Effect on Visual Outcome After Macular Hole Surgery When Staining theInternal Limiting Membrane With Indocyanine Green Dye

Objectives  To determine the effect on the visual outcome after macular hole surgerywhen staining the internal limiting membrane (ILM) with indocyanine green(ICG) dye and to study the mechanism of the adverse effects.

Patients and Methods  We studied 40 eyes of 38 patients with an idiopathic macular hole (size,<0.5 disc diameter; duration, <12 months). The concentration, exposuretime, and amount of the ICG solution that was minimally required to make theILM visible were determined. The patients were randomly divided into group1 (20 eyes of 19 patients) who underwent ILM peeling without ICG staining,and group 2 (20 eyes of 19 patients) who underwent ILM peeling with ICG staining.Routine examinations were conducted during the 12-month follow-up period.Multifocal electroretinogram, optical coherence tomography, and fluoresceinangiography were performed on 31 eyes of 30 patients.

Results  The macular hole was closed in all patients. Visual acuity was improvedin both groups, but it was significantly better in group 1 (median, 0.85)than in group 2 (median, 0.60; P = .02) after 12months. The improvement of visual acuity in group 1 (logarithm of the minimumangle of resolution [logMAR] units [SD], 0.82 [0.19]) was significantly betterthan that in group 2 (logMAR units, 0.67 [0.21]; P =.30). The multifocal electroretinogram and optical coherence tomographic findingswere not significantly different in the 2 groups. Fluorescein angiogram showedonly weak hyperfluorescence at the macula in some patients of both groups.

Conclusions  The results suggest that ICG staining should not be used as long asthe visibility of the retinal surface is good. However, ICG staining may beacceptable at a low concentration when a clear view of the retinal surfaceis unattainable. The results of the multifocal electroretinogram, opticalcoherence tomography, and fluorescein angiography suggest that the differencesin visual recovery were caused not only by pigment epithelial cell damageor retinal toxic effect but also probably by the effect of ICG staining onganglion cells and their axons.

In 1998, we described a new surgical technique of staining the capsuleof the crystalline lens with indocyanine green (ICG) dye during mature cataractsurgery (ICG with continuous circular capsulorhexis [ICG-CCC]).¹ Later,ICG staining was applied to the internal limiting membrane (ILM) to make itmore visible during macular hole surgery (ICG-ILMrhexis).^2,3 Althoughour report on ICG-CCC was a randomized control study, to our knowledge, ICG-ILMrhexishas not been studied in such a manner. In addition, the concentration of ICGsolution in ICG-CCC was determined from the results of animal experiments;however, to our knowledge, the optimal concentration for human ICG-ILMrhexishas not been determined.

Recently, we reported that the ICG dye remained in the eye for a longperiod after ICG-ILMrhexis.⁴ This is importantbecause there have been reports that ICG dye is a toxic substance,^5-7 and other groups havedescribed the efficacy and safety of ICG-ILMrhexis.^8-10 Furthermore,in vitro studies using human retinal pigment epithelial (RPE) cells have demonstratedthe toxic effect of ICG dye and illumination, an osmotic effect of the vehicleof ICG solution, and an ICG-induced decrease of mitochondrial enzyme activity.^11-14 Wehave performed a randomized control study to determine the effect of ICG-ILMrhexison the visual outcome after macular hole surgery and to study the mechanismof the adverse effects.

Forty eyes of 38 patients with an idiopathic macular hole and no othereye disease were studied prospectively. Eyes with an old macular hole (>12months), eyes with a large macular hole (>0.5 disc diameter), or eyes withatrophic RPE changes at the macular hole were excluded from the study. Afterobtaining informed consent, the study eyes were randomly divided before surgeryinto 2 groups using a coin-tossing method.

There were 20 eyes of 19 patients that underwent ILM peeling withoutICG staining in group 1, and 20 eyes of 19 patients that underwent ILM peelingwith ICG staining in group 2. Preoperatively, there was no significant differencein the visual acuity (range, group 1: 0.03-0.3, median, 0.125; group 2: 0.03-0.3,median, 0.15), the size and stage of the macular hole, the duration of thesymptoms, and age of the patient (mean [SD] age, group 1: 63.5 [6.9] years;group 2: 64.7 [6.9] years; Table 1).The stage of the macular hole was determined during surgery.

We use a 0.125% solution of ICG dye. The solution was made by injecting2.0 mL of sterile distilled water into a bottle of 25 mg of ICG dye (DaiichiPharmaceutical Co Ltd, Tokyo, Japan, identical to Acorn product), and thebottle was shaken until the powdered ICG was completely dissolved. Then, 1.5mL of the dissolved ICG was aspirated and 4.5 mL of an irrigating solution(BBS Plus; Alcon Inc, Forth Worth, Tex) was injected into the remaining 0.5mL of ICG in the bottle to make a final concentration of 0.125%. The osmolarityof the ICG solution in the bottle was 270 mOsm that is identical to that reportedin our original article.¹

A posterior vitreous detachment was formed with a cutter unless it wasalready present, and as much of the vitreous was removed as possible. Theinfusion of fluid was stopped, and 0.2 mL of the 0.125% ICG solution was flushedonto the surface of the retina. The ICG dye in the vitreous cavity was removedimmediately by aspiration with the cutter.

With an excellent view of the ILM, the ILM was successfully peeled inall patients. Fluid-gas exchange was performed, and the vitreous was replacedby 12% perfluoropropane. The light source of endoillumination was a 150-Whalogen lamp; no filter was used.

Before beginning this study, we determined the concentration, exposuretime, and amount of ICG solution that was minimally needed to make the ILMvisible during surgery (minimal staining). We found that 0.2 mL of a 0.125%ICG solution placed on the retina for 10 to 30 seconds was effective. The0.125% ICG solution was diluted immediately after injection by the residualvitreous; the volume in the vitreous cavity varied among the patients. Therefore,the concentration of ICG solution in the vitreous cavity was not constantand was estimated to be 0.05 to 0.06 mg/mL (0.005%-0.006% with an assumedvolume of the vitreous cavity of 4-5 mL). This concentration is much lowerthan the original report using ICG mixed with a viscomaterial.² However,the concentration of the dye in the vitreous may differ from the concentrationachieved at the vitreoretinal interface. The injected 0.125% ICG solutionwas aspirated immediately, but the exposure time of the ICG solution on theretina, the time from the start of the injection to the end of aspiration,was not constant, and was estimated to be 10 to 30 seconds from a review ofthe videotape taken during the surgery.

The visual acuity was measured preoperatively and postoperatively. Wehave reported that the preoperative visual acuity in patients with a macularhole is greatly affected by the patient's fixation and the use of a multipleletter chart can solve the problem.¹⁵ Unfortunately,the multiple letter chart was unavailable at the start of this study; therefore,we used a standard visual acuity chart (EA-117D; Takada Co Ltd, Tokyo Japan)that is widely used in Japan.

Multifocal stimulation and analyses were performed using the VERIS Science4.1 system (Electro-Diagnostic Imaging, San Mateo, Calif, and Meiyo, Aichi,Japan). Multifocal electroretinograms (mfERGs) were recorded from 16 eyesof 15 patients in group 1 and 15 eyes of 15 patients in group 2 from whominformed consent was obtained.

A stimulus array of 61 densely packed hexagons covered the central 50°.Within each frame of the cathode ray tube, each stimulus hexagon was eitherflashed on at an intensity of 2.67 candela (cd)//s per square meter or remaineddark (<0.01 cd · s · m²) and modulated at an m-sequence(215 − 1 steps).

The mfERGs were recorded with a Burian-Allen bipolar contact lens electrode,amplified (×50 000), and bandpass filtered between 10 and 300 Hz.K₁ and K₂ were extracted using the fast m-transformalgorithm, and we measured the implicit times and the amplitudes of the largestpeaks of the first-order kernel (b wave) of the central hexagon that fellon the fovea. Because the amplitudes of the mfERG are highly variable, theratio of the amplitudes, affected eye to fellow eye (A/F) ratio, was usedto compare the 2 groups.¹⁶

Cross-sectional images of the retina were obtained by optical coherencetomography (OCT) (Humphrey Instruments, San Leandro, Calif). The measurementof the foveal thickness was made manually using the longitudinal reflectivityprofile in a scan profile program of the OCT.^17,18 Forthe measurement, cursors were placed at the steepest part of each rising slopeproduced at the ILM and RPE. The retinal thickness used for analysis was theaverage of 4 measurements at the fovea. Optical coherence tomography was performedon 16 eyes of 15 patients in group 1 and 15 eyes of 15 patients in group 2.

Fluoresein angiography (FA), ophthalmoscopy, and Goldmann perimetrywere performed during the follow-up period.

Data are shown as the mean (SD). Statistical analysis software (SigmaStat;Jandel Scientific, San Rafael, Calif) was used for statistical comparisons.Comparisons for the 2 groups were performed using unpaired t test, and the Mann-Whitney test was performed for comparisons ofvisual acuity. Values of P<.05 were consideredto be statistically significant.

The median visual acuity in both groups was the same at 0.60 at 6 monthsafter surgery, but at 12 months, the median visual acuity in group 1 was significantlybetter at 0.85 than in group 2 at 0.60 (P = .02; Figure 1). The improvements of visual acuityin logMAR (logarithm of the minimum angle of resolution) units are shown in Figure 2 for both groups. At 12 months, theimprovements were significantly better in group 1 (0.82 [0.19]) than in group2 (0.67 [0.21], P = .03).

A comparison of the preoperative and postoperative visual acuities at12 months in all eyes is shown in Figure 3. All eyes had an improvement of visual acuity of more than 2 lineson the letter chart. The results of visual acuity in the 31 eyes of 30 patientswho had mfERGs, OCT, and FA tests were consistent. Although group 2 included5 more patients with a stage 4 macular hole (Table 1), their postoperative visual acuity ranged from 0.3 to 1.0(median, 0.6) and so was not the cause of the difference in postoperativevisual acuity.

There was no significant difference in the implicit time or the A/Fratio of the b-wave amplitudes preoperatively. The ratio of the amplitudesand peak time of the b wave of the mfERGs from the central hexagon were notsignificantly different for the 2 groups postoperatively (Table 2).

In all cases, OCT showed a complete closure of the macular hole, andthe thickness of the retina in the center of the fovea was not significantlydifferent in the 2 groups (mean [SD], group 1: 147.9 [29.7] µm; group2: 169.6 [58.2] µm; Table 2).

Fluorescein angiography showed only a faint hyperfluorescence in themacula in 7 of 16 eyes in group 1 and 8 of 15 eyes in group 2. Dilated fundusexamination revealed no obvious atrophy of the RPE. Goldmann perimetry revealedno scotoma or constriction of the peripheral field.

Our data showed that the visual acuity and the improvement of visualacuity in the 2 groups were not significantly different 6 months after surgery.However at 12 months, the recovery of the visual acuity was significantlybetter in eyes that were not stained with ICG than in eyes with ICG-ILMrhexis.This long-term effect is probably due to the delayed clearance of the dyeafter surgery because we have found that the dye remained in the eye for amean (SD) of 2.7 (1.4) months after macular surgery in contrast to the rapidclearance after ICG-CCC (6.0 [2.2] days).⁴ Bothgroups achieved good recovery of visual acuity. This suggests that ICG toxicitydid exist but is limited in this study.

Our results indicated that ICG staining might be acceptable at a lowconcentration in selected cases. For example, in an eye with media opacitythat prevents a clear view of the retinal surface, ICG staining may be usedto identify the ILM, but ICG staining should be avoided when the retinal surfaceis clearly visible.

The mechanism of the effect of ICG staining on the visual acuity at12 months has not been determined. The mfERGs were not significantly differentin the 2 groups suggesting that the retinal cells giving rise to the mfERGs(photoreceptors, bipolar cells, and Müller cells) were minimally affectedby ICG staining. The OCT-determined foveal thicknesses were also not significantlydifferent in the 2 groups suggesting that there was no retinal edema or atrophyat the fovea. Fluorescein angioagraphy showed only faint hyperfluorescencein some cases in both groups suggesting that RPE atrophy and retinal damagewere not obvious as described previously⁶ inboth groups.

Thus, these data suggest that the possible mechanism of the depressedvisual acuity in the eyes exposed to ICG dye is damage to the axons of theganglion cells or the nerve fibers at the optic nerve head. Recent reports^4,19,20 stated that ICG dyeremains at the optic nerve head for a long time after ICG-ILMrhexis. The findingsfrom an experimental report²¹ demonstratingICG dye in retinal ganglion cells after intravitreal injection support thispossibility. Furthermore, photodynamic effects may account for the nerve fiberlayer and ganglion cell damage.²² However,there may have been subclinical changes at the center of the fovea that couldnot be detected by mfERG, OCT, or FA. Several in vitro experiments^11-14 haveshown the toxic effect of ICG dye on the human RPE. We do not have a clinicaltest for measuring the function of RPE in the fovea. Therefore, we cannotexclude the possibility that dysfunction of foveal RPE affected visual acuity.To make a more definitive conclusion, further examinations of the visual fieldsand the RPE with longer follow-up periods are needed.

The use of ICG-ILMrhexis has been controversial. While some investigatorsreported poor results, other investigators reported good results. One of thereasons for this confusion is that the optimal concentration, exposure time,and quantity of ICG solution have not been determined. In our study, we determineda combination of the concentration, exposure time, and amount of ICG solutionfor minimal staining, which may be different among the hospitals because ofthe difference in instruments, source of the ICG dye, and color of the fundusin different races/ethnicities. Our results indicated that even a minimalstaining procedure affects the visual acuity slowly and slightly; thus, werecommend that ICG dye might be used only when the retinal surface is notclearly visible.

Visual recovery after macular hole surgery was affected by ICG-ILMrhexis,even when it was performed using a minimal staining technique. However, becausethe effect was slight, ICG staining may be acceptable at a low concentrationfor a short time in eyes with media opacity that prevents a clear view ofthe retinal surface.

Correspondence: Masayuki Horiguchi, MD, Department of Ophthalmology,Fujita Health University School of Medicine, 1-98 Toyoake, Aichi 470-1192,Japan (masayuki@fujita-hu.ac.jp).

Submitted for publication July 1, 2003; final revision received November17, 2003; accepted December 2, 2003.

Both authors contributed equally to the work and therefore should beregarded as equivalent senior authors.