Visual Field Defects After Intravitreous Administration of IndocyanineGreen in Macular Hole Surgery

Objectives  To report the findings on a patient cohort with visual field defectsafter macular hole surgery with indocyanine green (ICG)–assisted internallimiting membrane peeling and to investigate the correlation between the defectsand the use of ICG.

Design  Retrospective, noncomparative interventional case series.

Participants  Thirty-nine eyes of 38 patients having the clinical diagnosis of a macularhole who underwent pars plana vitrectomy between January 1, 2001, and December31, 2002, were enrolled in this study.

Intervention  Indocyanine green–assisted internal limiting membrane peelingwas performed on a series of 22 eyes: 12 eyes using a 0.5% ICG solution and3-minute exposure to the retina (group 1), 4 eyes using a 0.5% ICG solutionand immediate washout (group 2), and 6 eyes using a 0.25% ICG solution andimmediate washout (group 3). The remaining 17 eyes underwent vitrectomy withoutICG-assisted internal limiting membrane peeling (group 4).

Main Outcome Measures  Visual field, best-corrected visual acuity, and fundus photography wereevaluated.

Results  Postoperatively, all patients (100%) in group 1 and 1 (25%) of 4 eyesin group 2 had visual field defects. None of the patients in group 3 had avisual field defect. The visual field defects included 10 eyes (84%) withnasal defects, 1 eye (8%) with an inferotemporal defect, and 1 eye (8%) withan extensive visual field defect. Ophthalmoscopy revealed mild to moderateoptic disc pallor in 8 (62%) of 13 eyes with postoperative visual field defects.Only 1 patient in group 4 had an inferotemporal defect; none of the otherpatients in group 4 had visual field defects. There was no statistically significantdifference in postoperative visual acuity between patients with and withoutpostoperative visual field defects.

Conclusions  Although this study was limited by the few patients enrolled, our experienceindicates that visual field defects, specifically nasal defects, can occurafter macular hole surgery with ICG-assisted internal limiting membrane peeling,and that the incidence depends on the concentration of the ICG solution and/orthe exposure time to the retina. Further studies are needed to clarify thepathomechanism of visual field defects.

Although indocyanine green (ICG) is a helpful tool in peeling the internallimiting membrane (ILM) in macular surgery, the potential toxic effect ofICG has been recently reported.^1,2 Previousreports have shown the possible toxic effect of ICG on cultured human retinalpigment epithelial cells.^3,4 Inexperimental studies, morphologic and functional damage to the retina wasobserved using an animal model⁵ or human donoreyes.⁶ Furthermore, several authors have reportedretinal pigment epithelium changes after macular surgery using ICG to facilitatepeeling of the ILM.^7,8 More recently,visual field defects following vitrectomy for macular hole or pucker withICG-assisted ILM peeling have been observed.^9,10 Wealso reported on the development of postoperative visual field defects in4 patients who underwent vitrectomy for epiretinal membranes using ICG-assistedILM peeling.¹¹ Although the application ofICG is thought to be a possible cause of the defects in these reports, theassociation between the use of ICG and postoperative visual field defectsis not fully understood. The dual purpose of this article is to describe patientswith visual field defects after macular hole surgery using ICG-assisted ILMpeeling and to investigate the relationship between the use of ICG and theoccurrence of the defect.

This study was an interventional case series in which we performed anoncomparative retrospective review of a consecutive series of patients whohad undergone macular hole surgery with or without ICG-assisted ILM peeling.We retrospectively reviewed the medical records of 44 consecutive patients(45 eyes) who had undergone vitrectomy for idiopathic macular holes betweenJanuary 1, 2001, and December 31, 2002, at Kagoshima City Hospital, Kagoshima,Japan. Six patients were excluded for the following reasons: a history ofglaucoma (2 patients), failure to close the macular hole with a single surgicalprocedure (1 patient), a history of retinitis pigmentosa (1 patient), postoperativedense vitreous hemorrhage (1 patient), or less than 6 months' postsurgicalfollow-up (1 patient). Therefore, a total of 39 eyes of 38 patients (10 malesand 28 females; mean age, 63.5 years) were included in this study. The stageof macular holes included 16 in stage 2 (partial-thickness holes), 14 in stage3 (full-thickness holes), and 9 in stage 4 (advanced full-thickness holeswith vitreous separation from the optic disc and macula). None of these patientshad a postoperative retinal detachment. There was no recorded incidence ofincreased intraocular pressure above 30 mm Hg; there were no sustained increasesin intraocular pressure. This study was carried out at a site that has noinstitutional review board.

Surgery was performed by a single surgeon (A.U.). Informed consent wasobtained prior to surgical intervention in all patients. A standard 3-portpars plana vitrectomy was performed in all patients using retrobulbar anesthesia.The infusion cannula was placed in the inferotemporal quadrant. The crystallinelens was removed by phacoemulsification followed by intraocular lens implantationbefore pars plana vitrectomy in phakic eyes. After core vitrectomy, the posteriorvitreous was separated from the retina, if still attached. Posterior hyaloidseparation was achieved by active suction with the vitrectomy probe. In 22eyes (2 of 16 stage 2 holes, 12 of 14 stage 3 holes, and 8 of 9 stage 4 holes),ILM membranorhexis was performed using a microvitreoretinal blade and Tanoforceps (Synergetics Inc, St Charles, Mo) after staining the ILM with ICGas described below. During a fluid-air exchange, the air pressure was setat 40 mm Hg. An air-gas exchange was then performed using either 20% sulfurhexafluoride (SF₆) or 14% perfluoropropane (C₃F₈), and the patient was asked to maintain a face-down position for atleast 7 days.

The ICG solution was prepared as follows: 25 mg of ICG dye was reconstitutedwith 1.0 mL of sterile water. After the ICG was completely dissolved in thesterile water, 4.0 or 9.0 mL of balanced salt solution was added to attaina final ICG concentration of 0.5% or 0.25%. A total of 0.1 to 0.8 mL of ICGsolution was then injected into the fluid-filled eye using a 27-gauge bluntneedle. After removal of the ICG dye, the green-stained ILM was cut with amicrovitreoretinal blade. With Tano forceps, the ILM was lifted and peeledoff in a circular fashion. As a result, the ILM was removed from a 2 to 3disc diameter.

Twenty-two patients who had undergone ICG-assisted ILM peeling weredivided into 3 groups according to the concentration of the ICG solution andthe ICG staining time. In 12 patients (group 1), 0.6 to 0.8 mL of 0.5% ICGsolution was injected into the vitreous cavity. The ICG dye was left in thevitreous cavity for 3 minutes with scleral plugs in place and then washedout. In 4 patients (group 2), less than 0.2 mL of 0.5% ICG solution was injectedand removed immediately with active suction. In 6 patients (group 3), lessthan 0.2 mL of 0.25% ICG solution was used and it was aspirated immediately.

A Humphrey static perimetry test was performed preoperatively in allpatients. Humphrey static perimetry and Goldmann perimetry were performedbetween 3 and 6 months after surgery. Visual field, best-corrected visualacuity, and fundus photographs were evaluated. For statistical analysis, best-correctedvisual acuity measured at the 1-year follow-up visit was considered. Standardpostoperative examinations were performed at 1 week, 2 weeks, 1 month, 2 months,3 months, and every 3 months thereafter. The mean follow-up period was 17.4months (range, 12-31 months).

Visual acuity data were converted to the logarithm of the minimal angleof resolution (logMAR) and analyzed using Stat View version 4.5 (Abacus ConceptsInc, Berkeley, Calif). The t test was used to comparepreoperative and postoperative visual acuities.

All patients (100%) in group 1 had visual field defects. The defectsincluded 10 eyes (84%) with nasal defects, 1 (8%) with an inferotemporal defect,and 1 (8%) with an extensive visual field defect (Figure 1). In group 2, 1 patient (25%) had a nasal defect. In group3, there were no postoperative visual field defects. Nasal or extensive defectswere noted on both types of perimetry and the visual field changes were similarwhen present by both methods of testing. There was a statistically significantdifference in the presence of postoperative visual field defects between groups1 and 2 (P = .007, Fisher exact test) and betweengroups 1 and 3 (P<.001).

Ophthalmoscopy revealed optic disc pallor in 8 (62%) of 13 eyes withvisual field defects (Figure 2).Fundus examination revealed no detectable changes corresponding to the visualfield defects. Fluorescein angiography, which was performed in 3 patientswith visual field defects, showed no abnormalities such as retinal vesseldamage or alteration in the retinal pigment epithelium. Only 1 patient hadan inferotemporal defect; all of the remaining eyes that underwent macularhole surgery without ICG-assisted ILM peeling no had visual field defectspostoperatively.

The mean preoperative visual acuity (Snellen equivalent) was 20/122for group 1, 20/106 for group 2, and 20/100 for group 3. The mean postoperativevisual acuity at 1-year follow-up was 20/36 for group 1, 20/30 for group 2,and 20/27 for group 3. Twenty-one of 22 eyes in which ICG was used had postoperativevisual acuity of 20/40 or better (Table1). Mean visual acuity was significantly improved after surgeryin each group (P<.001). There was no significantdifference in postoperative visual acuity between patients with and withoutvisual field defects (Figure 3).

Our study showed that all patients in group 1 had visual field defectspostoperatively. Most of the defects occurred in the nasal field. Only 1 patientin group 2 had a nasal visual field defect. None of the patients in group3 had postoperative visual field defects. Additionally, visual field defectswere noted in only 1 of 17 patients who underwent macular hole surgery withoutICG-assisted ILM peeling. These data suggest that the occurrence of visualfield defects is strongly related to the use of ICG and that the incidencedepends on the concentration of ICG solution or tissue contact time of ICGor both, as proposed by Gandorfer et al.¹²

Visual field defects after vitrectomy for macular hole or pucker withthe use of ICG have been reported. In 1 series by Haritoglou et al,⁹ seven of 20 patients who had undergone macular holesurgery using ICG-assisted ILM peeling had visual field defects postoperatively.They injected 0.2 to 0.5 mL of 0.05% ICG solution to stain the ILM and thenwashed this out after about 1 minute. They also reported postoperative visualfield defects in 7 of 20 patients who had undergone macular pucker surgerywith the use of less than 0.5 mL of 0.05% ICG solution.¹⁰ Inour series, postoperative visual fields were normal in patients in whom lessthan 0.2 mL of a 0.25% ICG solution was used with immediate washout. Thesedata suggest that exposure time to the retina may be critical in the occurrenceof visual field defects as well as the dose and concentrations of ICG.

Eleven of the 13 patients with visual field defects after ICG-assistedILM peeling had nasal defects: one had an extensive defect and the other hadan inferotemporal defect. Nasal visual field defects detected in our seriesare similar to those reported previously.^9,10 Furthermore,1 patient had a more severe type of visual field defect that was describedin our previous article.¹¹ However, the reasonwhy the nasal field is predominantly affected remains unknown.

An inferotemporal visual field defect was detected in 1 of the group1 patients. From the shape of the defect, we speculated that it was causedby a dehydration injury of the nerve fiber layer during fluid-air exchange,^13-15 although we cannoteliminate the possibility that it was caused by the use of ICG. Given thefact that a similar defect was seen in a patient in whom ICG had not beenused, it is reasonable to suggest that the temporal defect is not correlatedwith the use of ICG.

Fundus examination of 8 of 14 patients with postoperative visual fielddefects showed a suggestion of diffuse optic disc pallor at the final follow-up.This is the first description of optic disc pallor in patients in whom ICGhas been used. We did not see optic disc pallor in our previous report ofvisual field defects following vitrectomy for epiretinal membrane.¹¹ The degree of optic disc pallor was different ineach eye: from slight pallor to complete optic disc atrophy. This findingwas evident only in eyes with postoperative visual field defects; therefore,optic disc pallor appears to be strongly related to both visual field defectsand the use of ICG. No other abnormalities, such as mottling of retinal pigmentepithelium or retinochoroidal atrophy seen in patients with air infusion–relatedvisual field defects, were observed in those eyes. Since fundus change maybecome apparent after years¹⁶; however, long-termfollow-up of these eyes is essential.

It is still controversial whether the use of ICG precludes good visualrecovery. Da Mata et al¹⁷ reported good visualoutcome after ICG-assisted ILM peeling. On the other hand, Kwok et al¹⁸ reported less favorable visual outcomes in a seriesof 10 patients. In our study, visual acuity improved postoperatively despitethe use of ICG. Even in patients with visual field defects, most gained 2Snellen lines or more postoperatively in contrast to the reports by Haritoglouand coworkers,^9,10 who reportedthat there was no statistically significant improvement in postoperative visualacuity after ICG-assisted ILM peeling. Although our results showed that theuse of ICG had no adverse effect on postoperative visual improvement, it islikely that it may affect visual results during long-term follow-up.

The mechanism of the induction of visual field defects by ICG is stillunclear. An experimental approach using an animal model showed that even low-doseintravitreous ICG induced functional damage of the retina without any apparentmorphologic damage.⁵ The results of previousstudies have shown the potential toxic effect of ICG to the retinal pigmentepithelium^7,8 and the accelerationof the toxic effect by the photodynamic effect of ICG.^6,19 Inour case series, no obvious changes in the retinal pigment epithelium weredetected in patients with visual field defects. Therefore, it cannot be reasonablyassumed that the toxic effect on the retinal pigment epithelium is correlatedwith the occurrence of visual field defects. Haritoglou and coworkers^9,10 found that cellular elements wereadherent to the retinal side of the ILM in eyes with ICG staining by histologicanalysis, but they concluded that the observed visual field defects couldnot be explained by the findings.

Recent studies have suggested retention of ICG staining, along withsome staining of the optic nerve after surgery.^20-25 Itis possible that the long-lasting staining of the retina and optic nerve withICG dye may have toxic effects on the retina and optic nerve, leading to visualfield defects. Therefore, we should pay attention to the progression of visualfield defects during long-term follow-up.

This study has limitations because of the few patients included in theretrospective data analysis. It is clear that the defects occurred after surgerybecause the preoperative Humphrey static perimetry test was normal in patientswith postoperative visual field defects. We use less than 0.2 mL of 0.05%ICG solution in only selected patients with long-standing macular holes andwash it out immediately after injection. As a result, we have not encounterednew patients with visual field defects. It is possible, however, that potentialtoxic damage to the retina may occur even if we use low concentrations ofICG and remove it immediately. Therefore, we may have to reconsider the useof ICG for macular surgery, and patients who have undergone vitrectomy withICG should have careful long-term follow-up even if they have no complicationsduring the early postoperative period.

We encountered 13 patients with postoperative visual field defects afterthe use of ICG in macular hole surgery. The incidence is strongly relatedto the concentrations of ICG solution or tissue contact time or both. Althoughlowering the concentration or immediate aspiration of ICG after injectionmay reduce the incidence of visual field defects, one should pay attentionto the potential toxic effect of ICG even if the dose is minimal. Furtherstudies such as electron microscopic studies in experimental settings in animalsor electrophysiologic testing in clinical settings will be needed to betterunderstand the implications of this study. Additionally, longer follow-upof those patients may offer additional information.

Correspondence: Akinori Uemura, MD, Department of Ophthalmology,Kagoshima City Hospital, 20-17 Kajiya-cho, Kagoshima-shi 892-8580, Japan (akiu@ml.kch.kagoshima.kagoshima.jp).

Submitted for publication November 20, 2003; final revision receivedApril 12, 2004; accepted May 7, 2004.