Since the first descriptions of pars plana vitrectomy (PPV) and gas for its treatment by Kelly and Wendell [1], macular hole has become one of the commonest reasons for vitrectomy surgery. Biomicroscopic staging of macular holes has been used since the descriptions of Gass [2, 3]. Success rates can be very high if the patients are operated upon early, but vary from 47 % to 100 % for hole closure [48] depending on the mix of stages of macular hole and the duration of the symptoms. In this study, we have re-examined the utility of the Gass staging system to predict macular hole closure rates and visual outcome in a large cohort of consecutive cases with prospective data collection.

Whilst various adjuvants have been tried in the management of the disorder [816], internal limiting membrane peeling (ILM) has become the method of choice, and its use improves the likelihood of primary anatomical closure [5, 1719]. To aid resection of the ILM, various vital dyes have been employed [2023]. Use of indocyanine green dye (ICG) persists as a methodology [5] despite its potential toxicity. ICG has been implicated in retinal pigment epithelial toxicity [2325], resulting in less visual improvement [27, 28]. The agent has been shown to persist in the eye up to 8 months after insertion [29, 30]. Some have suggested that ICG induces a different cleavage plane for ILM dissection [31] which may cause damage to retinal ganglion cells [32]. More recently, brilliant blue (BB) has been used, and is regarded as having a better safety profile [3338]. In our institution, ICG was available for surgeons to use prior to 2007, after which BB was available instead. Comparisons of the effects of the two agents on visual outcome are rare or confined to small numbers of patients [39]. In this study, prospectively collected data on consecutive patients with macular hole were examined for surgical methodology, complications, and visual outcome. In particular, outcomes related to macular hole stage and the use of BB and ICG were examined.

Materials and methods

Data from three datasets using the electronic medical record (EMR), VITREOR (Microsoft Access, available with the book Vitreoretinal Surgery, published by Springer) [40] on three sites in the Greater London area were analysed. All data were prospectively entered into the EMRs. The EMRs currently hold data on 14,085 operations. The duration of symptoms at time of surgery, stage of macular hole [3] (staged by biomicroscopy and optical coherence tomography, OCT), and visual acuity (logMAR) were recorded. Counting fingers, hand movements, and light perception were converted according to Holladay et al. [41] In addition, preoperative data on the status of the lens (including presence of visually significant cataract or pseudophakia) and date of prior cataract surgery were recorded. Operative data specified the use of internal limiting membrane peeling, intra-ocular dyes, intra-ocular gas, retinopexy, combined phaco-emulsification lens surgery, and listed complications. Follow-up required resolution of the gas bubble, and was for a minimum of 6 weeks for inclusion in the study. Ethical approval for the database study was obtained from the local research ethics committee (Guy's & St Thomas’ Hospitals Trust).

Only primary surgeries for idiopathic holes were included in the study. For patients who developed macular hole in both eyes, only the first eye was included in the present study. The diagnosis of postoperative hole closure was confirmed with OCT.

Data were analysed using statistical software (Analyse-it, Cambridge, UK). Means were compared using the t-test, medians by nonparametric methods, and proportions using Fisher’s exact test or Pearson Chi-squared test.


Three hundred and fifty-one macular hole surgeries with stages 2, 3, or 4 holes from 2001–2011 were included. Of these, 66.4 % were females. The mean age was 68.9 years, range 39–87 years, with 54.1 % right eyes. Follow-up duration was median 0.55 years range(range 6 weeks to 11.1 years). Mean preoperative spherical equivalent was +0.25 D, and mean intraocular pressure on presentation was 15.6 mmHg. At presentation, 15.4 % had a normal crystalline lens, 9.4 % pseudophakia, 63.8 % had nuclear sclerosis, 4.8 % had cortical lens opacities, 1.1 % posterior subcapsular opacity, and 5.4 % mixed cataract.

Twenty-gauge PPV was used in all except 44 patients who had 23-gauge surgery; 68.4 % had general anaesthesia, 16.2 % peribulbar anaesthesia, 14.2 % subtenons anaesthesia. In addition, three patients had topical anaesthesia, and one retrobulbar anaesthesia. In 18 patients, operation records were incomplete. Of the rest ILM peeling was performed with ICG, BB, trypan blue or without dye in 209, 109, eight, and seven patients respectively. At the completion of surgery, perfluoropropane (C3F8) was inserted in 92.7 % of cases, sulfahexafluoride (SF6) in 4.8 %, and sulphahexafluorane (C2F6) in 2.4 %; one patient had silicone oil inserted. Cryotherapy was used in 34.2 %, laser retinopexy in 2.3 %, and laser and cryotherapy combined in 1.4 %: panretinal photocoagulation was applied in one patient. Combined lens extraction by phaco-emulsification surgery at the time of the PPV was performed in 77 % of phakic eyes.

Iatrogenic breaks were identified at the time of surgery in 18.5 %, choroidal haemorrhage in two patients, and lens touch in one patient. Eighteen patients had cataract at final follow-up, three had glaucoma, and two silicone oil in situ. There were no patients with phthisis, and no retinal detachments at final follow-up.

Visual acuity

Vision was significantly improved from 0.97 (SD 0.45), Snellen equivalent 20/185 preoperatively to 0.65 (SD 0.51), 20/90 at final follow-up. Presenting VA correlated with final VA at follow up, r 2 = 0.225, p < 0.0001 (Fig. 1). More than or equal to 0.3 logMAR units (2 Snellen lines) of visual improvement was achieved in 52.3 % of patients.

Fig. 1
figure 1

The relationship between preoperative logMAR visual acuity and that at final follow-up

Stage of macular hole

One hundred and eighteen patients had stage 2 macular holes, 185 stage 3, and 48 stage 4. The proportion of patients who were female varied for the stage of hole, with stage 2 being 61.9 % female, stage 3 being 73 % female, and stage 4 being 52.1 % female, p = 0.01.

Stage two holes had significantly better mean presenting VA than either stage 3 or stage 4 holes [stage 2 mean of 0.78 (SD 0.39), 20/120, stage 3 mean 1.07 (SD 0.46), 20/240, stage 4 mean 1.05 (SD 0.43), 20/225], p < 0.0001. There was a significant difference in the mean duration of symptoms for the three stages of hole for the 204 patients with this variable recorded, i.e., stage 2, 0.53 years (SD 0.43), stage 3, 0.79 years (SD 0.68), and stage 4, 1.20 years (SD 1.26), p = 0.0002.

VA at final follow-up was significantly associated with stage of hole, with stage 2 achieving mean 0.42 (SD 0.33), 20/50, stage 3 0.75 (SD 0.53), 20/110, and stage 4 0.87 (SD 0.60), 20/145, p < 0.0001.

The improvement in VA was similar between stage 2 [0.37 (SD 0.37)] and stage 3 [0.31 (SD 0.55)] holes, but was significantly worse for stage 4 holes [0.18 (SD 0.54)] as compared with stage 2 holes, p = 0.008.

The proportions of patients improving more than or equal to 0.3 logMAR units (2 lines of improvement) were for stage 2 holes 58.5 %, stage 3 50.8 %, and stage 4 42.6 %, which were statistically similar. The proportions of patients achieving a final vision better than 0.3 logMAR (20/40) was greatest for stage 2 (37.4 %), and statistically less for stage 3 (11.6 %) and stage 4 (15.4 %), p < 0.0001. Closure rates of the holes were significantly different according to stage, with stage 2 closing in 95.8 %, stage 3 in 73.0 %, and stage 4 in 56.3 % of cases, p < 0.0001 (Fig. 2).

Fig. 2
figure 2

The percentage of macular holes which remained open or closed for each stage of hole

Closed holes achieved an improvement in visual acuity of 0.43 (SD 0.40), whereas open holes suffered a loss of vision of −0.11 (SD 0.57) postoperatively (Figs. 3 and 4). These results were statistically different, p < 0.0001. Stage 4 macular holes with closure (n = 27) had better vision after surgery at VA 0.57 (SD 0.48), 20/75, compared with 1.01 (SD 0.44), 20/200 preoperatively, p < 0.0001 (Fig. 3). Stage 4 (n = 21) without closure were unchanged at 1.2 (SD 0.49), 20/300 after surgery compared with 1.09 (SD 0.41), 20/250 pre-operatively (Fig. 4).

Fig. 3
figure 3

Post-operative mean LogMAR VA was significantly improved in patients whose macular holes closed for all stages of hole (stages 2,3, and 4) compared to pre-operative visual acuity, p < 0.0001

Fig. 4
figure 4

For macular holes which remained open post op there was no significant difference in pre-operative (preop) and post-operative (postop) logMAR VA across all stages of hole (2,3, and 4)

Indocyanine green and brilliant blue

Patients who received indocyanine green (ICG, 209 patients at 0.5 mg/ml) or brilliant blue dye (BB, 109 patients) to assist internal limiting membrane peel were compared to each other. Presenting VA was similar in the ICG group and the BB group (ICG 0.98, SD 0.47, 20/190 and BB 0.93, SD 0.40, 20/170). There were more females in the ICG group (70.3 %) than in the BB group (58.7 %, p = 0.05). Other presenting variables were similar between the groups. Postoperative VA was 0.71 (SD 0.53), 20/100 for ICG and 0.52 (SD 0.43), 20/70 which was significantly different, p = 0.003. Similarly the change in VA was significantly less for ICG (0.26, SD 0.51) than for BB (0.40, SD 0.47), p = 0.02.

The proportion of patients who achieved a closed hole was less for ICG (73.2 %) than BB (89.9 %), p = 0.0005.

In case the hole closure rate affected the VA results, the VA results for those patients with successfully closed holes were also tested separately. Final VA was then statistically similar between the groups, ICG 0.52 (SD 0.38), 20/70 and BB 0.44 (SD 0.31), 20/55 (p = 0.09). Change in VA was 0.39 (SD 0.41) for ICG and 0.48 (SD 0.40) for BB, which was not significant, p = 0.08. When the analysis was performed for only those patients with stage 2 hole, mean improvement in VA was significantly better for BB (0.47, SD 0.36) than for ICG (0.30, SD 0.31), p = 0.01 (Fig. 5).

Fig. 5
figure 5

The mean improvement in logMAR vision for holes of each stage that were closed successfully following the use of indocyanine green (ICG) or brilliant blue (BB) dyes. The visual improvement was statistically greater for BB compared with ICG for stage 2 macular holes only, p = 0.01


Overall, our closure rates are comparable to other studies of the success of surgery; Ruby et al. studying stage 2 holes reported a hole closure rate of 88 %, with 60 % achieving 20/50 or better [42]. Freeman et al. with stages 3 and 4 holes reported a hole closure rate of 69 % [43], and 60 % achieving 20/80 or better. In comparison, the hole closure rates in this study for stage 2 were 95.8 %, stage 3, 73.0 %, and stage 4, 56.3 %. Comparison of visual results between this and other studies is complicated by variation in the length of follow-up; median follow-up here was relatively short at 0.55 years, and so further improvement in eyes with anatomically successful surgery can be expected [44]. Nevertheless, visual acuity at final follow-up was significantly better by stage of hole, with stage 2 achieving mean 20/50, stage 3, 20/110 and stage 4, 20/145.

This study reinforces the contention that stage of hole is related to success of hole closure after surgery and to visual recovery. The results also illustrate that it is imperative for surgery to be optimized to achieve hole closure, as visual acuity improved in eyes in which hole closure was achieved, but deteriorated in eyes where there was a failure in hole closure. The stage of hole was related to the duration of symptoms, with stage 2 at 0.53 years of symptoms, stage 3 at 0.79 years, and stage 4 at 1.20 years, indicating that there is a need for prompt referral of these patients as soon as symptoms occur, and the avoidance of delay for surgery. In patients with macular holes of more than 1 year in duration, success rates have been shown to vary from 47 % [4] to 81 % [45]. This is reflected in the results of this study, with stage 3 and 4 holes with long duration achieving progressively poorer closure rates.

Many studies are confined to an examination of earlier onset holes, and therefore achieve high success rates. The reduced closure rates of later stages and older holes may discourage the surgeon from offering intervention. However, the improvement in vision seen in stage 3 and 4 holes when they are successfully closed shows that surgery should still be utilized in these patients. For example, in those patients with stage 4 holes in whom closure was achieved, the mean final vision was 20/75. More effective means of improving hole closure rates are needed for the later stages of the disorder, and this is an area for future innovations in surgical technique. In this study, a similar method of surgery was employed for all operations with PPV, and in most patients long-acting gas and ILM peel using one of two staining methods . The variety of dyes used allowed an examination of the effects of ICG and BB on outcomes. This study describes a comparison of cohorts of patients with the use of ICG and BB involving the largest number of patients to date. There remains an active debate over which agent to use, with slightly better visualisation with ICG but issues of toxicity [2432]. Low dosages of ICG as used in this study have been regarded as safer; however, this is uncertain. BB was introduced as an alternative dye with lower toxicity rates [33]. Our comparison found poorer visual outcomes with ICG compared with BB. This may have reflected the poorer hole closure rate seen with ICG. For this reason, an analysis was performed of the visual results of only those patients with hole closure. In this group, the visual acuities improvement were better when BB was used, a result which was only statistically different for stage 2 holes. This could be due to clinically undetectable low-grade toxicity from ICG, although there is also the potential for other uncontrolled differences between the comparison groups which could also be of significance. Nonetheless, given better closure rates and visual outcomes with BB, the results of this study would suggest the use of BB in preference to ICG.

In contrast to other studies, we found no correlation of visual outcome and age, but there was a relationship with increased preoperative visual acuity and hole stage [46]. A criticism of the study is that we did not examine OCT findings such as hole diameter which might have been helpful in predicting outcome [46]. OCT findings have been extensively studied elsewhere, and are well-understood. The study was of consecutive cases, and therefore suffers from problems related to comparing groups historically. Even so, the reduction in visual outcomes in stage 2 holes suggests that BB should be used in preference to ICG where available.

In conclusion, macular hole stage provides a useful measure to help predict the chance of postoperative hole closure and visual outcome. The relationship of duration of symptoms to increasing stage suggests these patients require prompt referral and early surgery. An examination of dyes used in ILM peel suggests that BB is preferable to ICG for better visual outcomes.