RETINAL MIGRATION AND SURGICAL OUTCOME AFTER HEMI-TEMPORAL INTERNAL LIMITING MEMBRANE PEELING VERSUS CONVENTIONAL PEELING FOR MACULAR HOLE

Abstract

Purpose:

To investigate the anatomical changes and surgical outcomes of hemi-temporal internal limiting membrane (ILM) peeling and conventional ILM peeling for idiopathic macular hole (MH).

Methods:

This randomized controlled trial was conducted at 3 centers and included 50 participants with MHs of <400 µm in minimum diameter for a duration of <6 months. All participants had undergone vitrectomy with either hemi-temporal ILM peeling (Hemi group) or 360° ILM peeling (360° group) with an injection of 5% sulfur hexafluoride gas, with or without simultaneous cataract surgery, from July 2017 to January 2021. The rate of MH closure and distance of retinal migration were examined.

Results:

Of 50 eyes randomized in the 3 centers, the Hemi group comprised 23 eyes, the 360° group 23 eyes, and 4 eyes were eliminated from final analysis. There was a significantly higher rate of primary MH closure in the 360° group (Hemi group: 73.9% vs. 360° group: 100%, P = 0.009). Retinal migration to the optic disk on the nasal side was significantly shorter in the Hemi group at 1, 3, and 6 months postoperatively than in the 360° group. There was no significant difference between the two groups in retinal migration to the optic disk on the temporal side.

Conclusion:

Nasal retinal migration in patients who underwent the hemi-temporal ILM peeling method was significantly less than in those who underwent the 360° ILM peeling method. However, less nasal retinal migration did not contribute to the MH closure rate.

It has been suggested that anteroposterior vitreomacular traction may be involved in the mechanism of pathogenesis of idiopathic full-thickness macular hole (MH).¹ Pars plana vitrectomy with posterior vitreous detachment and gas tamponade for successful MH closure was first reported by Kelly and Wendel.² Later, the efficacy of peeling of the internal limiting membrane (ILM) for MH has been reported.³ Recently, some reports have focused on retinal migration after macular surgery, especially including ILM peeling. Ishida et al⁴ found that in patients with idiopathic MH undergoing vitrectomy with ILM peeling, postoperative displacement of retinal vessels was greater toward the optic disk in the temporal than in the nasal region. Yoshikawa et al⁵ also found that in the treatment of diabetic macular edema, eyes treated with ILM peeling had a shorter papilla-to-orbital distance than eyes without detachment. Those reports indicated that the retina, which had been peeled the ILM, moved toward the optic disk. Although the mechanisms of retinal displacement after surgical treatment have not been fully clarified.

We hypothesized that reducing retinal migration on the nasal side of the MH would be advantageous for hole closure (Figure 1) and developed a hemi-temporal ILM peeling method, peeling only the temporal side of the MH,⁶ which achieved a closure rate equivalent to 360° ILM peeling. However, that previous study was a retrospective observational one, and therefore, we investigated the efficacy of the hemi-temporal ILM peeling method in a prospective, randomized, controlled, multicenter study.

Fig. 1.

Retinal migration. A. ILM peeling causes retinal migration in the region of the optic disk, and the migration distance is greater in the temporal retina than in the nasal retina. B and C. A 66-year-old woman who underwent 360° ILM peeling; the distance between the temporal margin of the OD and the intersection of retinal vessels was measured. B. Preoperative. The intersection of temporal vessels from OD was 4,636 µm, and the intersection of nasal vessels from OD was 3,047 µm. C. Six months after surgery. The intersection of temporal vessels from OD was 4,491 µm, the intersection of nasal vessels from OD was 2,920 µm. Migration to OD on temporal side was 145 µm and nasal side was 127 µm. D. The hemi-temporal ILM peeling technique is a surgical procedure that does not peel the ILM of the nasal retina, thus suppressing its migration. E and F. A 53-year-old woman who underwent hemi-temporal ILM peeling; the distance between the temporal margin of the OD and the intersection of retinal vessels was measured. E. Preoperative. The intersection of temporal vessels from OD was 5,488 µm, the intersection of nasal vessels from OD was 3,235 µm. F. Six months after surgery. The intersection of temporal vessels from OD was 5,333 µm, and the intersection of nasal vessels from OD was 3,218 µm. Migration to OD on temporal side was 155 µm and nasal side was 7 µm. OD, optic disk.

Patients and Methods

Patients

This prospective, randomized clinical trial was performed in three centers. The study protocol was approved by the Institutional Review Board of St. Marianna University School of Medicine and adhered to the tenets of the Declaration of Helsinki. The study was registered in the University Hospital Medical Information Network (https://www.umin.ac.jp) (UMIN 000027810). The registration period for participants was from July 2017 to January 2021. All patients provided written informed consent before enrollment. The inclusion criteria were patients >40 years of age diagnosed with idiopathic MH and a minimum hole diameter of <400 µm who provided written consent for study participation. Exclusion criteria were glaucoma, uveitis, large MH (minimum hole diameter >400 µm), highly myopic MH (axial length 0.28 mm), choroidal neovascularization, history of trauma, or proliferative vitreoretinopathy. Patients were randomly assigned in a 1:1 ratio to receive hemi-temporal ILM peeling or 360° ILM peeling. The institution was used as an assignment adjustment factor. Randomization was computer-generated in blocks of 4 (Figure 2). The allocation results were communicated to the surgeons immediately before performing the procedure, and the patients were masked to treatment allocation.

Fig. 2.

Flowchart of patient screening, exclusion, and randomization processes in the current multicenter, randomized, controlled study.

Surgical Technique

All patients underwent surgery with a 27-gauge minimally invasive vitrectomy system using the Alcon Constellation Vision System (Alcon Laboratories, Inc, Fort Worth, TX), including a three-port trocar cannula system (Total Plus Pak). Those with cataracts underwent phacoemulsification with concomitant intraocular lens implantation. Triamcinolone acetonide was used intraoperatively to facilitate visualization of the vitreous and posterior hyaloids. The ILM was stained with Brilliant Blue G for better visibility, as described by Enaida et al.⁷ In the 360° group, the area of ILM peeling was defined as a radius of approximately 2 to 3 disk diameters centered on the MH. In the Hemi group, the ILM was grasped at the temporal raphe and peeled within 180° to the temporal edge of the vascular arcade, and therefore, the area to be peeled was within a radius of 180° on the temporal side of the MH. After fluid–air exchange, 5% sulfur hexafluoride (SF6) was used for tamponade in primary surgery. All patients were asked to maintain a prone position for 3 days after surgery. When conducting this prospective study, we did not set any criteria for the selection of a procedure for unclosed cases. The reason for this is that there is a possibility of situations that require reoperation, such as proliferative membrane formation. Therefore, the choice of additional procedures was left to the operator's discretion.

Outcome Measures

Best-corrected visual acuity (BCVA) and migration length were measured at baseline and 1, 3, and 6 months postoperatively. The size and state of the MH before and after surgery were measured manually using B-scan images of spectral-domain optical coherence tomography (Cirrus high-definition optical coherence tomography [High-Definition Optical Coherence Tomography]; Carl Zeiss Meditec, Dublin, CA). The distance between the temporal margin of the optic disk and the intersection of retinal vessels was measured using the caliper function of spectral-domain optical coherence tomography, as previously described^6,8,9 (Figure 1). The measurements were performed by technicians masked to the allocation factors to avoid measurement bias. Retinal thickness was measured on the nasal and temporal side of the macula at the Early Treatment Diabetic Retinopathy Study subfield 3 mm using spectral-domain optical coherence tomography.

Statistical Analysis

All statistical analyses were performed using JMP pro 13 software (SAS Institute Inc, Cary, NC). Analyses of patient characteristics, postoperative BCVA, and distance of migration to the optic disk between the two patient groups were performed using the Mann–Whitney nonparametric test. Primary MH closure rates were analyzed using the chi-square test. P values of <0.05 were considered to represent statistically significant differences.

Results

Patient characteristics are shown in Table 1. Fifty patients were registered to participate in the study, of whom 23 underwent hemi-temporal ILM peeling and 23 underwent 360° ILM peeling. Four patients were excluded because they had a minimum MH diameter >400 µm. Age, preoperative BCVA, basal and minimum MH diameters, and axial length did not differ between the two groups. Postoperative ophthalmic data are shown in Table 2. No significant differences were found in postoperative BCVA at 1, 3, and 6 months postoperatively. The primary MH closure rate was 100% (23 of 23 eyes) in the 360° group and 73.9% (17 of 23 eyes) in the Hemi group (P = 0.009). Six eyes with unclosed MH underwent secondary surgery. Two eyes received additional nasal ILM peeling and 20% SF6 gas injection, and four eyes received additional 20% SF6 gas injection alone. In all cases, closure was achieved within 1 week after secondary surgery.

Table 1.

Patient Characteristics

Characteristics	Group		P
	360° ILM Peeling	Hemi-Temporal ILM Peeling
No. of patients	23	23
Age (year) (mean ± SD)	64.4 ± 9.2	67.4 ± 9.6	0.30
Gender (M/F)	11/12	11/12
Preoperative BCVA (logMAR) (mean ± SD)	0.42 ± 0.16	0.45 ± 0.20	0.99
Minimum MH diameter (μm) (mean ± SD)	239 ± 115	256 ± 89	0.60
Basal MH diameter (μm) (mean ± SD)	594 ± 294	615 ± 211	0.79
Axial length (mm) (mean ± SD)	25.0 ± 1.7	24.0 ± 1.4	0.054

F, female; logMAR, logarithm of minimum angle of resolution; M, male.

Table 2.

Postoperative Ophthalmic Outcomes

Outcome	Group		P
	360° ILM Peeling	Hemi-temporal ILM Peeling
Postoperative BCVA (logMAR)
1 month	0.20 ± 0.18	0.15 ± 0.20	0.162
3 months	0.10 ± 0.20	0.07 ± 0.12	0.299
6 months	0.06 ± 0.19	0.02 ± 0.13	0.283
Primary MH closure rate (%)	100	73.9	0.009
Final MH closure rate (%)	100	100
The intersection of nasal vessels from OD (μm)
Preoperative	2,830.0 ± 464.3	2,770.2 ± 413.8	0.692
1 month	2,742.8 ± 440.2	2,734.4 ± 386.0	0.952
3 months	2,732.0 ± 475.1	2,695.1 ± 371.4	0.809
6 months	2,703.0 ± 393.4	2,684.8 ± 393.0	0.902
The intersection of temporal vessels from OD (μm)
Preoperative	4,995.9 ± 461.9	5,320.8 ± 691.3	0.100
1 month	4,846.9 ± 477.1	5,160.4 ± 671.3	0.110
3 months	4,837.8 ± 468.2	5,086.9 ± 639.2	0.198
6 months	4,764.8 ± 445.5	5,034.8 ± 646.4	0.195
The nasal said retina thickness (μm)
Preoperative	356.9 ± 35.5	351.9 ± 28.8	0.796
1 month	341.5 ± 25.6	343.4 ± 18.8	0.627
3 months	339.0 ± 24.1	346.1 ± 22.6	0.268
6 months	329.2 ± 20.8	350.1 ± 26.5	0.013
The temporal said retina thickness (μm)
Preoperative	333.8 ± 31.4	325.8 ± 22.3	0.535
1 month	302.8 ± 25.4	308.7 ± 19.2	0.321
3 months	298.0 ± 23.8	303.6 ± 19.6	0.324
6 months	285.9 ± 30.0	300.3 ± 18.9	0.092

logMAR, logarithm of minimum angle of resolution; OD, optic disk.

Subsequently, 23 eyes in the 360° group and 17 eyes in the Hemi group with successful primary MH closure were analyzed. The mean migration length to the optic disk of the temporal retina was 149.0 ± 28.9 µm (±Standard Error) at 1 month, 176.9 ± 19.1 µm at 3 months, and 193.1 ± 33.7 µm at 6 months postoperatively in the 360° group. In the Hemi group, the retinal migration length of the temporal retina was 160.3 ± 20.5 µm at 1 month, 166.8 ± 24.6 µm at 3 months, and 186.2 ± 23.5 µm at 6 months postoperatively. In the temporal retina, there was no significant difference in the migration length between the two groups at any postoperative visit. The mean migration length to the optic disk of the nasal retina was 87.9 ± 15.8 µm at 1 month, 108.0 ± 17.1 µm at 3 months, and 118.0 ± 15.8 µm at 6 months postoperatively in the 360° peeling group. However, the mean retinal migration length was 35.8 ± 12.4 µm at 1 month, 53.7 ± 18.9 µm at 3 months, and 56.6 ± 22.9 µm at 6 months postoperatively in the Hemi group (Figure 3). These were significantly shorter than those of the 360° group at 1, 3, and 6 months (P < 0.05, respectively). The mean distances from the optic disk to the intersection of the nasal vessels were 2,830.0 ± 464.3 mm preoperatively, 2,742.8 ± 440.2 mm at 1 month, 2,732.0 ± 475.1 mm at 3 months, and 2,703.0 ± 393.4 at 6 months postoperatively in the 360° group. In the Hemi group, the distances were 2,770.2 ± 413.8 mm preoperatively and 2,734.4 ± 386.0 mm at 1 month, 2,695.1 ± 371.4 mm at 3 months, and 2,684.8 ± 393.0 mm at 6 months postoperatively, that, at any time point examined, there were no significant differences between them (P = 0.692, P = 0.952, P = 0.809, and P = 0.902, respectively). The mean distances from the optic disk to the point of intersection of the temporal vessels were 4,995.9 ± 461.9 mm before surgery and 4,846.9 ± 477.1 mm at 1 month, 4,837.8 ± 468.2 mm at 3 months, and 4,764.8 ± 445.5 mm at 6 months postoperatively in the 360° group, whereas they were 5,320.8 ± 691.3 mm before surgery and 5,160.4 ± 671.3 mm at 1 month, 5,086.9 ± 639.2 mm at 3 months, and 5,034.8 ± 646.4 mm at 6 months postoperatively in the Hemi group (Table 2). At no time point were significant differences found between the two groups (P = 0.100, P = 0.110, P = 0.198, and P = 0.195, respectively). The nasal side retinal thickness was not significantly different preoperatively and at 1 month and 3 months postoperatively but was significantly thicker in the Hemi group at 6 months postoperatively (P = 0.013). There was no significant difference in retinal thickness on the temporal side between the Hemi and 360° groups preoperatively and at 1, 3, and 6 months postoperatively (Table 2).

Fig. 3.

Retinal migration to the OD. Distances from the margin of the OD to the intersections of temporal vessels and nasal vessels were measured in near-infrared images. A. Differences in length before and after surgery over time on the temporal side. There was no significant difference in the migration length between the two patient groups in any month postsurgery. B. Differences in length before and after surgery over time on the nasal side. The migration lengths in the Hemi group were significantly shorter at 1, 3, and 6 months postsurgery (P < 0.05). OD, optic disk.

Discussion

To remove the tractional forces at the vitreomacular–macular interface, vitrectomy with ILM peeling has been widely used for two decades, leading to functional recovery.^10,11 The inverted ILM flap technique for refractory large MH improved the postoperative closure rate for a decade.¹² It thus seems that MH surgery has been established with good structural and functional outcomes. Seeking further progress, we focused on macular migration toward the optic disk after ILM peeling in several conditions.^4,5 We believed that reducing nasal retinal displacement to the optic disk by not performing ILM peeling would be advantageous for hole closure and devised the hemi-temporal ILM peeling method. We previously found that nasal retinal vascular migration after hemi-temporal ILM peeling was significantly less than that after 360° ILM peeling, with the same MH closure rate after the two procedures.

The mechanisms of retinal migration after ILM peeling have not been fully elucidated. It was reported that peeling the ILM upsets the balance between the nasal and temporal retina, causing displacement of the fovea to the optic disk.^9,10 Because it does not damage the central papillomacular bundle, hemi-temporal ILM peeling has been considered a useful and safe surgical technique.⁶ Table 2 shows the migration length to the optic disk at 1, 3, and 6 months after hemi-temporal peeling. Similar to our results, Liu et al⁹ also found that the migration length to the optic disk from 1 month after surgery tended to increase with time. Other groups reported that a macula in which the ILM had been peeled slipped toward the optic disk after surgical treatment for MH.^4,10,13 Significant retinal migration can occur after epiretinal membrane removal,⁸ which is thought to be a change associated with retinal elasticity.^14,15 The ILM is composed of very thin tissue, and its structural strength is approximately 1,000-fold greater than that of other cell layers. The ILM accounts for 50% of the retinal stiffness.^16,17 Thus, ILM peeling increases retinal elasticity, causing the fovea to move toward the disk.

However, nasal retinal migration was significantly less in the Hemi group than in the 360° group in the present study, suggesting that the unpeeled nasal ILM of an MH is more likely to limit the displacement of the retina in the direction of the optic disk. Ishida et al⁴ reported a more significant displacement of the temporal retina toward the optic disk than the nasal retina after MH surgery. Nerve fibers are mainly composed of microtubules, and it is known that axon shrinkage is caused by depolymerization of microtubules.¹⁸ The ILM is the basal lamina of the Müller cells, and permanent damage to part of the Müller cells leads to a loss of structural support of the retina and the retinal nerve fiber layer. It was thought that ILM peeling damages nerve fibers by injuring Müller cells, causing axon contraction. The optic nerve fibers are tethered to the lamina cribrosa,¹⁹ and the nerve fibers in the retina would move toward the optic disk after shrinkage of the fibers. The nasal retina has less retinal movement because of the shorter distance to the lamina cribrosa. However, in this study, retinal movement was further reduced by not performing ILM peeling.

After Kelly and Wendel reported the efficacy of vitrectomy with posterior vitreous detachment and gas tamponade for MH,² many studies showed that the closure rate was improved by performing ILM peeling.^3,20–22 An intraocular gas bubble applies anteroposterior force on the retina. Postoperatively, the face-down prone position causes the gas to adhere to the MH and promotes its closure.²³ It is also possible that the bubble adheres to the macula and serves as a scaffold for glial cell migration, promoting and maintaining the eventual closure of the hole.^24,25 The rationale behind ILM peeling is to relieve the traction forces occurring around the fovea and to ensure the removal of all epiretinal tissue, including the epiretinal membrane, which is responsible for foveal traction.²⁶ ILM peeling also stimulates gliosis across the MH, which is advantageous for hole closure.²⁷

We conducted this study because we thought that retinal migration with ILM peeling might be advantageous as a novel mechanism for MH closure. However, the primary MH closure rate was 73.9% in the Hemi group and 100% in the 360° group. As we hypothesized, there was significantly less nasal retinal migration in the Hemi group than in the 360° group, and there was no significant difference in temporal retinal migration between the two groups. However, the present results suggest that retinal migration has little effect on MH closure. In a study of daily spectral-domain optical coherence tomography imaging from the day after MH surgery, the average time required for MH closure was 1.2 days.²⁸ However, the amount of retinal migration increased monthly, as shown in the present study (Figure 3), and other groups reported similar results.^5,6,8–10 The retinal thickness is also thought to be affected by the retinal migration. The Hemi group tended to have thicker retina than 360° group because of retinal migration from the temporal side, but the difference was not significant until 6 months after surgery, so the retinal migration was considered to be a monthly change. In the present study, we found that the retinal migration associated with ILM peeling was not abrupt, and therefore, the lack of nasal ILM peeling did not have an advantage in closing the MH.

Another factor that may have influenced the decrease in closure rate was the small area of ILM peeling. A meta-analysis reported that ILM peeling is effective as a surgical method for MHs because it relieves the traction force generated around the fovea and reliably removes the epiretinal tissue that may cause foveal traction.²² Iezzi and Kapoor²³ reported that broad ILM peeling may enhance overall retinal compliance by restoring the intrinsic elasticity of the neurosensory retina, allowing it to relax after tangential forces are removed. As a result, they were able to achieve MH closure without the use of the face-down position by performing broad ILM peeling. Thus, hemi-temporal ILM peeling was thought to significantly reduce retinal compliance compared with 360° ILM peeling.

For additional treatment of patients in the present study in whom MH closure was not achieved, two eyes received additional nasal ILM peeling and 20% SF6 gas injection, and four eyes received additional 20% SF6 gas injection alone. In all cases, closure was achieved within 1 week after the additional treatment. This suggested that the Hemi group required more time for closure than the 360° group. Therefore, it was thought that the use of long-term retention gas from the initial surgery would lead to an improvement in the closure rate.

We recognize several limitations in this study. First, surgical outcomes using long-term retention gas had not previously been performed. Therefore, if retinal compliance contributed to the closure rate, it was considered necessary to investigate the use of 20% SF6 gas. Second, it is unclear whether the two eyes that received additional nasal ILM peeling could have been closed with additional gas alone without the need for additional ILM peeling. If all cases had initially received 20% SF6, then 100% closure could possibly have been achieved in both groups, with the potential benefit of no damage to the central papillomacular bundle. In cases of glaucoma or high myopia with thinning of the retinal thickness, the initial surgery may be considered to be a hemi-temporal ILM peeling with 20% SF6 to preserve the ILM, but in small MH, 360° peeling is advantageous for closure.

In conclusion, nasal retinal migration after the hemi-temporal ILM peeling method was significantly less than that after the 360° ILM peeling method. However, this prospective, randomized controlled trial showed that less nasal retinal migration did not contribute to an increased rate of MH closure.