, in the future, clinicians may select more dilute preparations on the basis of their experiences with ICG. 

Cell Culture

Experiments were undertaken on RPE and glial cell lines, by using culture techniques described previously. ¹⁴ Briefly, human RPE cells (ARPE-19, passage 23; American Type Culture Collection, Manassas, VA) were cultured in Ham’s F-10 medium (pH 7.4; Sigma-Aldrich, Poole, UK), supplemented with 2 mM glutamine, 25 mM/L HEPES, 10 IU/mL penicillin, 10 μg/mL streptomycin, and 15% heat-inactivated fetal calf serum (Sigma-Aldrich). Glial cell experiments were undertaken in a spontaneously immortalized Müller cell line (passage 57) obtained from a 68-year-old female donor. Cells were grown in Dulbecco’s modified Eagle’s medium containing a commercial cell culture medium (glutaMAX; Invitrogen-Gibco, Paisley, Scotland, UK), supplemented with 2 mM glutamine, 10 IU/mL penicillin, 10 μg/mL streptomycin, and 10% heat-inactivated fetal calf serum (Sigma-Aldrich). The isolation and characterization of these cells is described elsewhere. ¹⁵ Cells were grown to confluence in an incubator with a humidified atmosphere of 5% CO², 95% air at 37°C and were trypsinized and seeded into 96-well flat-bottomed plates (TPP, Switzerland) and 8-well chamber slides (Nunc, Inc., Naperville, IL). 

Dye Exposure

Once cells reached confluence, the culture medium was removed and 50 μL IFCG was placed in each well. After 5 minutes, IFCG was removed with a suction pipette, and the wells were rinsed three times with phosphate-buffered saline (PBS; sodium chloride 120 mM, potassium chloride 2.7 mM, phosphate buffer salts 10 mM [pH 7.4] at 25°C; 294 mOsM; Sigma-Aldrich). The culture medium was then replaced, and cells were returned to the incubator. Cells were also exposed to 0.15% trypan blue, as this has been used in conjunction with IFCG—so-called double staining. ¹⁰ Monolayers undergoing double staining were stained first with trypan blue, then IFCG, to simulate the reported clinical use of these agents. ¹⁰ Cells were rinsed three times after exposure to each agent. Other cells were also exposed to trypan blue alone, to determine whether the cellular response to double-staining more closely matched that of IFCG, trypan blue, or otherwise. After dye exposure the culture medium was replaced, and cell viability was measured at days 1, 5, and 15. 

Negative control cultures were cells exposed to PBS. These had undergone identical rinsing routines. At least 16 wells were used as the negative control for each 96-well plate. The positive control was cells killed by exposure to 30% methanol for 1 minute. Previous experiments have shown that the cell culture grade trypan blue is comparable to medical grade trypan blue (DORC, Zuidland, The Netherlands) used clinically. ¹⁴  

Light Exposure

Experiments using IFCG were repeated in the presence of illumination from a surgical endolight powered by a medical, 300-W, xenon light source with a total lamp output of 5000 lumens. A xenon light source was selected over the more commonly used a quartz halogen light source, because it provided more stimulus for IFCG excitation and more chance of detecting IFCG-mediated phototoxicity. ¹⁶ Some of the major manufacturers of vitrectors are now offering the option of xenon light sources for endoillumination, because these units provide brighter illumination than halogen. This light source was coupled to a standard fiber-optic endoilluminating “light pipe” with a 45% illumination transmittance and an angular spread of 8 NA (manufacturer’s information). The cone angle of the light was measured in PBS containing 3% fat-free milk and found to be 36°. The output from the endolight was measured with a light meter (Graseby Optronics, Orlando, FL) and found to be 4.86 × 10⁻² mW/mm² at 5 mm. Cells were first exposed to IFCG for 5 minutes and then were rinsed only once with PBS, leaving a residual stain on the monolayer. The wells were then filled with PBS, and the endolight was held in a retention stand, 5 mm above the monolayer for 5 minutes. Unilluminated cells exposed to PBS acted as the main negative control, but controls were also provided in the form of cells incubated with PBS and exposed to light in the same manner. All cells were exposed to ambient illumination for the duration of the experiment. After illumination, wells were rinsed twice with PBS, and the growth medium was replaced. Cells were then returned to the incubator, and viability was measured at day 1. 

Cell-Viability Assays

A quantitative measure of cell viability was made with an MTT assay, based on our study of ICG. ¹⁴ There is a theoretical risk that residual IFCG and trypan blue will interfere with the assay of the blue formazan reaction product that is measured with the MTT assay. Although previous control experiments ¹⁴ with ICG and trypan blue suggested that this might not occur, preliminary analysis of results with IFCG suggest that it produces more residual staining than ICG and that this interferes with the MTT assay. To establish whether this was the case, cells were exposed to IFCG and rinsed in the usual manner; then, MTT was added. Immediately after this the optical density was measured with a microplate reader, so that results reflected the effect of residual IFCG, as the cells had not had time to produce the formazan reaction product. The results of this preliminary analysis are shown in Figure 1 . Monolayers had decreased optical density, relative to the negative control, that was statistically significant for up to 24 hours in RPE cells and 15 days in glia. This interaction could produce a falsely low reading that might simulate cell damage. 

The experiments presented in this article were therefore adjusted to account for residual staining of the monolayers with IFCG. A baseline assessment of optical density was made immediately after MTT was added to the wells. This baseline reading was subtracted from the final reading, taken 4 hours later. Hence, the effect of any residual stain was not included in this adjusted assay reading. 

A qualitative assessment of cell viability was made with a commercially available fluorescent live–dead probe (Molecular Probes, Eugene, OR). Live cells were stained green with calcein-AM (CAM), dead cells were stained red with ethidium homodimer-1 (EH-1). Cells were grown in eight-well chamber slides and underwent the same exposure routines as those in the 96-well plates. Negative control cultures were also included on each chamber slide, but not all chambers included a positive (dead-cell) control, given the uniformity of dead-cell staining. For light exposure experiments, only the cells immediately underneath the endolight were examined, as the area of the well extended beyond the area of illumination. Staining routines, microscopy, and photography were as described previously. ¹⁴  

Statistics

Cells were defined as having reduced viability if their optical density on the MTT assay was below two standard deviations of the negative control. Group comparisons were also made using the unpaired t-test, with the Welch correction, if assumption tests showed a significant difference in the standard deviation of the groups. Nonparametric (Mann-Whitney) tests were used if normality tests (Kolmogorov-Smirnov) indicated that data had a non-Gaussian distribution. P ≤ 0.05 was considered significant. The data presented as confidence intervals (CIs) are expressed in relation to the mean of the negative control for each experiment, which was set at 1.0. The 95% CI shows the difference between the mean result of the negative control (cells exposed to PBS) and the comparison group, unless otherwise stated. 

Results

Retinal Pigment Epithelium

MTT Assay.

MTT assays of RPE cells exposed to IFCG at a range of concentrations from 0.03125% to 0.5% did not show toxicity (Fig. 2) . Although all data points were below the mean of the negative control, none fell below the predetermined limit (negative control minus 2 SD). The lowest data point occurred in the 0.25% group, but this was within 1 SD of the negative control, and this difference was not significant (P = 0.1357; CI −0.21 to 1.20). IFCG 0.5% was not significantly reduced relative to the negative control (P = 0.651; CI −0.28 to 0.45). There was no clear relationship between the concentration of IFCG and cell viability (R ² of linear trend line = 0.2909). 

There was no evidence of delayed toxicity in cells exposed to 0.5% IFCG, trypan blue, and combined IFCG and trypan blue, when followed up for 15 days (Fig. 3) . 

Cells exposed to IFCG and illumination (Fig. 4) had reduced viability compared with those exposed to PBS without illumination (P = 0.0047; CI 0.08–0.41) but not compared with IFCG without illumination (P = 0.366; CI −0.20 to 0.52), or PBS with illumination (P = 0.540; CI −0.20 to 0.38). The viability of cells exposed to PBS with illumination was less than that of cells in PBS without illumination, but the difference was not significant (P = 0.2984; CI −0.45 to 0.14). 

Light and Fluorescence Microscopy.

It was noted that cell monolayers exposed to 0.5% IFCG retained a green hue that persisted for 15 days. In addition, there were commonly areas in which the IFCG appeared as a localized, discrete, patchy, film on the monolayers immediately after rinsing with PBS. There were occasional small areas of denuded cells that were possibly more common in the IFCG groups, despite having rinsing routines identical with those in the control group. These areas had repopulated by day 5. By day 15, the plaquelike, patchy IFCG staining was not common, but some wells showed focal green spots of IFCG surrounded by a rosette of cells. 

Staining with CAM and EH-1 did not reveal any consistent difference from the negative control. 

Glial Cells

MTT Assay.

MTT assays of Müller cells exposed to IFCG over a range of concentrations from 0.03125% to 0.5% did not show toxicity, in that all data points were within the predetermined limit. However, the 0.25% and 0.5% group were lower than the negative control (Fig. 5) . This difference was significant in the 0.5% group (P = 0.0003; CI 0.10–0.31), but not the 0.25% group (P = 0.218; CI −0.14729 to 0.57). There was a tendency toward a linear decrease in viability with increasing concentration, but this association was not close (R ² = 0.5226). 

Viability continued to decrease in the IFCG 0.5% group at day 5 (P = 0.008; CI 0.32–1.04), with values below 2 SD of the negative control. Cells exposed to both IFCG and trypan blue were also below 2 SD of the negative control at day 5. Viability in both groups was similar to that in the negative control cultures by day 15 (Fig. 6) . Cells exposed to trypan blue did not show toxicity, with all data points above the predetermined limit. 

Illumination of cells exposed to 0.5% IFCG did not further reduce viability. The viability of glia exposed to IFCG and illumination (Fig. 7) was significantly less than that of those exposed to PBS without illumination (P = 0.026; CI 0.01–0.16), but not PBS with illumination (P = 0.08; CI −0.01 to 0.19). There was no difference between the PBS with and without light groups (P = 0.8476; CI −0.08 to 0.09). 

Light and Fluorescence Microscopy.

Most groups stained with the fluorescent live–dead probe were similar to the negative control. There were, however, some scattered dead cells in the 0.5% IFCG group stained at days 1 and 5 (Figs. 8A 8B) . In addition, some of the monolayers in the day-5 IFCG group showed alterations in morphology, with cells appearing more rounded and clumped than in the negative control cultures (Fig. 8C) . Trypan blue did not noticeably alter the appearance of cells. 

Examination with light microscopy showed a surface staining pattern similar to that in RPE monolayers, with patchy areas of discrete staining at day 1, in addition to the green hue that stained the entire surface up to day 15. Rosettes of cells surrounding green spots of IFCG were also seen at day 15 (Figs. 8D 8E) . These focal collections of cells sometimes appeared without an associated spot of IFCG. 

Discussion

The safety of ICG remains a contentious issue. There are several clinical and experimental reports suggesting that higher concentrations of ICG are damaging to neuroretina and RPE, ⁵ but it is still in common use, with recent clinical reports showing favorable results when used in macular hole surgery. ^{17 18 19} Some studies suggest ICG itself is damaging, ²⁰ others that it is the combination of ICG and hypo-osmolarity that produces cell damage. ⁶ The response of many surgeons has been to reduce the concentration of ICG, ^{17 18 19} using preparations that are near to iso-osmolar with ocular tissue. 

Another response has been to try other agents that may prove safer. ²¹ Hence, Stalmans et al. ¹⁰ recently reported the clinical use of IFCG. As already noted, this agent is similar to ICG but can be dissolved directly into glucose, avoiding hypo-osmotic preparations. It is therefore possible that it may be less toxic. Preliminary ex vivo work by the same researchers did not show significant damage in RPE cells exposed to IFCG, although average results were lower than the control. ⁶ Alternatively, IFCG’s chemical similarity to ICG may mean that it has the same effect on retinal tissue, either in isolation or in combination with endoillumination. As is the case of ICG, the clinical use of IFCG has predated extensive ex vivo experiments. We sought to investigate this issue by performing safety testing of IFCG, using RPE and glial cell cultures. 

The main findings were as follows. There was no evidence of acute or delayed damage when RPE cells were exposed to IFCG up to a concentration of 0.5%. This was also true when IFCG was used in conjunction with trypan blue, referred to as double staining. The combination of IFCG and endoillumination significantly reduced RPE viability relative to the negative control, but not relative to cells that had undergone illumination without IFCG exposure. Glial cells were more vulnerable, with 0.5% IFCG producing acute damage that persisted at day 5, but had recovered by day 15. There was no apparent phototoxicity when glial cells were exposed to IFCG and endoillumination. 

Cell damage was more apparent with the MTT assay than the live–dead fluorescent probe. This was similar to our observations using ICG, ¹⁴ and other reports that found more damage with the MTT assay than was seen with light and electron microscopy. ⁷ This may suggest that ICG and IFCG have more effect on mitochondrial function than on cytoplasmic esterase activity and cell wall integrity, as the former is measured using the MTT assay and the latter two are assessed using the live–dead probe. 

Another possibility is that IFCG interferes with the MTT assay. Our earlier studies ¹⁴ of ICG suggested that this was not the case, but there is a theoretical risk given the spectral overlap of ICG and the blue formazan reaction product measured by the MTT assay. Unlike the situation with ICG, preliminary studies using IFCG suggested that higher concentrations interfere with the MTT assay. For this reason, the assay was adapted by subtracting a baseline reading from the final reading, so that the effect of any residual dye was negated. MTT assays of cells exposed to IFCG without this adjustment may produce falsely low estimates of cell viability. 

It is not known why IFCG produced more surface staining than ICG. It is possible that the saline-based fluid used to rinse the wells (PBS) produced flocculation of residual IFCG on the monolayers. A glucose solution may have been more effective in rinsing the monolayers, but would differ from the saline-based solutions used clinically for vitreous infusion. Internationally, surgeons select from several proprietary brands of intraocular solutions, and those originally designed for intravenous use, such as Hatmann’s solution. In the absence of a universal intraocular infusion, PBS was selected because it is similar to most solutions used clinically and is well established for use with cell cultures. 

Some other findings of this study are more similar to our observations with ICG. ¹⁴ Both experiments showed that glial cells were more vulnerable to damage than RPE cells, and that the maximum damage appears at day 5.The viability of glial cells exposed to IFCG and endoillumination was not reduced relative to those exposed to IFCG alone, unlike the situation with ICG. One hypothesis is that the shift in the absorbance spectrum of IFCG relative to ICG makes it less likely to mediate phototoxic damage. ¹²  

Initial inspection of the results in RPE suggests a phototoxic effect, with the viability of cells exposed to IFCG with illumination significantly reduced relative to the negative control (PBS without illumination). Closer inspections of these results suggests otherwise, as viability was not reduced relative to the PBS with illumination group. Although the difference between the PBS with and without illumination groups did not reach significance, the data in Figure 4 suggest that illumination was primarily responsible for cell damage, and this effect was not significantly altered by the presence of IFCG. 

Although the statistical tests do not support the conclusion that the xenon light source was damaging to RPE, this trend merits further investigation, particularly as these light sources have recently been introduced for endoillumination during vitreoretinal surgery. Compared with halogen light sources, xenon produces much more intense output over a broad spectral range. Despite this increased potential for phototoxic damage, it is noted that our previous study ¹⁴ did not show reduced viability in the control groups that underwent illumination without ICG, despite using the same xenon light source and illumination distance. In the present study, a longer exposure time was used; 5 minutes instead of 1. This suggests a dose–response effect. 

Yam et al. ⁸ also investigated ARPE-19 cells exposed to light and ICG. Although light alone produced less damage than it did in combination with ICG, the expression of the c-Fos gene was increased in BSS control groups exposed to a standard surgical endolight, suggesting cell damage. It was not our intent to investigate the effect of xenon light sources on cells in culture, but it would be interesting to repeat these observations with the interposition of optical filters in the light path and to compare halogen and xenon light sources directly. Direct comparison of both light sources might be problematic, however, as the power sources, filters, and endoscopic probes differ between manufacturers, making standardization difficult. Given what is known about the spectral absorption of IFCG ¹² and emission of these light sources, ^{16 22 23 24} halogen is probably less likely to cause phototoxic damage than xenon. 

The purpose of this study was instead to perform safety testing of IFCG. A cell culture model was selected for several reasons. Cell culture affords relatively precise control of experimental parameters, including those that might not be possible to control clinically or in animals. It has the obvious advantage that it does not subject humans or animals to potentially damaging conditions and is certainly appropriate for preliminary testing. This method also allows individual cell lines to be studied. The two chosen are probably the most relevant, as clinical and experimental reports suggest that RPE may be damaged by ICG, ⁵ as may Müller cells. ⁹ Müller cells are also in intimate contact with dyes that stain the ILM. An important disadvantage of a cell culture model is that it does not fully replicate the situation that occurs in vivo, and findings cannot be directly extrapolated to the clinical environment. It is possible also that the immortalized cell lines used in this study behave differently from those in vivo and may be more robust than primary cell cultures. This latter fact may mean that our findings tend to underestimate the potential damage caused by IFCG. 

In this study, we repeated several experiments that were undertaken on cells exposed to ICG, ¹⁴ but used IFCG instead. A comparison of both these studies is summarized in Table 1 . One notable difference was that there was no evidence of IFCG-mediated phototoxicity, despite the longer illumination times. Although IFCG avoids the hypo-osmolarity of ICG preparations, it remains damaging to glial cells when used at higher concentrations. 

 Submitted for publication April 5, 2004; revised June 8, 2004; accepted June 12, 2004.

 Disclosure: T.L. Jackson, Keeler Instruments UK (F); B. Vote, None; B.C. Knight, None; A. El-Amir, None; M.R. Stanford, None; J. Marshall, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Results of a background study investigating the effect of IFCG on the MTT assay used to assess cell viability. RPE monolayers were stained with 0.5% IFCG for 5 minutes, then rinsed three times with PBS. At days 0 to 15, they were placed in the microplate reader, immediately after MTT had been added, but before cells were able to produce the formazan reaction product that is measured with this assay. Hence, the data shown represent the effect of residual dye on the assay and are compared with results from cells exposed to PBS. IFCG readings were lower than the control at each time point (day 0, P < 0.0001; day 1, P < 0.0001; day 5, P = 0.0807; day 15, P = 0.397). Similar results were obtained with glial cells although staining persisted longer (P < 0.0004, all time points). Cells exposed to trypan blue are shown for comparison. This dye had notably less residual staining than IFCG. Error bars, ±1 SD.

Figure 2.

 

The results of an MTT viability assay of RPE cells exposed to IFCG over a range of concentrations from 0.03125% to 0.5%. Exposure time was 5 minutes, with viability read at day 1. The y-axis shows viability expressed as a percentage of the negative control. Solid line: mean of the negative control; dotted lines: mean of negative control ±1 SD. Error bars, SD for individual data points. Statistical tests are described in the text.

Figure 3.

 

Results of an MTT assay of an RPE cell line exposed to 0.5% IFCG, with and without 0.15% trypan blue, and to trypan blue alone. Viability was read at days 1, 5, and 15. Note that the x-axis is categorical and nonlinear. Data presentation is as per Figure 2 .

Figure 4.

 

MTT assay of an RPE cell line exposed to 0.5% IFCG, with or without endoillumination from a xenon light source. Viability was measured at day 1. Statistical comparisons are provided in the text. Error bars, 1 SD.

Figure 5.

 

The results of an MTT viability assay of a Müller cell line exposed to IFCG over a range of concentrations from 0.03125% to 0.5%. Exposure time was 5 minutes, with viability read at day 1. Data presentation is as in Figure 2 .

Figure 6.

 

An MTT assay of a Müller cell line exposed to 0.5% IFCG, with and without 0.15% trypan blue, and to trypan blue alone. Viability was measured at days 1, 5, and 15. Data presentation is as in Figure 2 .

Figure 7.

 

MTT assay of Müller cells exposed to 0.5% IFCG with or without endoillumination from a xenon light source. Viability was assessed at day 1. Statistical comparisons are provided in the text. Error bars, 1 SD.

Figure 8.

 

(A) Fluorescence photomicrograph of Müller cells exposed to 0.5% IFCG for 5 minutes, and then stained with the live–dead probe at day 1. Dead cells labeled red with ethidium homodimer-1 were scattered among the live cells stained green with calcein. (B) The same image through a red filter, with the nuclear staining pattern of dead cells seen more easily. (C) Another Müller cell monolayer at day 5, with rounded cell outlines and clumping, relative to the more normal cell morphology shown in (A). (D) A light photomicrograph of two discrete green spots on the surface of a Müller cell monolayer at day 15. This spotty staining was also observed in day-15 RPE monolayers (not shown). (E) Another Müller cell monolayer at day 15, with rosette-shaped, focal collections of cells. Bars, 100 μm.

Table 1.

 

Main Findings in the Present Study of IFCG Compared with a Previous Study ¹⁴ of ICG

Study Group	Infracyanine Green							Indocyanine Green
		RPE			Glia			RPE			Glia
		MTT	Microscopy	MTT	Microscopy	MTT	Microscopy	MTT	Microscopy
Acute toxicity (0.0313%–0.5%)	No	No	Yes (0.5% only)	Occasional dead cells (0.5% only)	No	No	No	No
Delayed toxicity (0.5%)	No	No increase in cell death, but rosettes of cells surrounding spotty green deposits at day 15	Yes (day 1 and 5,but not day 15)	Increased cell death at day 5 with associated cell clumping. Rosettes and green deposits seen at day 15.	No*	No*	No, †	No, †
Additional toxicity when combined with xenon illumination	No	No	No	No	No	No	Yes	No

 The ICG study was undertaken using ICG solutions of various osmolarities, but the results shown are those of the standard, 275 mOsM solution. Boxes marked Yes indicate that the study group showed a statistically significant difference from the negative control, with the MTT assay. Qualitative differences, using either light or fluorescence microscopy, are also noted. Fluorescence microscopy was undertaken, using a live–dead probe containing ethidium homomdimer-1 and calcein AM. The negative control was physiologic saline for all groups except those showing the effect of xenon illumination. In this series of experiments, cells were exposed to a 0.5% solution of IFCG or ICG, rinsed once with saline, and then illuminated with a xenon light source. Cells were compared to those exposed to saline and xenon illumination. The numbers shown in brackets (column 1) indicate the concentration of dye as a percentage weight/volume.

*  No delayed RPE toxicity was observed with a 0.5% solution, but a 1% solution showed a linear decrease in viability over time, so that the day-15 data point was significantly lower than the negative control with the MTT assay. The live–dead fluorescent probe did not show any consistent damage in this day-15 group.

†  No delayed glial toxicity was observed with the 0.5% solution, but a 1% solution produced a significant reduction in cell viability, with the MTT assay at day 5. This day-5 group showed scattered dead cells with the live–dead fluorescent probe.

The authors thank G. Astrid Limb (Institute of Ophthalmology, London, UK) for providing the glial cell cultures used in this study and Paul Constable for assistance with RPE cell culture.