Macular hole surgery has become a common treatment for idiopathic macular holes, and refinement of techniques has improved the success rate of surgery with fewer complications. The procedure typically consists of a pars plana vitrectomy, removal of the posterior hyaloid, and fluid-air or fluid-gas exchange with some period of prone positioning following surgery. A number of surgical technique variations have been used to attempt to improve anatomic success (closure of the macular hole) and functional success (improved visual acuity). Some studies have suggested that removal of the internal limiting membrane (ILM) is helpful in improving the percentage of eyes that achieve closure of the macular hole.^1–6 In contrast, one study found no improvement in macular hole surgery when the ILM was peeled.⁷ There is controversy about whether the ILM should be stained with indocyanine green (ICG) or other agents, such as trypan blue. Staining the ILM improves the ability to completely remove the ILM and usually decreases the operative time needed to remove the ILM during vitrectomy. Some studies have found that ICG-assisted removal of the ILM improves results,^8,9 whereas other have found that ICG makes no difference.^3,10–15 Some studies have found that ICG has a negative effect on the results of macular hole surgery.^16–19 A recent randomized study found no significant differences in visual acuity comparing ICG with trypan blue.²⁰ The effect of removing ILM and staining ILM with ICG was evaluated in the following study.

The results of macular hole surgery were evaluated in a retrospective, consecutive case series of 123 eyes that underwent surgery between 1994 and 2005 to examine the effect of epimacular dissection or removal of the ILM on the results of surgery. The study was performed in the author’s practice and was in conformity with the Declaration of Helsinki, US federal, and HIPAA guidelines, although no institutional review board approval was requested. The diagnosis of a macular hole was made by the surgeon and confirmed by fluorescein angiography in earlier operated eyes in the series and by optical coherence tomography in the later operated eyes in the series. All patients were treated by a single surgeon (J.T.T.) using 1 of 3 techniques for epimacular dissection. The techniques were used sequentially such that the surgeon used the same technique on all eyes until he switched to a different technique. Prior to the onset of the study, the surgeon had performed over 100 macular hole surgeries, so the results do not appear due primarily to a “learning effect” for performing macular hole surgery. The remaining aspects of the surgeries were identical.

Patients were eligible if they had a primary idiopathic or traumatic macular hole with symptoms of blurred vision of 2 years or less. Eyes with other significant macular diseases such as diabetic retinopathy or prior vitrectomy were excluded. The preoperative data collected included patient age, approximate age of the macular hole, macular hole stage (confirmed intraoperatively), preoperative lens opacities using the LOCS II grading system,²¹ visual acuity measured on a Snellen chart with current correction, and intraocular pressures. Patients were examined postoperatively at 1 day, 1 to 2 weeks, 6 weeks, 3 months, 6 months, 12 months, and variable times following 12 months. Additional postoperative visits were performed as necessary. Results of a standard ophthalmologic examination including current corrected Snellen visual acuity, status of the macular hole, cataract grade using the LOCS II system, intraocular pressures, and any postoperative complications were also recorded.

The primary outcome variables were complete macular hole closure and visual acuity at 3 months, 12 months, and the final examination following 12 months. Categorical variables were compared using chi-square tests, and numerical variables were compared using t tests or analysis of variance with post hoc testing for multiple comparisons, when appropriate.

The second technique was ILM removal, where vitreoretinal forceps were used to pinch and lift the ILM around the macular hole in 32 eyes. The goal was to remove the ILM completely surrounding the macular hole. Confirmation of ILM removal was primarily determined by the presence of a pale edema in the perimacular retina where the ILM had been successfully removed. This allowed the “edge” of the removed ILM to be identified in some eyes so further removal could be performed. It was often difficult to verify complete removal of the ILM, since the ILM was not stained.

The third technique was called the ICG technique, because ICG was used to stain the ILM and remove all of the perimacular ILM. The use of ICG to stain the ILM represents a common off-label use of ICG. The ILM was stained by performing a fluid-air exchange and placing 1 or 2 drops of 0.5 mg/mL ICG over the macular hole. The ICG was left for about 60 seconds, and then the air was replaced with fluid and the ICG was lavaged from the eye over several minutes until the vitreous fluid was no longer stained green. Removal of the ILM was started usually by grasping the ILM with flat-tipped vitreoretinal forceps and lifting the ILM until a break in the ILM developed. The edge of the ILM was grasped with the forceps and peeled circumferentially around the macular hole for a distance of at least 3 disc diameters. Occasionally, a sharp microvitreoretinal blade was used to create a small nick in the ILM to lift an edge to grasp with the forceps. Rarely would ILM removal extend beyond the temporal vascular arcades. A fluid-air exchange was then performed, and residual intravitreal fluid was again aspirated about 5 minutes after the initial fluid-air exchange to try to obtain a complete gas fill. Later in the series, eyes had preplacement of the sclerotomy sutures prior to the fluid-air exchange to allow rapid closure to try to prevent dehydration injury to the nerve fiber layer, which has been associated with postoperative visual field defects.^22–27

All eyes received a 16% perfluoropropane bubble, and about 50 mL of 16% perfluoropropane was lavaged through the eye to ensure the correct gas concentration. Patients were placed immediately prone in the operating room after dressing the eye and remained prone full-time for 14 days. Patients were instructed to sleep prone for a total of 6 weeks. The intraocular gas bubbles absorbed by 6 to 8 weeks following surgery.

The baseline characteristics of the three groups (ERD, ILM, and ICG) were similar and are given in the Table. Visual acuity was slightly worse in the ICG group, but this difference was not statistically significant. The ILM group had a significantly shorter duration of the macular hole (P = .032), and the time of the final examination was significantly greater for the ERD and ILM groups, which had surgery before the ICG group (P < .001).

TABLE BASELINE CHARACTERISTICS OF THE THREE GROUPS THAT UNDERWENT MACULAR HOLE SURGERY

The macular hole was closed at 3 months in 26 of 29 eyes (89.7%) in the ERD group, 31 of 32 eyes (96.9%) in the ILM group, and 61 of 62 eyes (98.4%) in the ICG group (Figure 1). The one open macular hole in the ICG group closed spontaneously between 3 and 6 months. There were 3 persistent macular holes in the ERD group, and 4 additional macular holes recurred following initial successful closure at various times during follow-up. A second macular hole surgery was performed in 6 of 7 ERD eyes with persistent or recurrent macular holes, and the holes were closed at the final examination in 27 of 29 eyes (93.1%). Of the remaining open holes, 1 patient refused surgery and the other eye had 2 surgeries with unsuccessful closure, possibly due to severe cystoid macular edema that developed following cataract surgery.

FIGURE 1 Macular hole closure at 3 months. Closure was successful in a higher percentage of eyes where the ILM was removed (ILM and ICG groups) compared to eyes where the ILM was not removed (ERD group), although the differences were not statistically significant. (more ...)

There was 1 persistent macular hole (3.1%) in the ILM group and 1 additional macular hole (3.1%) reopened between the 1-year and final examination. Both eyes had 1 additional surgery with successful closure of the macular holes for a final closure rate of 32 of 32 eyes (100%) All eyes in the ICG group maintained closure of the macular hole with no recurrent macular holes, for a final closure rate of 62 of 62 eyes (100%). Figure 2 presents the cumulative macular hole closure rates, defined as the percentage of eyes with closure of the macular hole with 1 surgery throughout the entire follow-up period.

FIGURE 2 Cumulative macular hole closure rate. This is the percentage of eyes that had long-term closure of the macular hole with one operation. Eyes with persistent and recurrent macular hole months or years later were subtracted from the total. ILM removal resulted (more ...)

The mean visual acuities at 3 months, 1 year, and the final examination are plotted in Figure 3. Visual acuity improved significantly more in eyes in the ERD and ILM groups at 3 months compared to the ICG group (P < .001 comparing the ERD to ICG group, and P = .036 comparing the ILM to ICG group). There were no statistically significant differences in the visual acuities between the 3 groups at the 1-year or final examination. Eyes in the ERD and ILM groups showed improvement at 3 months compared to preoperatively, with some decline at 1 year due to nuclear sclerotic cataracts in most phakic eyes and further improvement at the final examination when over two-thirds of eyes were pseudophakic. The eyes in the ICG group did not show as much improvement at 3 months but did show gradual improvement at the 1-year and final examinations.

FIGURE 3 Mean visual acuities for the 3 groups plotted over time. Eyes in the ERD only group (without removal of the ILM) had significantly more improvement at 3 months compared to both groups where the ILM was removed, despite a lower rate of successful closure. (more ...)

Figure 4 presents the mean change in visual acuity by the number of Snellen lines (1 Snellen line = 0.1 logMAR unit). Again, the ERD eyes showed the greatest improvement at 3 months, which was significantly better than the ICG group (P = .004). The mean visual acuity improvements were similar by the 1-year and final examinations (+3.6 lines for ERD, +3.9 lines for ILM, and +3.2 lines for ICG groups).

FIGURE 4 Mean visual acuity gain. Gain was higher in the eyes without removal of ILM at 3 months compared to the other 2 groups, although the visual acuity gain was similar at the final examination in all 3 groups. ERD, epiretinal dissection; ICG, indocyanine (more ...)

Figure 5 shows the percentage of eyes that gained at least 0.3 logMAR unit (3 or more Snellen lines) in the ERD, ILM, and ICG groups. The ERD eyes again showed a statistically significant improvement at 3 months (20/29 eyes, or 69%, compared to 23/62 eyes, or 37%; P = .007). There were no statistically significant differences at the 1-year or final examinations.

FIGURE 5 The percentage of eyes that gained 3 or more Snellen lines. Gain was significantly higher in eyes where the ILM was not removed at 3 months, but was very similar at the final examination. ERD, epiretinal dissection; ICG, indocyanine green; ILM, internal (more ...)

Other visual parameters (Figure 6) showed similar trends with a higher percentage of eyes in the ERD group attaining a visual acuity of 20/40 or better at 3 months (10/29 eyes [35%]) compared to the ILM (5/32 eyes [16%]) and ICG (3/62 eyes [ 5%]) groups. The difference between the ERD and ICG groups was significant (P = .0004), whereas the difference between the ILM and ICG groups did not reach statistical significance (P = .12). The percentage of eyes that were 20/40 or better at the 1-year and final examinations was similar in the 3 groups (11/29 [38%] of ERD, 11/32 [34%] of ILM, and 18/62 [29%] of ICG groups, with no significant differences between the 3 groups).

FIGURE 6 Percentage of eyes that attained a visual acuity of 20/40 or better. Percentage was significantly higher in eyes where the ILM was not removed at 3 months, but the percentage of eyes with 20/40 or better visual acuity was very similar at the final examination (more ...)

One eye in the ERD group developed a peripheral rhegmatogenous retinal detachment (with an attached macula), for an 0.8% incidence of retinal detachment in 123 eyes. One eye in the ICG group developed a presumed bacterial scleritis at the location of one of the sclerotomy sites 3 months following surgery, which responded to topical antibiotics. There were no other serious complications, such as infectious endophthalmitis or recognized macular ICG toxicity, in the ICG group. Intraocular pressure became transiently elevated in some eyes immediately following surgery, but all eyes had an intraocular pressure of less than 25 mm Hg by the 3-month, 1-year, and final examinations. Progression of nuclear sclerotic cataracts was expected in phakic patients, and all patients over age 50 years showed increased nuclear sclerosis. The mean nuclear sclerosis scores at 3 months were 0.96 in the ERD group, 1.027 in the ILM group, and 0.84 in the ICG group. The mean nuclear sclerosis scores at 1 year were 2.0 in the ERD group, 2.17 in the ILM group, and 1.38 in the ICG group. Cataracts had been removed in an additional 4 eyes (13.8%) in the ERD group, 11 eyes (34.4%) in the ILM group, and 17 eyes (27.4%) in the ICG group at 1 year. The mean nuclear sclerosis scores at the final examination were 1.94 in the ERD group, 2.31 in the ILM group, and 1.25 in the ICG group. Cataracts had been removed in an additional 7 eyes (24.1%) in the ERD group, 9 eyes (28.1%) in the ILM group, and 9 eyes (14.5%) in the ICG group between 1 year and the final examination. Only 9 of 29 eyes (31%) in the ERD group, 8 of 32 eyes (25%) in the ILM group, and 11 of 62 eyes (17.7%) in the ICG group were still phakic at the final examination.

The success of macular hole surgery has improved over the past 20 years with refinements in surgical techniques to achieve a higher likelihood of closing the macular hole and improving visual acuity compared to early studies. Some of the early innovators of macular hole surgery, such as Kelly and Wendel, advocated removal of what they described as cortical vitreous strands from the perimacular retina before the fluid-air exchange to help release any traction from residual cortical vitreous.²⁸ Their technique resulted in limited dissection around the edges of the macular hole, usually with forceps. The success rate for closing macular hole was 73% in their early series,²⁸ so they and other surgeons attempted to refine their surgical technique to achieve macular hole closure and visual acuity improvement in a larger percentage of eyes.

Adjuvants such as transforming growth factor β2, plasma, serum, and platelet lysate were used extensively in the 1990s to try to improve the results of macular hole surgery, but the use of adjuvants has largely been abandoned. Some surgeons began to experiment with removal of the ILM, and the first large series was reported by Brooks.¹ He was able to close the macular holes in 116 of 116 eyes (100%) with ILM peeling vs 36 of 44 eyes (82%) without ILM peeling, with a statistically significant improvement by removing the ILM. He also noted that ILM peeling seemed most beneficial in eyes with larger macular holes. Most recent studies have confirmed that ILM removal improves the results of macular hole surgery,^1–6 and one study that found no improvement had a sample size of only 22 eyes.⁷

The role of ICG staining of the ILM is more controversial where some studies found a benefit (or at least no harm) while other found evidence of toxic effects, resulting in poorer visual acuity or increased central scotomas. This suggests that the concentration of ICG, the way it is used to stain the ILM around the macular hole, and how completely the ICG is removed may have a role in potentiating or avoiding the harmful effects of ICG. Lochhead⁸ found improved results with ICG staining of the ILM compared to ILM removal without ICG use. Studies by Slaughter,¹⁰ Rivett,¹¹ Husson-Danan,¹² Mavrofrides,¹³ Kumagai,¹⁴ Villota-Deleu,²⁹ Rufer,³⁰ Oficjalska-Mlynczak,³¹ and Ullern³² found no significant differences between macular hole surgery with and without use of ICG to peel the ILM. Studies by Ando,¹⁶ Gass,¹⁷ Yamashita,¹⁸ Ferencz,¹⁹ Brasil,³³ and Tsuiki³⁴ have reported that ICG impairs the results of macular hole surgery. Ando reported a follow-up study using a lower concentration of .05% ICG and found no toxicity¹⁵ compared to his earlier study, though.¹⁶

ICG can cause macular toxicity, as was noted with early ILM staining techniques for macular holes that used unfiltered ICG.³⁵ The technique by which ICG is used may influence whether ICG toxicity occurs in an individual eye, although some eyes may be more susceptible to ICG toxicity. The present study attempted to minimize the amount of ICG in the vitreous cavity by staining only the ILM around the macular hole with 1 or 2 drops of ICG in an air-filled eye. The ICG was positioned over the macular hole by rotating the eye as necessary. Once the perimacular ILM was stained, the ICG was lavaged from the eye. ICG can be excluded from entering the subretinal space through the macular hole by placing a perfluorocarbon bubble over the macular hole before ICG is placed into the eye .³⁶ This technique is especially helpful when the macular hole is associated with a retinal detachment.

Two published studies compared ILM removal using ICG or trypan blue. Lee³⁷ found no significant differences between macular hole closure, but postoperative visual acuities were significantly better in the eyes treated with trypan blue. A study by Beutel³⁸ found no statistically significant difference between visual acuity in eyes treated with ICG vs trypan blue, but visual improvement was somewhat better (not significant though) with trypan blue with fewer central scotomas. The higher success rate of closing small macular holes without ILM removal has caused Tadayoni³⁹ to advocate ILM removal only in macular holes over 400 μm.

The current study has several limitations inherent in a retrospective, nonrandomized study. One limitation was a smaller sample size in eyes without ILM peeling and eyes with ILM peeling without ICG, compared to removal of the ILM using ICG. Because the series was a consecutive series, little selection bias would be expected, since the same technique was used in all eyes with macular holes until a new technique was adopted. All other features of the macular hole surgery were identical, such as the surgeon, gauge of the vitrectomy system (20-gauge), intraocular gas tamponade (16% C₃F₈), and prone positioning (2 weeks). Another limitation is that visual acuities were taken using current correction rather than refracted visual acuities using a standardized ETDRS visual acuity chart and refraction. A third limitation is that cataracts were not removed in all phakic eyes by the last follow-up visit. The author believes that the study limitations would not likely change the primary findings of the study, though.

This study found no long-term deleterious effect of ICG on the results of macular hole surgery, but the short-term visual acuity recovery was delayed in eyes treated with ICG and also somewhat delayed in eyes with ILM removal, without use of ICG. This suggests that complete removal of the ILM slows the visual recovery, especially if ICG is used, but ILM removal results in a higher long-term macular hole closure rate. The delayed visual recovery in ICG eyes may be due to ICG itself or more complete removal of the ILM. Terasaki and associates⁴⁰ found delayed recovery in the macular b-wave 6 months following surgery in eyes with ILM removal, even when ICG was not used to peel the ILM. The slowed visual recovery can explain some of the discrepancies between studies that found differing results in macular holes treated with ICG-assisted ILM removal compared to eyes where ICG was not used. Studies that showed poorer results with ICG had visual acuities reported at varying times following surgery, so some of the visual results with ICG may have improved with longer follow-up. Internal limiting membrane removal with ICG in phakic eyes often results in decreased vision until after cataract surgery, since progressing cataracts decrease visual acuity by the time the macula has recovered from ICG-assisted ILM removal. The long-term results with and without use of ICG appear relatively equivalent, although eyes without ILM removal will have a higher probability of requiring reoperation for persistent or recurrent macular holes. The study suggests that excellent results can be obtained with removal of the ILM using ICG, but the use of ICG may not be necessary in all eyes with macular holes. It may be difficult to identify which macular holes require ILM removal preoperatively. Surgeons are reluctant to abandon a technique for macular hole surgery with ILM removal, which ensures close to a 100% rate of successful macular hole closure. Patients are also often reluctant to agree to a second surgery if the first operation was not successful, so closure of the macular hole with 1 surgery is very desirable.

Trypan blue has been offered as an alternative staining agent for macular hole surgery, but visualization of the ILM with trypan blue is substantially poorer than ILM staining with ICG. Triamcinolone particles can be dusted on the macula to identify where ILM has been removed, but triamcinolone does not stain ILM and allow easy identification of the edges of ILM. Further studies will hopefully identify other dyes that stain the ILM better than trypan blue but that allow more rapid visual recovery than ICG.

The main strength of this study is that it involved a single surgeon in a highly technical procedure. Its weaknesses include its nonrandomized design, lack of protocol refracted visual acuity measurement, and perhaps underpowered statistics.

Important baseline factors differed between the groups, such as the initial visual acuity and the type, stage, and duration of the macular hole. The study included traumatic holes, which carry different prognoses from idiopathic holes, and are not distinguished from the others. Also, relative to the ICG group, the ILM group had a statistically shorter duration of macular hole (3.1 vs 6.4 months), and fewer stage 4 holes (9.4% vs 30.6%). Indeed, the ILM group had a slightly better visual result than the ICG group, probably because it had a better prognosis at the start. What is surprising is that the epiretinal dissection (ERD) group had final visual acuity results about as good as any of the other groups, despite having a lower hole closure rate. How might this be?

Aside from differing baseline factors, retinal or retinal pigment epithelial toxicity from ICG deserves mention. Such has been reported to occur at concentrations as low as 0.1%.³ A concentration of half of that was used in the current study, but following fluid-air exchange, that can increase its effective concentration at the retinal surface. It is notable that the ICG group had the slowest rate of visual recovery of the study eyes and the worst visual outcome, in spite of a 100% hole closure rate. Could subtle ICG toxicity have limited visual recovery?

Finally, Dr Thompson’s study might be underpowered to detect small differences in outcome between the treatment groups. Standard deviations for visual acuity results were not given, so that the magnitude of differences required to assign significance cannot be calculated. Nevertheless, the mean change in acuity differed between groups by no more than 0.7 lines, suggesting that differences, if any, may not be clinically meaningful.

In summary, Dr Thompson’s study shows that lower reoperation rates can be obtained with ILM removal, but that removal may not be essential for good outcomes. He should be congratulated for contributing to our knowledge of macular hole surgery.

Financial Disclosures: None.

Regarding the comments of George Waring about measuring visual acuity, this is the problem we face. You know what good visual acuity is and you are absolutely right in stating this is an imperfect system of measurement. The projector chart was used for determining visual acuity in all the patients with their current spectacle correction. The Snellen visual acuities, such as 20/70-2, were converted to a log MAR scale, and then calculations were used with the log MAR scale values. How many lines of visual acuity gain or loss are clinically significant? In the retina community we really believe the standard is three lines or more. Visual acuity can vary by one or two lines from time to time. The comments of Travis Meredith regarding the Bascom Palmer Eye Institute series are interesting. I would like to follow all patients and measure vision acuities after cataract surgery; however, some of them have already been lost to follow-up. In a recent study Ingrid Scott reported the visual acuities of patients after macular hole surgery with a minimum of five year follow-up. Many of these patients had participated in a an earlier study by Bill Smiddy to determine the efficacy of transforming growth factor beta (TGF-beta) on macular hole surgery. The results were amazingly good and many patients continued to improve years later. The surgery is a worthwhile surgery. In reference to Dr. Grossniklaus question, the ILM was not submitted for histopathology but other studies have shown glial cells and presumed fragments of Mueller cells attached to the ILM fragments.

Funding/Support: None

Financial Disclosures: None

Author Contributions: The author designed the study, collected, analyzed, and interpreted the data, as well as prepared the manuscript.