- Email: [email protected]
An experimental method for testing novel retinal vital stains
Experimental Eye Research 81 (2005) 446–454 www.elsevier.com/locate/yexer
An experimental method for testing novel retinal vital stains Timothy L. Jacksona,d,*, Lewis Griffinb,d, Brendan Votea, Jost Hillenkampa,c, John Marshalla a
Academic Department of Ophthalmology, The Rayne Institute, St Thomas’ Hospital, Lambeth Palace Road, London SE1 7EH, UK b Imaging Sciences, King’s College London, London, UK c Eye Hospital, University of Regensburg, Regensburg, Germany d Vitrocetinal Unit, Moorfields Eye Hospital, London EC1V 2PD, UK Received 10 November 2004; accepted in revised form 7 March 2005 Available online 31 May 2005
Abstract There is uncertainty surrounding the safety of the vital stains currently used to assist macular surgery, and there may be other agents that are more suitable. This study aimed to validate a method of screening retinal vital stains for their potential surgical utility. Bovine retina was exposed to test agents at a range of concentrations. Masked observers determined the minimum dye concentration that reliably stained the retina, defined as the minimum visible concentration (MVC). Computer image analysis (CIE94 colour difference equation) was used to estimate the magnitude of the colour difference between stained and unstained retina. Agents that had favourable staining characteristics underwent safety testing using a retinal pigment epithelium and glial cell culture model. Cells were exposed to each agent and viability was assessed with a mitochondrial enzyme (MTT) assay, and fluorescent live–dead probe (ethidium homodimer-1/calcein-AM). Frozen sections were used to determine which retinal layers were stained. Techniques were tested on the following agents: alcian blue; diethyloxadicarbocyanine; Evan’s blue; fast green; fluorescein; Janus green; methylene blue; naphthol green; neutral red; procian (reactive) yellow; rose bengal; and trypan blue. For most dyes, the results of image analysis showed that colour differences increased linearly with dye concentration, although some displayed a more exponential relationship. Five agents showed favourable staining characteristics: Evan’s blue, rose bengal, naphthol green, neutral red, and trypan blue (MVC 0.02, 0.01, 0.1, 0.002, 0.01%, respectively). Safety testing of these five agents did not show toxicity, except in glial cells exposed to rose bengal. Relative to the negative control (saline), these showed a 48% reduction in viability using the MTT assay (p!0.001; tZ4.71; CI 30–75%), and qualitative damage on fluorescence microscopy. Frozen sections showed that some agents produced diffuse staining of all retinal layers, others produced selective inner retinal staining. There are thousands of biological stains available and many of these may be more effective or safer than those currently used for retinal surgery. This study provides a means of screening potentially useful vital stains. q 2005 Elsevier Ltd. All rights reserved. Keywords: vital stain; retina; macula; toxicity; retinal pigment epithelium; Mu¨ller cell; glia
1. Introduction Indocyanine green (ICG) assisted macular surgery is well described. By staining the retinal internal limiting membrane (ILM), ICG makes an otherwise optically transparent structure much easier to visualise and remove. There has, however, been debate regarding its safety (Kampik and Sternberg, 2003; Sebag, 2004; Jackson, in press). Some recent clinical studies reported favourable results (Weinberger et al., 2002; Kwok et al., 2003; Kwok and Lai, 2003; * Corresponding author. E-mail address: [email protected] (T.L. Jackson).
0014-4835/$ - see front matter q 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.exer.2005.03.004
Slaughter and Lee, 2004); others suggested that it caused a functional visual loss (Gandorfer et al., 2001; Engelbrecht et al., 2002; Haritoglou et al., 2002; Uemura et al., 2003). Experimental studies have shown possible damage in human retinal pigment epithelium (RPE) (Sippy et al., 2001; Stalmans et al., 2002; Yam et al., 2003; Ho et al., 2003; Jackson et al., 2004a), retinal ganglion cells (Iriyama et al., 2004), and Mu¨ller cells (Jackson et al., 2004a). In vivo animal experiments also suggested a toxic effect (Enaida et al., 2002; Maia et al., 2004), but only when using prolonged contact times or concentrations that were higher than those used clinically. More recently, infracyanine green has been advocated as an alternative vital stain (Stalmans et al., 2003). This agent is closely related to ICG but is theoretically safer, as it comes with a glucose diluent that results in a more iso-osmotic solution. Experimental
T.L. Jackson et al. / Experimental Eye Research 81 (2005) 446–454
studies suggested that infracyanine green also produces deleterious effects in cultured human glia (Jackson et al., 2004b) and cadaver eyes (Haritoglou et al., 2004). Another alternative macular vital stain is trypan blue. This has been noted to produce deleterious effects on the sclera and optic nerve (Funahashi et al., 1980), and animal studies found retinal damage with high concentrations and prolonged contact times (Veckeneer et al., 2001). Other cell culture experiments (Jackson et al., 2004a) and clinical studies (Feron et al., 2002; Li et al., 2003; Stalmans et al., 2003; Perrier and Sebag, 2003a, 2003b) did not show toxicity. It is possible that ICG and trypan blue were initially selected as vital stains because clinicians and scientists are familiar with their use. ICG is well known for its use in choroidal angiography, and in laboratory settings trypan blue is commonly used to distinguish live cells that it exclude it, from dead cells that do not, the so-called trypan blue exclusion test. This property has been exploited to determine endothelial cell viability in corneal grafts. Trypan blue is also used to stain the crystaline lens capsule to assist capsulorhexus during cataract surgery (Melles et al., 1999). However, these agents are only two of an extremely large range of biological stains (Lillie, 1977; Jackson, 2003), and it seems likely that several other agents might be equally, if not more suitable. The ideal macular vital stain would have a number of characteristics. Most importantly, it would be safe for intraocular use. A number of factors will influence safety, but in general acid chromophores are thought to be safer than unmodified basic dyes (Sorsby et al., 1942). Agents with an established safety record of intraocular use would be advantageous, but this may limit the number of agents available. A second important attribute would be the ability to reliably and selectively stain the ILM. Rapid elimination from the eye would be another advantage, as there are concerns (Weinberger et al., 2001) that ICG may persist for many weeks, if not months. The effect of this prolonged contact time is not known, but it may possibly increase the risk of adverse effects. A suitable absorption profile may also reduce the risk of light mediated damage when used in conjunction with surgical endoillumination, or maximise the visualisation of stained tissue against the predominantly red-orange retinal background. This study aimed to develop a technique to test the potential surgical utility of retinal vital stains. Identifying a better vital stain than those currently available would offer surgeons a safer and more effective tool to assist ILM removal during macular hole surgery.
2. Materials and methods 2.1. Selection of vital stains A selection of agents were used to validate the experimental method. These agents were thought to have
447
the potential clinical utility, or have staining characteristics that might suggest groups of chemically related agents that would themselves have potential clinical use. The selected chromophores were: alcian blue; diethyloxadicarbocyanine (DiOC); Evan’s blue; fast green; fluorescein; Janus green; methyl blue; naphthol green; neutral red; procian yellow (reactive yellow); rose bengal; and trypan blue. 2.2. Testing of vital stains Bovine eyes were obtained from a local abattoir and transported on ice. Experiments were done on the day of slaughter, or the morning after evening slaughter. The anterior segment was dissected free and the vitreous removed en bloc using gravity and a metal scoop. The retina was then separated from the optic nerve using dissecting scissors, and then approximately two-by-two millimetre flat-mounts were isolated from mid-peripheral, non-tapetal retina. These were floated onto a metal spatula and transferred individually to a small glass, watch-maker’s bowl containing the test agent. Reported exposure times for vital staining range from less than 30 sec to 5 min, so exposure time was arbitrarily set at 3 min. A briefer exposure might fail to identify dyes that require a few minutes to bind effectively, but dyes binding more slowly than this would extend surgical times and might limit clinical acceptability. After 3 min, tissue was transferred into phosphate buffered saline (PBS; Sigma-Aldrich, Poole, UK), and gently agitated to remove any dye that was not bound to retinal tissue. Each dye was initially tested at a minimum of five concentrations, typically 0.0001, 0.001, 0.01, 0.1, 1, and 10% weight/volume. PBS was used as the diluent. Other concentrations were sometimes selected based on clinical reports or availability of stock solutions. Further testing was then bracketed around the most useful concentrations, to account for the fact that the clinical application of vital stains varies, with some delivered into an air filled eye, and others injected into a fluid filled eye, over the macular. In this setting the exact concentration at the retinal surface is unknown, and will vary. Agents were prepared fresh each day and filtered through hardened ashless filter paper (Whatman, Maidstone, UK). Dyes were obtained from Sigma-Aldrich, except fluorescein and rose bengal (Chauvin Pharmaceuticals, Surrey, UK). A descriptive assessment of staining characteristics and potential clinical utility was made at by an ophthalmic surgeon (TLJ). This included features such as the uniformity and reliability of retinal staining, speed of dye uptake, and ability to rinse the dye from stained tissue. Retina was then photographed on a white background using a macrostand, Olympus OM-10 single lens reflex camera, Kodak Ektachrome T64 film, and standardized illumination and exposure routines. The developed photographic slides were then viewed in a standardized manner by two masked observers, to determine the minimum visible concentration (MVC). This was defined as the concentration at which they
448
T.L. Jackson et al. / Experimental Eye Research 81 (2005) 446–454
could correctly discriminate a difference from unstained retina in five out of five attempts. Control experiments were undertaken using retina that had been killed by pre-treatment with 70% methanol for 10 min, to determine if the staining pattern in experimental groups was caused by post-mortem cell death. These experiments were also repeated using heat-damaged tissue. Using a modification of a technique designed to maintain conjunctival hydration during experiments (Whittle et al., 1998), a slice of cucumber was placed in the bottom of a petri-dish and the retina was floated, ILM uppermost, over this. The fluid level was then lowered with a syringe so that the retinal tissue was supported above the fluid level by the cucumber. The high water content of the cucumber served to maintain tissue hydration, but by temporarily keeping the retina out of fluid in the petri-dish, a central retinal burn could be created. A 23-gauge (0.64 mm diameter) needle was heated until red hot and then used to create a needle puncture in the centre of the flat-mount. The retina was then stained in the usual manner. Damaged tissue at the margin of the burn was compared to healthy neighbouring tissue. 2.3. Image analysis The relationship between dye concentration and degree of staining was quantified by computer image analysis. The 35 mm photographic slides of stained retina were converted to high resolution tagged information file format using a photographic slide scanner (LS-4000, Nikon, Kingston upon Thames, UK). The CIE94 colour difference equation (McDonald and Smith, 1995) was used to estimate the magnitude of the colour difference between stained and unstained retina. This was done by converting the red–green–blue (RGB) values averaged over each image into XYZ values. It was assumed that the RGB sensitivites of the combined camera plus digitisation pipeline were calibrated for reproduction on a Recommendation 709 monitor using a display gamma of 2.4 (International Telecommunications Union, 1990). Von Kries’s adaptation (1970) was used to transform the XYZ values so that the unstained retina mapped to the D65 white point. The CIE94 equation was then applied to these transformed colour co-ordinates. As this was only intended as a screening test that could be applied to potentially large numbers of dyes, some of the more sensitive measurement techniques, such as full spectral characterisation, were not undertaken. 2.4. Safety testing Agents that showed potential use as a retinal vital stain underwent safety testing on cultured human retinal pigment epithelium (RPE) and Mu¨ller cells using techniques described previously (Jackson et al., 2004a). Briefly, ARPE-19 (passage 23; American Type Culture Collection, Manassas, VA) were cultured in Ham’s F-10 media (pH 7.4; Sigma-Aldrich), supplemented with 2 mM glutamine, 25 mM LK1
Hepes, 10 IU mLK1 penicillin, 10 mg mLK1 streptomycin, and 15% heat-inactivated fetal calf serum (Sigma-Aldrich). The spontaneously immortalized human Mu¨ller cell line MIO-M1 (Limb et al., 2002), gift of GA Limb, was cultured in Dulbecco’s modified Eagles medium containing L-glutamax 1 (Gibco BRL, Paisley, Scotland), supplemented with 2 mM glutamine, 10 IU mLK1 penicillin, 10 mg mLK1 streptomycin, and 10% heat-inactivated fetal calf serum (SigmaAldrich). The isolation and characterisation of these cells is described elsewhere (Limb et al., 2002). RPE and glial cells were grown to confluence in an incubator with a humidified atmosphere of 5% CO2, 95% air at 378C, then seeded into 96-well flat-bottom plates (TPP, Trasadingen, Switzerland) and 8-well chamber slides (Nunc, Inc., Naperville,IL). Once cells reached confluence the culture media was removed and 50 mL of the test agent was placed in each well. A standard 0.02% concentration was used for all dyes to allow comparison. This concentration was selected as it produced discernable retinal staining (at or above the MVC) with most agents. After 5 min, the dye was removed and the monolayer was carefully rinsed three times with PBS. This was then replaced with culture media and the cells were returned to the incubator. Cells exposed to PBS were used as the negative controls and were subject to identical rinsing routines. Cells killed by exposure to 30% methanol were used as positive controls. At day 1, a quantitative measure of cell viability was made using an MTT assay. Because of the theoretical risk that residual dye might alter the optical density of the cell monolayer and interfere with the assay, an adjustment was made by subtracting baseline readings (immediately after MTT was added to the well) from the final reading taken 4 hr later (Jackson et al., 2004b). Hence, the final adjusted value was not affected by any spectral overlap of the dye and the blue formazan reaction product tested with this assay. Cells were defined as having reduced viability if their optical density on the microplate reader was below 2 S.D. of the negative control. The optical density of each group was expressed as a percentage of the negative control (set at 100%), hence values less than 100% represented lower concentrations of formozan reaction product and reduced cell viability. Group comparisons were made using the unpaired t test and p-values less than 0.05 were considered significant. A qualitative assessment of cell viability was made using a fluorescent live–dead probe (Molecular probes, Eugene, OR) comprising 2 mM calcein-AM (CAM) and 2 mM ethidium homodimer-1 (EH-1). Staining routines, microscopy, and photography have been reported (Jackson et al., 2004a). 2.5. Frozen section To determine the effect of staining just the inner retinal surface (as occurs during macular surgery), in situ retina
T.L. Jackson et al. / Experimental Eye Research 81 (2005) 446–454
was exposed to those dyes that showed useful staining characteristics in isolated retina, and no toxicity during cell culture testing. Frozen sections were used to determine which layers were stained. The anterior segment was dissected from fresh bovine eyes and the vitreous was removed en bloc. The resulting eyecups were filled with approximately 1 mL of the candidate dye at concentrations bracketed around the MVC. The eyecups were positioned so that the dye pooled over the inferior, non-tapetal retina. After 3 min the dye was rinsed three times with saline and then a retina-RPE-scleral trephine was taken from the area exposed to the dye. Care was taken to ensure that only the inner retinal surface contacted the dye. Trephines were embedded (Bright Instruments, Huntingdon, UK) then immersed in a eutectic solution of isopentate (K1508C), cooled in a liquid nitrogen bath. Frozen sections (10–15 mm) were cut on a cryostat (Anglia Scientific Instruments, Cambridge, UK), placed on plain microscope slides, and air-dried. To prevent dye dispersion, no mounting medium was applied. Slides were viewed and photographed (Kodak Ectachrome 64T) on a light microscope (Leitz, Wezlar, Germany).
449
3. Results 3.1. Testing of vital stains Of the dyes tested, Evan’s blue, rose bengal, naphthol green, neutral red and trypan blue appeared to show the most potential as vital stains. These five agents produced reliable staining of the retinal surface in no more than 1 min. An example is shown in Fig. 1A–F. Although all were relatively easy to rinse free, neutral red tended to produce a more prolonged staining than the other four. The MVC that reliably stained the retinal surface is shown in Table 1. The pattern of staining altered markedly in retina pre-treated with methanol or those with heat damage. Both had increased dye uptake (Fig. 1G and H), suggesting that the pattern of retinal staining in other specimens was not caused by post-mortem cell death. Of those that showed less potential, alcian blue produced relatively little staining at concentrations less than 2%. By contrast, janus green and diethyloxadicarbocyanine were effective in staining the retinal surface at low concentrations, but sometimes produced spotty staining,
Fig. 1. Figure shows examples of bovine retina stained with vital stains. Figures A–E show serial dilutions of rose bengal. When shown to masked observers, the minimum visible concentration that could be repeatedly distinguished from the control (F) was 0.01%. Image G shows retina that had been perforated with a heated needle and then stained with fast green. The heat-damaged tissue shows marked uptake of dye relative to undamaged tissue. Image H shows the same experiment repeated following pre-treatment with methanol. The entire retina stained green and the selective staining of heat-damaged tissue was not as evident. Image I shows one of the dyes (janus green) that produced unreliable staining, with patchy uptake in comparison to the more even staining seen with rose bengal. BarsZ100 mm (A–F and I), and 500 mm (G and H).
450
T.L. Jackson et al. / Experimental Eye Research 81 (2005) 446–454
Table 1 Table shows the minimum dye concentration that produced retinal staining that could be detected by two masked observers in five out of five attempts Chromophore
Minimum visible concentration, MVC (% weight/volume)
Alcian blue Diethyloxadicarbocyanine Evan’s blue Fast green Fluorescein Janus green Methylene blue Naphthol green Neutral red Reactive yellow Rose bengal Trypan blue
Observer 1
Observer 2
2 0.0024 0.02 2 0.2 0.04 0.2 1 0.002 1 0.01 0.1
2 0.0024 0.02 2 0.2 0.04 0.2 0.1 0.002 0.1 0.01 0.01
even, reliable staining at low concentrations, but appeared to alter the retinal structure, with tissue becoming less compliant to handling. As observed previously with trypan blue (Jackson and Marshall, 2004), both janus green and rose bengal produced some selective staining of the cut outer edge of the retinal flat-mount. This appeared to constitute increased uptake in the cells that were traversed or damaged by the cut used to isolate the flat-mount from whole retina. 3.2. Image analysis
particularly with higher concentrations (Fig. 1I). Fast green produced homogenous staining in some retinal specimens at concentrations as low as 0.05%, but in others up to 2% was required. The cause of this variability was not identified. Fluorescein stained the retina reliably at concentrations of 0.2%, but the yellow hue it imparted produced relatively little colour contrast with unstained retina. Similar findings were made with procian yellow. Methylene blue produced an
The results of image analysis are shown in Fig. 2. For nine of the 12 dyes, the calculated colour differences increase monotonically with dye concentration. The exceptions were Evan’s blue, fluorescein and rose bengal, for which the colour difference for the highest concentration was less than that of the second highest. From visual inspection of the graphs a threshold of 35DE94 units was selected as best corresponding to the MVC. This is shown in Fig. 2 as the horizontal line on each graph. For most dyes this fell within 1 data point (1 log unit of dye concentration) of the MVC (shown as vertical lines). The exceptions were fluorescein, neutral red, methylene blue, and alcian blue. For these dyes the 35DE94 threshold was within 2 log units of the MVC.
DiOC
Alcian Blue
Evan's Blue
FastGreen
125
125
125
125
100
100
100
100
75
75
75
75
50
50
50
50
25
25
25
25
0.0002 0.002
0.02
0.2
2
0.00024 0.0024 0.024
Fluorescein
0.24
2.4
0.002
0.02
0.2
2
0.002
20
MethyleneBlue
JanusGreen 125
125
125
100
100
100
100
75
75
75
75
50
50
50
50
25
25
25
25
0.02
0.2
2
20
0.0004 0.004
NeutralRed
0.04
0.4
0.002
4
ProcianYellow
0.02
0.2
2
20
0.002
125
125
100
100
100
100
75
75
75
75
50
50
50
50
25
25
25
25
0.2
2
0.001
0.01
0.1
1
10
0.0001 0.001
0.01
0.1
2
20
0.2
2
20
TrypanBlue
125
0.02
0.02
RoseBengal
125
0.0002 0.002
0.2
NaphtholGreen
125
0.002
0.02
1
0.002
0.02
0.2
2
20
Fig. 2. Graphs showing the CIE94 colour differences (y-axis) between stained and unstained retina for varying dye concentrations (x-axis). The vertical error bars show G1 S.D. and were calculated using the data from different retinal samples (nR3). The vertical dashed lines show the minimum visible concentration that produced reliable staining of the retinal surface, judged by two blinded observers; where only one line is shown the observers agreed. The horizontal dashed line at 35DE94 units is the calculated colour difference threshold that best agrees with the observers’ judgements.
T.L. Jackson et al. / Experimental Eye Research 81 (2005) 446–454
451
Fig. 3. Graph shows an MTT assay of human retinal pigment epithelium and Mu¨ller cells exposed to various dyes. There was no significant reduction in viability relative to the negative control (saline) in any group except Mu¨ller cells exposed to rose bengal. These showed a statistically significant reduction in viability (p!0.001). Error barsG1 S.D.
3.3. Safety testing The viability of glial cells exposed to rose bengal was significantly reduced to 48% of the negative control (p!0.001; tZ4.71; CI of difference 30–75%) when tested using the MTT assay (Fig. 3). The live–dead probe also showed cell damage in this group, with large numbers of dead glia (Fig. 4). The toxicity data for trypan blue has been presented previously: no RPE or glial damage was observed with concentrations of up to 0.2% (Jackson et al., 2004a). The viability of glial cells exposed to Evan’s blue was slightly reduced to 87% of the negative control, but this difference was not significant (pZ0.2815; tZ1.09; CI K11 to 38%), and no consistent damage was observed with the live–dead probe. The other groups did not show toxicity with either the MTT assay or live–dead probe.
Black (1947) and Kutschera (1969) attempted similar experiments with local rather than systemic administration of dyes. Since that time there has been relatively little interest in the subject until the recent, widespread introduction of ICG as a macular vital stain, and more recent reports advocating infracyanine green and trypan blue. These agents have been introduced without a published, systematic appraisal of why they are the most suitable dyes. Rather, the dyes that are most readily available in the laboratory and clinic have been adapted for macular use, often before extensive experimental safety studies have been completed. This has probably been possible because these vital stains have been employed as surgical devices, avoiding the more rigorous safety testing required for pharmacotherapeutic agents. This approach is
3.4. Frozen section Frozen sections of retina exposed to naphthol green, Evan’s blue, and neutral red all showed staining of the ILM, but diffusion into the retinal substrate varied (Fig. 5). Overall, Evan’s blue produced the most selective staining of the ILM. Despite producing easily visible macroscopic retinal staining, methylene blue showed much less distinct staining in section, with light uptake in all retinal layers in some sections—others appeared similar to the unstained control. Trypan blue showed only faint staining on frozen section and was often difficult to distinguish from unstained retina.
4. Discussion Sorsby (Sorsby et al., 1937; Sorsby, 1938, 1939a, 1939b) described in vivo retinal staining more than 65 years ago, and
Fig. 4. Fluorescence photomicrograph of Mu¨ller cells exposed to rose bengal 0.02%, then stained with the live–dead probe one day later. The cytoplasm of live cells are stained green with calcein, and the nuclear matter of dead or dying cells are stained red with ethidium homodimer-1. Dead cells can be seen scattered throughout the image, suggesting a toxic effect that was not seen in the negative control.
452
T.L. Jackson et al. / Experimental Eye Research 81 (2005) 446–454
Fig. 5. Photomicrograph of bovine retinal frozen sections exposed to methylene blue, neutral red, Evan’s blue, and naphthol green (left to right). The methylene blue stained section reveals a very faint blue stain in all retinal layers, but the appearance was only marginally different to the unstained control. Trypan blue showed even less definite staining on frozen section and could not reliably be distinguished from unstained retina (not shown). Neutral red showed staining of the internal limiting membrane (ILM), inner nerve fibre layer, and possibly inner half of the retina. Evan’s blue staining was predominantly restricted to the ILM. Naphthol green showed preferential binding in the ILM and nerve fibre layer, but some green staining was visible throughout all retinal layers. Scale: retinal thickness measures approximately 150 mm.
not ideal, and the present uncertainty (Kampik and Sternberg, 2003; Sebag, 2004; Jackson, 2005) regarding the safety of ICG highlights this fact. It is likely that ICG, infracyanine green, and trypan blue represent only a small proportion of the agents that might be suitable for vital staining. In general, biological stains have specific atomic groupings (CaS, CaN, NaN, NaO and NO2) that are known to impart colour (Lillie, 1977). A number of these have been used as vital stains in other clinical specialties such as dentistry (van de Rijke, 1991) and urology (Gill and Strauss, 1984). Many hundreds of chromophores are used by the food and textile industries, and in medical and scientific laboratories to stain tissue for light microscopy. This study tested only a small proportion of available agents. It used a series of dyes to validate a systematic screening method that could be applied to a wide range of dyes. This in turn may help to identify a vital stain that is less toxic than ICG, and more effective than trypan blue. This has the potential to improve the ease and safety of surgical ILM removal; a procedure that is thought to contribute to the success of macular hole surgery. Twelve dyes were selected for testing: alcian blue; diethyloxadicarbocyanine; Evan’s blue; fast green; fluorescein; janus green; methylene blue; naphthol green; neutral
red; procian yellow (reactive yellow); rose bengal; and trypan blue. Of these, five showed favourable staining characteristics: Evan’s blue; naphthol green; neutral red; rose bengal; and trypan blue. One of these, trypan blue, has recently been marketed as a macular vital stain and was included for comparison (Feron et al., 2002; Stalmans et al., 2003; Rodrigues, 2003; Perrier and Sebag, 2003a, 2003b). All five agents showed an ability to rapidly stain the retina and were relatively easy to rinse free, unlike agents such as infracyanine green (Jackson et al., 2004b). Safety testing of these five agents did not show any toxicity except in glial cells exposed to rose bengal. These showed significant damage using a mitochondrial enzyme assay and a live–dead probe that identifies cells with compromised cell membranes. Although cell culture cannot be used to conclude on the safety of these agents as a surgical tool, it is a useful preliminary test to rule out agents with clear toxicity. The findings suggest rose bengal may not be appropriate for retinal use, but with respect to the other dyes it is only possible to conclude that further safety testing would be worthwhile. Use of these agents in humans would not yet be appropriate, with the exception of trypan blue. Establishing the safety of any candidate dye would ideally involve experiments in an animal with rods and cones, a duplex retinal circulation, and a globe size adequate to perform vitrectomy. Exposure routines would involve focal delivery and removal of the dye after a brief interval, and adequate follow up to detect delayed toxicity. Although image analysis was useful in determining the relationship of dye concentration to degree of retinal staining, it had some limitations. For three agents, Evan’s blue, fluorescein, and rose bengal, the highest dye concentration produced a lower colour difference value than the second highest concentration (Fig. 4). Inspection of a representative image (Fig. 1) shows that high concentrations produced dark retinal staining, which necessarily means that samples were less saturated with the hue of the dye. Colour difference equations such as CIE94 attempt to correctly weight differences in hue, saturation, and lightness. It is not known whether the use of CIE94 produces valid weighting at these high concentrations. This does not, however, limit the applicability of this technique, as the finding only occurred at concentrations well above those that might be used clinically. It should be noted that although the image analysis provides a relatively simple means providing quantitative data, it would not be able to replace the MVC as a screening method. Further, image analysis considered the effect of staining isolated retina that had both inner and outer surfaces exposed to dye. Image analysis could therefore exclude dyes that failed to effectively bind retinal tissue, but it could not identify those that selectively stained the ILM. To determine which dyes selectively stain the ILM, frozen sections were viewed on a light microscope (Fig. 5). Retina exposed to dye on only the inner retinal surface showed
T.L. Jackson et al. / Experimental Eye Research 81 (2005) 446–454
different staining patterns, depending on which dye was used. Agents such as methylene blue produced easily visible macroscopic staining with concentrations of 0.2%, but staining was dispersed throughout the retinal layers. Hence, 10–15 mm frozen sections showed only faint staining. By contrast, the concentration required to produce macroscopic staining with Evan’s blue (0.02%) produced definite staining on frozen section, as dye uptake was predominantly at the level of the ILM. This selectivity may be advantageous in a vital stain designed to delineate peeled ILM. This study only considered agents that would be employed as chromophores, but it would be interesting to investigate certain fluorophores. This is a potentially useful route of enquiry as many of the fluorophores used histologically have extremely high levels of tissue specificity. Relatively simple optical modification of the operating microscope means that they could be selectively viewed during surgery, as barrier filters are already employed to protect surgeons from macular injury during endoscopic laser application. We have previously shown that it is possible to selectively identify glial cells with damaged cell membranes during surgery, using a fluorophore tagged antibody to the intracellular filament, glial fibrillary acidic protein (GFAP) (Jackson and Marshall, 2004). It is possible that removal of ILM would also result in selective staining with this agent, as the ILM is formed by the basement membrane of the Mu¨ller cell foot-plates. There are many other fluorophores that might also be useful as vital stains. Unlike the introduction of present-day macular vital stains, it would be helpful to have a systematic approach to test potentially useful agents, with preliminary ex vivo safety testing prior to animal, then human studies. This study attempted to validate a screening system by testing it on a selection of chromophores. There are, however, many more chromophores available, and considerable scope for further research.
Acknowledgements Supported by Deutsche Forschungsgemeinschaft (DFG) grant Hi 758/1-1, The Allerton Fund, and Special Trustees of Guys and St Thomas’ Hospital. The authors thank G. Astrid Limb (Institute of Ophthalmology, London, UK) for providing the Mu¨ller cells used in this study, and Paul Constable and Bruce Knight for assistance with RPE cell culture.
References Black, G.W., 1947. Some aspects of the treatment of simple detachment of the retina, including vital staining of the retina by methylene blue. Trans. Ophthalmol. Soc. UK 67, 313–322. Enaida, H., Sakamoto, T., Hisatomi, T., Goto, Y., Ishibashi, T., 2002. Morphological and functional damage of the retina caused by intravitreous indocyanine green in rat eyes. Graefes Arch. Clin. Exp. Ophthalmol. 240, 209–213.
453
Engelbrecht, N.E., Freeman, J., Sternberg Jr., P., Aaberg Sr., T.M., Aaberg Jr., T.M., Martin, D.F., Sippy, B.D., 2002. Retinal pigment epithelial changes after macular hole surgery with indocyanine green-assisted internal limiting membrane peeling. Am. J. Ophthalmol. 133, 94. Feron, E.J., Veckeneer, M., Parys-Van Ginderdeuren, R., Van Lommel, A., Melles, G.R.J., Stalmans, P., 2002. Trypan blue staining of epiretinal membranes in proliferative vitreoretinopathy. Arch. Ophthalmol. 120, 141–144. Funahashi, T., Kondo, K., Kimura, M., 1980. The vital staining of the eye. Acta Soc. Ophthalmol. Jpn. 84, 1826–1880. Gandorfer, A., Haritoglou, C., Gass, C.A., Ulbig, M.W., Kampik, A., 2001. Indocyanine green-assisted peeling of the internal limiting membrane may cause retinal damage. Am. J. Ophthalmol. 132, 431–433. Gill, W.B., Strauss, F.H., 1984. In vivo mapping of bladder cancer (chromocystoscopy for in vivo detection of neoplastic urothelial surfaces). Urology 23, 63–66. Haritoglou, C., Gandorfer, A., Gass, C.A., Schaumberger, M., Ulbig, M.W., Kampik, A., 2002. Indocyanine green-assisted peeling of the internal limiting membrane in macular hole surgery affects visual outcome: a clinicopathologic correlation. Am. J. Ophthalmol. 134, 836–841. Haritoglou, C., Gandorfer, A., Gass, C.A., Kampik, A., 2004. Histology of the vitreoretinal interface after staining of the internal limiting membrane using glucose 5% diluted indocyanine green and infracyanine green. Am. J. Ophthalmol. 137, 345–348. Ho, J.D., Chen, H.-C., Chen, S.-N., Tsai, R.J.-F., 2003. Cytotoxicity of indocyanine green on retinal pigment epithelium: implications for macular hole surgery. Arch. Ophthalmol. 121, 1423–1429. International Telecommunications Union, 1990. BT. 709. Basic parameter values for the HDTV standard for the studio and for international programme exchange. International Telecommunications Union, Geneva. Iriyama, A., Uchida, S., Yanagi, Y., Tamaki, Y., Inoue, Y., Matsuura, K., Kadonosono, K., Araie, M., 2004. Effects of indocyanine green on retinal ganglion cells. Invest. Ophthalmol. Vis. Sci. 45, 943–947. Jackson, T.L., 2003. An investigation of novel techniques to enhance the detection of retinal breaks in detachment surgery. University of London, PhD Thesis. pp. 191–197. Jackson, T.L., 2005. Indocyanine green accused: the case for and against ICG-assisted macular surgery. Br. J. Ophthalmol 89, 395–396. Jackson, T.L., Marshall, J., 2004. Fluorophore assisted retinal break detection using antibodies to glial fibrillary acidic protein. Invest. Ophthalmol. Vis. Sci. 45, 993–1001. Jackson, T.L., Hillenkamp, J., Knight, B.C., Zhang, J.J., Thomas, D., Stanford, M.R., Marshall, J., 2004a. Safety testing of indocyanine green and trypan blue using retinal pigment epithelium and glial cell cultures. Invest. Ophthalmol. Vis. Sci. 45, 2778–2785. Jackson, T.L., Vote, B., Knight, B.C., El-Amir, A., Standford, M.R., Marshall, J., 2004b. Safety testing of infracyanine green using retinal pigment epithelium and glial cell cultures. Invest. Ophthalmol. Vis. Sci. 45, 3697–3703. Kampik, A., Sternberg, P., 2003. Indocyanine green in vitreomacular surgery (why) is it a problem?. Am. J. Ophthalmol. 136, 527–529. Kutschera, E., 1969. [Vital staining of the detached retina with retinal breaks]. Albrecht Von Graefes Archiv fur Klinische und Experimentelle Ophthalmologie 178, 72–87. Kwok, A.K., Lai, T.Y., 2003. Internal limiting membrane removal in macular hole surgery for severely myopic eyes: a case–control study. Br. J. Ophthalmol. 87, 885–889. Kwok, A.K., Lai, T.Y., Man-Chan, W., Woo, D.C., 2003. Indocyanine green assisted retinal internal limiting membrane removal in stage 3 or 4 macular hole surgery. Br. J. Ophthalmol. 87, 71–74. Li, K., Wong, D., Hiscott, P., Stanga, P., Groenewald, C., McGalliard, J., 2003. Trypan blue staining of internal limiting membrane and epiretinal membrane during vitrectomy: visual results and histopathological findings. Br. J. Ophthalmol. 87, 216–219.
454
T.L. Jackson et al. / Experimental Eye Research 81 (2005) 446–454
Lillie, R.D., 1977. H.J. Conn’s Biological Stains, ninth ed. Williams & Wilkins, Baltimore. Limb, G.A., Salt, T.E., Munroe, P.M.G., Moss, S.E., Khaw, P.T., 2002. In vitro characterization of a spontaneously immortalized human Mu¨ller cell line. Invest. Ophthalmol. Vis. Sci. 43, 864–869. Maia, M., Pieramici, D.J., Margalit, E., Farah, M.E., Lakhanpal, R.R., Au Eong, K.-G., 2004. Retinal pigment epithelial abnormalities after internal limiting membrane peeling guided by indocyanine green staining. Retina 24, 157–160. McDonald, R., Smith, K.J., 1995. CIE94—a new colour-difference formula. J. Soc. Dyers Colourists 111, 376–379. Melles, G.R.J., de Waard, P.W.T., Pameyer, J.H., Beekhuis, W.H., 1999. Trypan blue capsule staining to visualize the capsulorhexis in cataract surgery. J. Cataract Refract. Surg. 25, 7–9. Perrier, M., Sebag, M., 2003a. Epiretinal membrane surgery assisted by trypan blue. Am. J. Ophthalmol. 135, 909–911. Perrier, M., Sebag, M., 2003b. Trypan blue-assisted peeling of the internal limiting membrane during macular hole surgery. Am. J. Ophthalmol. 135, 903–905. Rodrigues, E.B., 2003. Trypan blue stains the epiretinal membrane but not the internal limiting membrane. Br. J. Ophthalmol. 87, 1431–1432. Sebag, J., 2004. Indocyanine green-assisted macular hole surgery: too pioneering?. Am. J. Ophthalmol. 4, 744–746. Sippy, B.D., Engelbrecht, N.E., Hubbard, G.B., Moriarty, S.E., Jiang, S., Aaberg Jr., T.M., Aaberg Sr., T.M., Grossniklaus, H.E., Sternberg Jr., P., 2001. Indocyanine green effect on cultured human retinal pigment epithelial cells: implications for macular hole surgery. Am. J. Ophthalmol. 132, 433–435. Slaughter, K., Lee, I.L., 2004. Macular hole surgery with and without indocyanine green assistance. Eye 18, 376–378. Sorsby, A., 1938. Two patients with vital staining of the fundi. Trans. Ophthal. Soc. UK 58, 275. Sorsby, A., 1939a. Vital staining of the fundus. Trans. Ophthal. Soc. UK 59, 727–730. Sorsby, A., 1939b. Vital staining of the retina: preliminary clinical note. Br. J. Ophthalmol. 23, 20–24. Sorsby, A., Elkeles, A., Goodhart, G.W., Morris, I.B., 1937. Experimental staining of the retina in life. Proc. R. Soc. Med. 30, 1271–1273.
Sorsby, A., Wright, A.D., Elkeles, A., 1942. Vital staining in brain surgery. A preliminary note. Proc. R. Soc. Med. 36, 137–144. Stalmans, P., Van Aken, E.H., Veckeneer, M., Feron, E.J., Stalmans, I., 2002. Toxic effect of indocyanine green on retinal pigment epithelium related to osmotic effects of the solvent. Am. J. Ophthalmol. 134, 282–285. Stalmans, P., Feron, E.J., Parys-Van Ginderdeuren, R., Van Lommel, A., Melles, G.R., Veckeneer, M., 2003. Double vital staining using trypan blue and infracyanine green in macular pucker surgery. Br. J. Ophthalmol. 87, 713–716. Uemura, A., Kanda, S., Sakamoto, Y., Kita, H., 2003. Visual field defects after uneventful vitrectomy for epiretinal membrane with indocyanine green-assisted internal limiting membrane peeling. Am. J. Ophthalmol. 136, 252–257. van de Rijke, J.W., 1991. Use of dyes in cariology. Int. Dent. J. 41, 111–116. Veckeneer, M., van Overdam, K., Monzer, J., Kobuch, K., van Marle, W., Spekreijse, H., van Meurs, J., 2001. Ocular toxicity study of trypan blue injected into the vitreous cavity of rabbit eyes. Graefes Archiv. Clin. Exp. Ophthalmol. 239, 698–704. Von Kries, J., 1970. Chromatic adaptation. In: Sources of Colour Science. MIT Press, Cambridge, MA. Weinberger, A.W.A., Kirchhof, B., Mazinani, B.E., Schrage, N.F., 2001. Persistent indocyanine green (ICG) fluorescence 6 weeks after intraocular ICG administration for macular hole surgery. Graefes Arch. Clin. Exp. Ophthalmol. 239, 388–390. Weinberger, A.W., Schlossmacher, B., Dahlke, C., Hermel, M., Kirchhof, B., Schrage, N.F., 2002. Indocyanine-green-assisted internal limiting membrane peeling in macular hole surgery—a follow-up study. Graefes Archiv. Clin. Exp. Ophthalmol. 240, 913–917. Whittle, R.M., Wallace, G.R., Whiston, R.A., Dumonde, D.C., Stanford, M.R., 1998. Human antiretinal antibodies in toxoplasma retinochoroiditis. Br. J. Ophthalmol. 82, 1017–1021. Yam, H.F., Kwok, A.K., Chan, K.P., Lai, T.Y., Chu, K.Y., Lam, D.S., Pang, C.P., 2003. Effect of indocyanine green and illumination on gene expression in human retinal pigment epithelial cells. Invest. Ophthalmol. Vis. Sci. 44, 370–377.
An experimental method for testing novel retinal vital stains Timothy L. Jacksona,d,*, Lewis Griffinb,d, Brendan Votea, Jost Hillenkampa,c, John Marshalla a
Academic Department of Ophthalmology, The Rayne Institute, St Thomas’ Hospital, Lambeth Palace Road, London SE1 7EH, UK b Imaging Sciences, King’s College London, London, UK c Eye Hospital, University of Regensburg, Regensburg, Germany d Vitrocetinal Unit, Moorfields Eye Hospital, London EC1V 2PD, UK Received 10 November 2004; accepted in revised form 7 March 2005 Available online 31 May 2005
Abstract There is uncertainty surrounding the safety of the vital stains currently used to assist macular surgery, and there may be other agents that are more suitable. This study aimed to validate a method of screening retinal vital stains for their potential surgical utility. Bovine retina was exposed to test agents at a range of concentrations. Masked observers determined the minimum dye concentration that reliably stained the retina, defined as the minimum visible concentration (MVC). Computer image analysis (CIE94 colour difference equation) was used to estimate the magnitude of the colour difference between stained and unstained retina. Agents that had favourable staining characteristics underwent safety testing using a retinal pigment epithelium and glial cell culture model. Cells were exposed to each agent and viability was assessed with a mitochondrial enzyme (MTT) assay, and fluorescent live–dead probe (ethidium homodimer-1/calcein-AM). Frozen sections were used to determine which retinal layers were stained. Techniques were tested on the following agents: alcian blue; diethyloxadicarbocyanine; Evan’s blue; fast green; fluorescein; Janus green; methylene blue; naphthol green; neutral red; procian (reactive) yellow; rose bengal; and trypan blue. For most dyes, the results of image analysis showed that colour differences increased linearly with dye concentration, although some displayed a more exponential relationship. Five agents showed favourable staining characteristics: Evan’s blue, rose bengal, naphthol green, neutral red, and trypan blue (MVC 0.02, 0.01, 0.1, 0.002, 0.01%, respectively). Safety testing of these five agents did not show toxicity, except in glial cells exposed to rose bengal. Relative to the negative control (saline), these showed a 48% reduction in viability using the MTT assay (p!0.001; tZ4.71; CI 30–75%), and qualitative damage on fluorescence microscopy. Frozen sections showed that some agents produced diffuse staining of all retinal layers, others produced selective inner retinal staining. There are thousands of biological stains available and many of these may be more effective or safer than those currently used for retinal surgery. This study provides a means of screening potentially useful vital stains. q 2005 Elsevier Ltd. All rights reserved. Keywords: vital stain; retina; macula; toxicity; retinal pigment epithelium; Mu¨ller cell; glia
1. Introduction Indocyanine green (ICG) assisted macular surgery is well described. By staining the retinal internal limiting membrane (ILM), ICG makes an otherwise optically transparent structure much easier to visualise and remove. There has, however, been debate regarding its safety (Kampik and Sternberg, 2003; Sebag, 2004; Jackson, in press). Some recent clinical studies reported favourable results (Weinberger et al., 2002; Kwok et al., 2003; Kwok and Lai, 2003; * Corresponding author. E-mail address: [email protected] (T.L. Jackson).
0014-4835/$ - see front matter q 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.exer.2005.03.004
Slaughter and Lee, 2004); others suggested that it caused a functional visual loss (Gandorfer et al., 2001; Engelbrecht et al., 2002; Haritoglou et al., 2002; Uemura et al., 2003). Experimental studies have shown possible damage in human retinal pigment epithelium (RPE) (Sippy et al., 2001; Stalmans et al., 2002; Yam et al., 2003; Ho et al., 2003; Jackson et al., 2004a), retinal ganglion cells (Iriyama et al., 2004), and Mu¨ller cells (Jackson et al., 2004a). In vivo animal experiments also suggested a toxic effect (Enaida et al., 2002; Maia et al., 2004), but only when using prolonged contact times or concentrations that were higher than those used clinically. More recently, infracyanine green has been advocated as an alternative vital stain (Stalmans et al., 2003). This agent is closely related to ICG but is theoretically safer, as it comes with a glucose diluent that results in a more iso-osmotic solution. Experimental
T.L. Jackson et al. / Experimental Eye Research 81 (2005) 446–454
studies suggested that infracyanine green also produces deleterious effects in cultured human glia (Jackson et al., 2004b) and cadaver eyes (Haritoglou et al., 2004). Another alternative macular vital stain is trypan blue. This has been noted to produce deleterious effects on the sclera and optic nerve (Funahashi et al., 1980), and animal studies found retinal damage with high concentrations and prolonged contact times (Veckeneer et al., 2001). Other cell culture experiments (Jackson et al., 2004a) and clinical studies (Feron et al., 2002; Li et al., 2003; Stalmans et al., 2003; Perrier and Sebag, 2003a, 2003b) did not show toxicity. It is possible that ICG and trypan blue were initially selected as vital stains because clinicians and scientists are familiar with their use. ICG is well known for its use in choroidal angiography, and in laboratory settings trypan blue is commonly used to distinguish live cells that it exclude it, from dead cells that do not, the so-called trypan blue exclusion test. This property has been exploited to determine endothelial cell viability in corneal grafts. Trypan blue is also used to stain the crystaline lens capsule to assist capsulorhexus during cataract surgery (Melles et al., 1999). However, these agents are only two of an extremely large range of biological stains (Lillie, 1977; Jackson, 2003), and it seems likely that several other agents might be equally, if not more suitable. The ideal macular vital stain would have a number of characteristics. Most importantly, it would be safe for intraocular use. A number of factors will influence safety, but in general acid chromophores are thought to be safer than unmodified basic dyes (Sorsby et al., 1942). Agents with an established safety record of intraocular use would be advantageous, but this may limit the number of agents available. A second important attribute would be the ability to reliably and selectively stain the ILM. Rapid elimination from the eye would be another advantage, as there are concerns (Weinberger et al., 2001) that ICG may persist for many weeks, if not months. The effect of this prolonged contact time is not known, but it may possibly increase the risk of adverse effects. A suitable absorption profile may also reduce the risk of light mediated damage when used in conjunction with surgical endoillumination, or maximise the visualisation of stained tissue against the predominantly red-orange retinal background. This study aimed to develop a technique to test the potential surgical utility of retinal vital stains. Identifying a better vital stain than those currently available would offer surgeons a safer and more effective tool to assist ILM removal during macular hole surgery.
2. Materials and methods 2.1. Selection of vital stains A selection of agents were used to validate the experimental method. These agents were thought to have
447
the potential clinical utility, or have staining characteristics that might suggest groups of chemically related agents that would themselves have potential clinical use. The selected chromophores were: alcian blue; diethyloxadicarbocyanine (DiOC); Evan’s blue; fast green; fluorescein; Janus green; methyl blue; naphthol green; neutral red; procian yellow (reactive yellow); rose bengal; and trypan blue. 2.2. Testing of vital stains Bovine eyes were obtained from a local abattoir and transported on ice. Experiments were done on the day of slaughter, or the morning after evening slaughter. The anterior segment was dissected free and the vitreous removed en bloc using gravity and a metal scoop. The retina was then separated from the optic nerve using dissecting scissors, and then approximately two-by-two millimetre flat-mounts were isolated from mid-peripheral, non-tapetal retina. These were floated onto a metal spatula and transferred individually to a small glass, watch-maker’s bowl containing the test agent. Reported exposure times for vital staining range from less than 30 sec to 5 min, so exposure time was arbitrarily set at 3 min. A briefer exposure might fail to identify dyes that require a few minutes to bind effectively, but dyes binding more slowly than this would extend surgical times and might limit clinical acceptability. After 3 min, tissue was transferred into phosphate buffered saline (PBS; Sigma-Aldrich, Poole, UK), and gently agitated to remove any dye that was not bound to retinal tissue. Each dye was initially tested at a minimum of five concentrations, typically 0.0001, 0.001, 0.01, 0.1, 1, and 10% weight/volume. PBS was used as the diluent. Other concentrations were sometimes selected based on clinical reports or availability of stock solutions. Further testing was then bracketed around the most useful concentrations, to account for the fact that the clinical application of vital stains varies, with some delivered into an air filled eye, and others injected into a fluid filled eye, over the macular. In this setting the exact concentration at the retinal surface is unknown, and will vary. Agents were prepared fresh each day and filtered through hardened ashless filter paper (Whatman, Maidstone, UK). Dyes were obtained from Sigma-Aldrich, except fluorescein and rose bengal (Chauvin Pharmaceuticals, Surrey, UK). A descriptive assessment of staining characteristics and potential clinical utility was made at by an ophthalmic surgeon (TLJ). This included features such as the uniformity and reliability of retinal staining, speed of dye uptake, and ability to rinse the dye from stained tissue. Retina was then photographed on a white background using a macrostand, Olympus OM-10 single lens reflex camera, Kodak Ektachrome T64 film, and standardized illumination and exposure routines. The developed photographic slides were then viewed in a standardized manner by two masked observers, to determine the minimum visible concentration (MVC). This was defined as the concentration at which they
448
T.L. Jackson et al. / Experimental Eye Research 81 (2005) 446–454
could correctly discriminate a difference from unstained retina in five out of five attempts. Control experiments were undertaken using retina that had been killed by pre-treatment with 70% methanol for 10 min, to determine if the staining pattern in experimental groups was caused by post-mortem cell death. These experiments were also repeated using heat-damaged tissue. Using a modification of a technique designed to maintain conjunctival hydration during experiments (Whittle et al., 1998), a slice of cucumber was placed in the bottom of a petri-dish and the retina was floated, ILM uppermost, over this. The fluid level was then lowered with a syringe so that the retinal tissue was supported above the fluid level by the cucumber. The high water content of the cucumber served to maintain tissue hydration, but by temporarily keeping the retina out of fluid in the petri-dish, a central retinal burn could be created. A 23-gauge (0.64 mm diameter) needle was heated until red hot and then used to create a needle puncture in the centre of the flat-mount. The retina was then stained in the usual manner. Damaged tissue at the margin of the burn was compared to healthy neighbouring tissue. 2.3. Image analysis The relationship between dye concentration and degree of staining was quantified by computer image analysis. The 35 mm photographic slides of stained retina were converted to high resolution tagged information file format using a photographic slide scanner (LS-4000, Nikon, Kingston upon Thames, UK). The CIE94 colour difference equation (McDonald and Smith, 1995) was used to estimate the magnitude of the colour difference between stained and unstained retina. This was done by converting the red–green–blue (RGB) values averaged over each image into XYZ values. It was assumed that the RGB sensitivites of the combined camera plus digitisation pipeline were calibrated for reproduction on a Recommendation 709 monitor using a display gamma of 2.4 (International Telecommunications Union, 1990). Von Kries’s adaptation (1970) was used to transform the XYZ values so that the unstained retina mapped to the D65 white point. The CIE94 equation was then applied to these transformed colour co-ordinates. As this was only intended as a screening test that could be applied to potentially large numbers of dyes, some of the more sensitive measurement techniques, such as full spectral characterisation, were not undertaken. 2.4. Safety testing Agents that showed potential use as a retinal vital stain underwent safety testing on cultured human retinal pigment epithelium (RPE) and Mu¨ller cells using techniques described previously (Jackson et al., 2004a). Briefly, ARPE-19 (passage 23; American Type Culture Collection, Manassas, VA) were cultured in Ham’s F-10 media (pH 7.4; Sigma-Aldrich), supplemented with 2 mM glutamine, 25 mM LK1
Hepes, 10 IU mLK1 penicillin, 10 mg mLK1 streptomycin, and 15% heat-inactivated fetal calf serum (Sigma-Aldrich). The spontaneously immortalized human Mu¨ller cell line MIO-M1 (Limb et al., 2002), gift of GA Limb, was cultured in Dulbecco’s modified Eagles medium containing L-glutamax 1 (Gibco BRL, Paisley, Scotland), supplemented with 2 mM glutamine, 10 IU mLK1 penicillin, 10 mg mLK1 streptomycin, and 10% heat-inactivated fetal calf serum (SigmaAldrich). The isolation and characterisation of these cells is described elsewhere (Limb et al., 2002). RPE and glial cells were grown to confluence in an incubator with a humidified atmosphere of 5% CO2, 95% air at 378C, then seeded into 96-well flat-bottom plates (TPP, Trasadingen, Switzerland) and 8-well chamber slides (Nunc, Inc., Naperville,IL). Once cells reached confluence the culture media was removed and 50 mL of the test agent was placed in each well. A standard 0.02% concentration was used for all dyes to allow comparison. This concentration was selected as it produced discernable retinal staining (at or above the MVC) with most agents. After 5 min, the dye was removed and the monolayer was carefully rinsed three times with PBS. This was then replaced with culture media and the cells were returned to the incubator. Cells exposed to PBS were used as the negative controls and were subject to identical rinsing routines. Cells killed by exposure to 30% methanol were used as positive controls. At day 1, a quantitative measure of cell viability was made using an MTT assay. Because of the theoretical risk that residual dye might alter the optical density of the cell monolayer and interfere with the assay, an adjustment was made by subtracting baseline readings (immediately after MTT was added to the well) from the final reading taken 4 hr later (Jackson et al., 2004b). Hence, the final adjusted value was not affected by any spectral overlap of the dye and the blue formazan reaction product tested with this assay. Cells were defined as having reduced viability if their optical density on the microplate reader was below 2 S.D. of the negative control. The optical density of each group was expressed as a percentage of the negative control (set at 100%), hence values less than 100% represented lower concentrations of formozan reaction product and reduced cell viability. Group comparisons were made using the unpaired t test and p-values less than 0.05 were considered significant. A qualitative assessment of cell viability was made using a fluorescent live–dead probe (Molecular probes, Eugene, OR) comprising 2 mM calcein-AM (CAM) and 2 mM ethidium homodimer-1 (EH-1). Staining routines, microscopy, and photography have been reported (Jackson et al., 2004a). 2.5. Frozen section To determine the effect of staining just the inner retinal surface (as occurs during macular surgery), in situ retina
T.L. Jackson et al. / Experimental Eye Research 81 (2005) 446–454
was exposed to those dyes that showed useful staining characteristics in isolated retina, and no toxicity during cell culture testing. Frozen sections were used to determine which layers were stained. The anterior segment was dissected from fresh bovine eyes and the vitreous was removed en bloc. The resulting eyecups were filled with approximately 1 mL of the candidate dye at concentrations bracketed around the MVC. The eyecups were positioned so that the dye pooled over the inferior, non-tapetal retina. After 3 min the dye was rinsed three times with saline and then a retina-RPE-scleral trephine was taken from the area exposed to the dye. Care was taken to ensure that only the inner retinal surface contacted the dye. Trephines were embedded (Bright Instruments, Huntingdon, UK) then immersed in a eutectic solution of isopentate (K1508C), cooled in a liquid nitrogen bath. Frozen sections (10–15 mm) were cut on a cryostat (Anglia Scientific Instruments, Cambridge, UK), placed on plain microscope slides, and air-dried. To prevent dye dispersion, no mounting medium was applied. Slides were viewed and photographed (Kodak Ectachrome 64T) on a light microscope (Leitz, Wezlar, Germany).
449
3. Results 3.1. Testing of vital stains Of the dyes tested, Evan’s blue, rose bengal, naphthol green, neutral red and trypan blue appeared to show the most potential as vital stains. These five agents produced reliable staining of the retinal surface in no more than 1 min. An example is shown in Fig. 1A–F. Although all were relatively easy to rinse free, neutral red tended to produce a more prolonged staining than the other four. The MVC that reliably stained the retinal surface is shown in Table 1. The pattern of staining altered markedly in retina pre-treated with methanol or those with heat damage. Both had increased dye uptake (Fig. 1G and H), suggesting that the pattern of retinal staining in other specimens was not caused by post-mortem cell death. Of those that showed less potential, alcian blue produced relatively little staining at concentrations less than 2%. By contrast, janus green and diethyloxadicarbocyanine were effective in staining the retinal surface at low concentrations, but sometimes produced spotty staining,
Fig. 1. Figure shows examples of bovine retina stained with vital stains. Figures A–E show serial dilutions of rose bengal. When shown to masked observers, the minimum visible concentration that could be repeatedly distinguished from the control (F) was 0.01%. Image G shows retina that had been perforated with a heated needle and then stained with fast green. The heat-damaged tissue shows marked uptake of dye relative to undamaged tissue. Image H shows the same experiment repeated following pre-treatment with methanol. The entire retina stained green and the selective staining of heat-damaged tissue was not as evident. Image I shows one of the dyes (janus green) that produced unreliable staining, with patchy uptake in comparison to the more even staining seen with rose bengal. BarsZ100 mm (A–F and I), and 500 mm (G and H).
450
T.L. Jackson et al. / Experimental Eye Research 81 (2005) 446–454
Table 1 Table shows the minimum dye concentration that produced retinal staining that could be detected by two masked observers in five out of five attempts Chromophore
Minimum visible concentration, MVC (% weight/volume)
Alcian blue Diethyloxadicarbocyanine Evan’s blue Fast green Fluorescein Janus green Methylene blue Naphthol green Neutral red Reactive yellow Rose bengal Trypan blue
Observer 1
Observer 2
2 0.0024 0.02 2 0.2 0.04 0.2 1 0.002 1 0.01 0.1
2 0.0024 0.02 2 0.2 0.04 0.2 0.1 0.002 0.1 0.01 0.01
even, reliable staining at low concentrations, but appeared to alter the retinal structure, with tissue becoming less compliant to handling. As observed previously with trypan blue (Jackson and Marshall, 2004), both janus green and rose bengal produced some selective staining of the cut outer edge of the retinal flat-mount. This appeared to constitute increased uptake in the cells that were traversed or damaged by the cut used to isolate the flat-mount from whole retina. 3.2. Image analysis
particularly with higher concentrations (Fig. 1I). Fast green produced homogenous staining in some retinal specimens at concentrations as low as 0.05%, but in others up to 2% was required. The cause of this variability was not identified. Fluorescein stained the retina reliably at concentrations of 0.2%, but the yellow hue it imparted produced relatively little colour contrast with unstained retina. Similar findings were made with procian yellow. Methylene blue produced an
The results of image analysis are shown in Fig. 2. For nine of the 12 dyes, the calculated colour differences increase monotonically with dye concentration. The exceptions were Evan’s blue, fluorescein and rose bengal, for which the colour difference for the highest concentration was less than that of the second highest. From visual inspection of the graphs a threshold of 35DE94 units was selected as best corresponding to the MVC. This is shown in Fig. 2 as the horizontal line on each graph. For most dyes this fell within 1 data point (1 log unit of dye concentration) of the MVC (shown as vertical lines). The exceptions were fluorescein, neutral red, methylene blue, and alcian blue. For these dyes the 35DE94 threshold was within 2 log units of the MVC.
DiOC
Alcian Blue
Evan's Blue
FastGreen
125
125
125
125
100
100
100
100
75
75
75
75
50
50
50
50
25
25
25
25
0.0002 0.002
0.02
0.2
2
0.00024 0.0024 0.024
Fluorescein
0.24
2.4
0.002
0.02
0.2
2
0.002
20
MethyleneBlue
JanusGreen 125
125
125
100
100
100
100
75
75
75
75
50
50
50
50
25
25
25
25
0.02
0.2
2
20
0.0004 0.004
NeutralRed
0.04
0.4
0.002
4
ProcianYellow
0.02
0.2
2
20
0.002
125
125
100
100
100
100
75
75
75
75
50
50
50
50
25
25
25
25
0.2
2
0.001
0.01
0.1
1
10
0.0001 0.001
0.01
0.1
2
20
0.2
2
20
TrypanBlue
125
0.02
0.02
RoseBengal
125
0.0002 0.002
0.2
NaphtholGreen
125
0.002
0.02
1
0.002
0.02
0.2
2
20
Fig. 2. Graphs showing the CIE94 colour differences (y-axis) between stained and unstained retina for varying dye concentrations (x-axis). The vertical error bars show G1 S.D. and were calculated using the data from different retinal samples (nR3). The vertical dashed lines show the minimum visible concentration that produced reliable staining of the retinal surface, judged by two blinded observers; where only one line is shown the observers agreed. The horizontal dashed line at 35DE94 units is the calculated colour difference threshold that best agrees with the observers’ judgements.
T.L. Jackson et al. / Experimental Eye Research 81 (2005) 446–454
451
Fig. 3. Graph shows an MTT assay of human retinal pigment epithelium and Mu¨ller cells exposed to various dyes. There was no significant reduction in viability relative to the negative control (saline) in any group except Mu¨ller cells exposed to rose bengal. These showed a statistically significant reduction in viability (p!0.001). Error barsG1 S.D.
3.3. Safety testing The viability of glial cells exposed to rose bengal was significantly reduced to 48% of the negative control (p!0.001; tZ4.71; CI of difference 30–75%) when tested using the MTT assay (Fig. 3). The live–dead probe also showed cell damage in this group, with large numbers of dead glia (Fig. 4). The toxicity data for trypan blue has been presented previously: no RPE or glial damage was observed with concentrations of up to 0.2% (Jackson et al., 2004a). The viability of glial cells exposed to Evan’s blue was slightly reduced to 87% of the negative control, but this difference was not significant (pZ0.2815; tZ1.09; CI K11 to 38%), and no consistent damage was observed with the live–dead probe. The other groups did not show toxicity with either the MTT assay or live–dead probe.
Black (1947) and Kutschera (1969) attempted similar experiments with local rather than systemic administration of dyes. Since that time there has been relatively little interest in the subject until the recent, widespread introduction of ICG as a macular vital stain, and more recent reports advocating infracyanine green and trypan blue. These agents have been introduced without a published, systematic appraisal of why they are the most suitable dyes. Rather, the dyes that are most readily available in the laboratory and clinic have been adapted for macular use, often before extensive experimental safety studies have been completed. This has probably been possible because these vital stains have been employed as surgical devices, avoiding the more rigorous safety testing required for pharmacotherapeutic agents. This approach is
3.4. Frozen section Frozen sections of retina exposed to naphthol green, Evan’s blue, and neutral red all showed staining of the ILM, but diffusion into the retinal substrate varied (Fig. 5). Overall, Evan’s blue produced the most selective staining of the ILM. Despite producing easily visible macroscopic retinal staining, methylene blue showed much less distinct staining in section, with light uptake in all retinal layers in some sections—others appeared similar to the unstained control. Trypan blue showed only faint staining on frozen section and was often difficult to distinguish from unstained retina.
4. Discussion Sorsby (Sorsby et al., 1937; Sorsby, 1938, 1939a, 1939b) described in vivo retinal staining more than 65 years ago, and
Fig. 4. Fluorescence photomicrograph of Mu¨ller cells exposed to rose bengal 0.02%, then stained with the live–dead probe one day later. The cytoplasm of live cells are stained green with calcein, and the nuclear matter of dead or dying cells are stained red with ethidium homodimer-1. Dead cells can be seen scattered throughout the image, suggesting a toxic effect that was not seen in the negative control.
452
T.L. Jackson et al. / Experimental Eye Research 81 (2005) 446–454
Fig. 5. Photomicrograph of bovine retinal frozen sections exposed to methylene blue, neutral red, Evan’s blue, and naphthol green (left to right). The methylene blue stained section reveals a very faint blue stain in all retinal layers, but the appearance was only marginally different to the unstained control. Trypan blue showed even less definite staining on frozen section and could not reliably be distinguished from unstained retina (not shown). Neutral red showed staining of the internal limiting membrane (ILM), inner nerve fibre layer, and possibly inner half of the retina. Evan’s blue staining was predominantly restricted to the ILM. Naphthol green showed preferential binding in the ILM and nerve fibre layer, but some green staining was visible throughout all retinal layers. Scale: retinal thickness measures approximately 150 mm.
not ideal, and the present uncertainty (Kampik and Sternberg, 2003; Sebag, 2004; Jackson, 2005) regarding the safety of ICG highlights this fact. It is likely that ICG, infracyanine green, and trypan blue represent only a small proportion of the agents that might be suitable for vital staining. In general, biological stains have specific atomic groupings (CaS, CaN, NaN, NaO and NO2) that are known to impart colour (Lillie, 1977). A number of these have been used as vital stains in other clinical specialties such as dentistry (van de Rijke, 1991) and urology (Gill and Strauss, 1984). Many hundreds of chromophores are used by the food and textile industries, and in medical and scientific laboratories to stain tissue for light microscopy. This study tested only a small proportion of available agents. It used a series of dyes to validate a systematic screening method that could be applied to a wide range of dyes. This in turn may help to identify a vital stain that is less toxic than ICG, and more effective than trypan blue. This has the potential to improve the ease and safety of surgical ILM removal; a procedure that is thought to contribute to the success of macular hole surgery. Twelve dyes were selected for testing: alcian blue; diethyloxadicarbocyanine; Evan’s blue; fast green; fluorescein; janus green; methylene blue; naphthol green; neutral
red; procian yellow (reactive yellow); rose bengal; and trypan blue. Of these, five showed favourable staining characteristics: Evan’s blue; naphthol green; neutral red; rose bengal; and trypan blue. One of these, trypan blue, has recently been marketed as a macular vital stain and was included for comparison (Feron et al., 2002; Stalmans et al., 2003; Rodrigues, 2003; Perrier and Sebag, 2003a, 2003b). All five agents showed an ability to rapidly stain the retina and were relatively easy to rinse free, unlike agents such as infracyanine green (Jackson et al., 2004b). Safety testing of these five agents did not show any toxicity except in glial cells exposed to rose bengal. These showed significant damage using a mitochondrial enzyme assay and a live–dead probe that identifies cells with compromised cell membranes. Although cell culture cannot be used to conclude on the safety of these agents as a surgical tool, it is a useful preliminary test to rule out agents with clear toxicity. The findings suggest rose bengal may not be appropriate for retinal use, but with respect to the other dyes it is only possible to conclude that further safety testing would be worthwhile. Use of these agents in humans would not yet be appropriate, with the exception of trypan blue. Establishing the safety of any candidate dye would ideally involve experiments in an animal with rods and cones, a duplex retinal circulation, and a globe size adequate to perform vitrectomy. Exposure routines would involve focal delivery and removal of the dye after a brief interval, and adequate follow up to detect delayed toxicity. Although image analysis was useful in determining the relationship of dye concentration to degree of retinal staining, it had some limitations. For three agents, Evan’s blue, fluorescein, and rose bengal, the highest dye concentration produced a lower colour difference value than the second highest concentration (Fig. 4). Inspection of a representative image (Fig. 1) shows that high concentrations produced dark retinal staining, which necessarily means that samples were less saturated with the hue of the dye. Colour difference equations such as CIE94 attempt to correctly weight differences in hue, saturation, and lightness. It is not known whether the use of CIE94 produces valid weighting at these high concentrations. This does not, however, limit the applicability of this technique, as the finding only occurred at concentrations well above those that might be used clinically. It should be noted that although the image analysis provides a relatively simple means providing quantitative data, it would not be able to replace the MVC as a screening method. Further, image analysis considered the effect of staining isolated retina that had both inner and outer surfaces exposed to dye. Image analysis could therefore exclude dyes that failed to effectively bind retinal tissue, but it could not identify those that selectively stained the ILM. To determine which dyes selectively stain the ILM, frozen sections were viewed on a light microscope (Fig. 5). Retina exposed to dye on only the inner retinal surface showed
T.L. Jackson et al. / Experimental Eye Research 81 (2005) 446–454
different staining patterns, depending on which dye was used. Agents such as methylene blue produced easily visible macroscopic staining with concentrations of 0.2%, but staining was dispersed throughout the retinal layers. Hence, 10–15 mm frozen sections showed only faint staining. By contrast, the concentration required to produce macroscopic staining with Evan’s blue (0.02%) produced definite staining on frozen section, as dye uptake was predominantly at the level of the ILM. This selectivity may be advantageous in a vital stain designed to delineate peeled ILM. This study only considered agents that would be employed as chromophores, but it would be interesting to investigate certain fluorophores. This is a potentially useful route of enquiry as many of the fluorophores used histologically have extremely high levels of tissue specificity. Relatively simple optical modification of the operating microscope means that they could be selectively viewed during surgery, as barrier filters are already employed to protect surgeons from macular injury during endoscopic laser application. We have previously shown that it is possible to selectively identify glial cells with damaged cell membranes during surgery, using a fluorophore tagged antibody to the intracellular filament, glial fibrillary acidic protein (GFAP) (Jackson and Marshall, 2004). It is possible that removal of ILM would also result in selective staining with this agent, as the ILM is formed by the basement membrane of the Mu¨ller cell foot-plates. There are many other fluorophores that might also be useful as vital stains. Unlike the introduction of present-day macular vital stains, it would be helpful to have a systematic approach to test potentially useful agents, with preliminary ex vivo safety testing prior to animal, then human studies. This study attempted to validate a screening system by testing it on a selection of chromophores. There are, however, many more chromophores available, and considerable scope for further research.
Acknowledgements Supported by Deutsche Forschungsgemeinschaft (DFG) grant Hi 758/1-1, The Allerton Fund, and Special Trustees of Guys and St Thomas’ Hospital. The authors thank G. Astrid Limb (Institute of Ophthalmology, London, UK) for providing the Mu¨ller cells used in this study, and Paul Constable and Bruce Knight for assistance with RPE cell culture.
References Black, G.W., 1947. Some aspects of the treatment of simple detachment of the retina, including vital staining of the retina by methylene blue. Trans. Ophthalmol. Soc. UK 67, 313–322. Enaida, H., Sakamoto, T., Hisatomi, T., Goto, Y., Ishibashi, T., 2002. Morphological and functional damage of the retina caused by intravitreous indocyanine green in rat eyes. Graefes Arch. Clin. Exp. Ophthalmol. 240, 209–213.
453
Engelbrecht, N.E., Freeman, J., Sternberg Jr., P., Aaberg Sr., T.M., Aaberg Jr., T.M., Martin, D.F., Sippy, B.D., 2002. Retinal pigment epithelial changes after macular hole surgery with indocyanine green-assisted internal limiting membrane peeling. Am. J. Ophthalmol. 133, 94. Feron, E.J., Veckeneer, M., Parys-Van Ginderdeuren, R., Van Lommel, A., Melles, G.R.J., Stalmans, P., 2002. Trypan blue staining of epiretinal membranes in proliferative vitreoretinopathy. Arch. Ophthalmol. 120, 141–144. Funahashi, T., Kondo, K., Kimura, M., 1980. The vital staining of the eye. Acta Soc. Ophthalmol. Jpn. 84, 1826–1880. Gandorfer, A., Haritoglou, C., Gass, C.A., Ulbig, M.W., Kampik, A., 2001. Indocyanine green-assisted peeling of the internal limiting membrane may cause retinal damage. Am. J. Ophthalmol. 132, 431–433. Gill, W.B., Strauss, F.H., 1984. In vivo mapping of bladder cancer (chromocystoscopy for in vivo detection of neoplastic urothelial surfaces). Urology 23, 63–66. Haritoglou, C., Gandorfer, A., Gass, C.A., Schaumberger, M., Ulbig, M.W., Kampik, A., 2002. Indocyanine green-assisted peeling of the internal limiting membrane in macular hole surgery affects visual outcome: a clinicopathologic correlation. Am. J. Ophthalmol. 134, 836–841. Haritoglou, C., Gandorfer, A., Gass, C.A., Kampik, A., 2004. Histology of the vitreoretinal interface after staining of the internal limiting membrane using glucose 5% diluted indocyanine green and infracyanine green. Am. J. Ophthalmol. 137, 345–348. Ho, J.D., Chen, H.-C., Chen, S.-N., Tsai, R.J.-F., 2003. Cytotoxicity of indocyanine green on retinal pigment epithelium: implications for macular hole surgery. Arch. Ophthalmol. 121, 1423–1429. International Telecommunications Union, 1990. BT. 709. Basic parameter values for the HDTV standard for the studio and for international programme exchange. International Telecommunications Union, Geneva. Iriyama, A., Uchida, S., Yanagi, Y., Tamaki, Y., Inoue, Y., Matsuura, K., Kadonosono, K., Araie, M., 2004. Effects of indocyanine green on retinal ganglion cells. Invest. Ophthalmol. Vis. Sci. 45, 943–947. Jackson, T.L., 2003. An investigation of novel techniques to enhance the detection of retinal breaks in detachment surgery. University of London, PhD Thesis. pp. 191–197. Jackson, T.L., 2005. Indocyanine green accused: the case for and against ICG-assisted macular surgery. Br. J. Ophthalmol 89, 395–396. Jackson, T.L., Marshall, J., 2004. Fluorophore assisted retinal break detection using antibodies to glial fibrillary acidic protein. Invest. Ophthalmol. Vis. Sci. 45, 993–1001. Jackson, T.L., Hillenkamp, J., Knight, B.C., Zhang, J.J., Thomas, D., Stanford, M.R., Marshall, J., 2004a. Safety testing of indocyanine green and trypan blue using retinal pigment epithelium and glial cell cultures. Invest. Ophthalmol. Vis. Sci. 45, 2778–2785. Jackson, T.L., Vote, B., Knight, B.C., El-Amir, A., Standford, M.R., Marshall, J., 2004b. Safety testing of infracyanine green using retinal pigment epithelium and glial cell cultures. Invest. Ophthalmol. Vis. Sci. 45, 3697–3703. Kampik, A., Sternberg, P., 2003. Indocyanine green in vitreomacular surgery (why) is it a problem?. Am. J. Ophthalmol. 136, 527–529. Kutschera, E., 1969. [Vital staining of the detached retina with retinal breaks]. Albrecht Von Graefes Archiv fur Klinische und Experimentelle Ophthalmologie 178, 72–87. Kwok, A.K., Lai, T.Y., 2003. Internal limiting membrane removal in macular hole surgery for severely myopic eyes: a case–control study. Br. J. Ophthalmol. 87, 885–889. Kwok, A.K., Lai, T.Y., Man-Chan, W., Woo, D.C., 2003. Indocyanine green assisted retinal internal limiting membrane removal in stage 3 or 4 macular hole surgery. Br. J. Ophthalmol. 87, 71–74. Li, K., Wong, D., Hiscott, P., Stanga, P., Groenewald, C., McGalliard, J., 2003. Trypan blue staining of internal limiting membrane and epiretinal membrane during vitrectomy: visual results and histopathological findings. Br. J. Ophthalmol. 87, 216–219.
454
T.L. Jackson et al. / Experimental Eye Research 81 (2005) 446–454
Lillie, R.D., 1977. H.J. Conn’s Biological Stains, ninth ed. Williams & Wilkins, Baltimore. Limb, G.A., Salt, T.E., Munroe, P.M.G., Moss, S.E., Khaw, P.T., 2002. In vitro characterization of a spontaneously immortalized human Mu¨ller cell line. Invest. Ophthalmol. Vis. Sci. 43, 864–869. Maia, M., Pieramici, D.J., Margalit, E., Farah, M.E., Lakhanpal, R.R., Au Eong, K.-G., 2004. Retinal pigment epithelial abnormalities after internal limiting membrane peeling guided by indocyanine green staining. Retina 24, 157–160. McDonald, R., Smith, K.J., 1995. CIE94—a new colour-difference formula. J. Soc. Dyers Colourists 111, 376–379. Melles, G.R.J., de Waard, P.W.T., Pameyer, J.H., Beekhuis, W.H., 1999. Trypan blue capsule staining to visualize the capsulorhexis in cataract surgery. J. Cataract Refract. Surg. 25, 7–9. Perrier, M., Sebag, M., 2003a. Epiretinal membrane surgery assisted by trypan blue. Am. J. Ophthalmol. 135, 909–911. Perrier, M., Sebag, M., 2003b. Trypan blue-assisted peeling of the internal limiting membrane during macular hole surgery. Am. J. Ophthalmol. 135, 903–905. Rodrigues, E.B., 2003. Trypan blue stains the epiretinal membrane but not the internal limiting membrane. Br. J. Ophthalmol. 87, 1431–1432. Sebag, J., 2004. Indocyanine green-assisted macular hole surgery: too pioneering?. Am. J. Ophthalmol. 4, 744–746. Sippy, B.D., Engelbrecht, N.E., Hubbard, G.B., Moriarty, S.E., Jiang, S., Aaberg Jr., T.M., Aaberg Sr., T.M., Grossniklaus, H.E., Sternberg Jr., P., 2001. Indocyanine green effect on cultured human retinal pigment epithelial cells: implications for macular hole surgery. Am. J. Ophthalmol. 132, 433–435. Slaughter, K., Lee, I.L., 2004. Macular hole surgery with and without indocyanine green assistance. Eye 18, 376–378. Sorsby, A., 1938. Two patients with vital staining of the fundi. Trans. Ophthal. Soc. UK 58, 275. Sorsby, A., 1939a. Vital staining of the fundus. Trans. Ophthal. Soc. UK 59, 727–730. Sorsby, A., 1939b. Vital staining of the retina: preliminary clinical note. Br. J. Ophthalmol. 23, 20–24. Sorsby, A., Elkeles, A., Goodhart, G.W., Morris, I.B., 1937. Experimental staining of the retina in life. Proc. R. Soc. Med. 30, 1271–1273.
Sorsby, A., Wright, A.D., Elkeles, A., 1942. Vital staining in brain surgery. A preliminary note. Proc. R. Soc. Med. 36, 137–144. Stalmans, P., Van Aken, E.H., Veckeneer, M., Feron, E.J., Stalmans, I., 2002. Toxic effect of indocyanine green on retinal pigment epithelium related to osmotic effects of the solvent. Am. J. Ophthalmol. 134, 282–285. Stalmans, P., Feron, E.J., Parys-Van Ginderdeuren, R., Van Lommel, A., Melles, G.R., Veckeneer, M., 2003. Double vital staining using trypan blue and infracyanine green in macular pucker surgery. Br. J. Ophthalmol. 87, 713–716. Uemura, A., Kanda, S., Sakamoto, Y., Kita, H., 2003. Visual field defects after uneventful vitrectomy for epiretinal membrane with indocyanine green-assisted internal limiting membrane peeling. Am. J. Ophthalmol. 136, 252–257. van de Rijke, J.W., 1991. Use of dyes in cariology. Int. Dent. J. 41, 111–116. Veckeneer, M., van Overdam, K., Monzer, J., Kobuch, K., van Marle, W., Spekreijse, H., van Meurs, J., 2001. Ocular toxicity study of trypan blue injected into the vitreous cavity of rabbit eyes. Graefes Archiv. Clin. Exp. Ophthalmol. 239, 698–704. Von Kries, J., 1970. Chromatic adaptation. In: Sources of Colour Science. MIT Press, Cambridge, MA. Weinberger, A.W.A., Kirchhof, B., Mazinani, B.E., Schrage, N.F., 2001. Persistent indocyanine green (ICG) fluorescence 6 weeks after intraocular ICG administration for macular hole surgery. Graefes Arch. Clin. Exp. Ophthalmol. 239, 388–390. Weinberger, A.W., Schlossmacher, B., Dahlke, C., Hermel, M., Kirchhof, B., Schrage, N.F., 2002. Indocyanine-green-assisted internal limiting membrane peeling in macular hole surgery—a follow-up study. Graefes Archiv. Clin. Exp. Ophthalmol. 240, 913–917. Whittle, R.M., Wallace, G.R., Whiston, R.A., Dumonde, D.C., Stanford, M.R., 1998. Human antiretinal antibodies in toxoplasma retinochoroiditis. Br. J. Ophthalmol. 82, 1017–1021. Yam, H.F., Kwok, A.K., Chan, K.P., Lai, T.Y., Chu, K.Y., Lam, D.S., Pang, C.P., 2003. Effect of indocyanine green and illumination on gene expression in human retinal pigment epithelial cells. Invest. Ophthalmol. Vis. Sci. 44, 370–377.