Long-Term Quantitative Analysis of Inner Retinal Dimples and Visual Function Post Internal Limiting Membrane Peeling in Macular Diseases

Internal limiting membrane (ILM) peeling provides promising therapeutic outcomes in macular disease, such as increasing the closure rate of idiopathic macular hole (iMH), reducing the recurrence of idiopathic epiretinal membrane (iERM), assisting foveal reattachment in myopic foveoschisis (mFS). However, the effect of ILM peeling on visual function remains controversial [1]. ILM peeling has been reported to be associated with the decrease of macular thickness [2] and the recovery postponement of b-wave amplitudes in focal macular electroretinograms [3]. In 2001, Tadayoni et al. [4] first found inner retinal dimples (IRDs) on a B-scan of optical coherence tomography (OCT) after epiretinal membrane removal, which coincided with the dark arcuate striae along the course of the optic nerve fibers on a C-scan of OCT, which was the so called “dissociated optic nerve fiber layer (DONFL).” Since then, IRDs (or DONFL) have been reported to appear after ILM peeling in various diseases [5‐7]. Some studies [8‐10] pointed out that IRDs might contribute to the decrease of visual function after ILM peeling. Others [11‐13] insisted no difference in retinal sensitivity (RS) between arcuate striae and normal retina. However, the majority of previous studies only depicted IRDs in iMH, overlooking other macular diseases such as iERM and mFS. Additionally, the measure equipment or method in the above-mentioned studies may have failed to detect a mild degree of inner retinal defect in whole ILM peeling area. Using the newest Nidek’s microperimetry (MP-3), we previously confirmed high incidence of scotomas in iMH patients with IRDs [7]. Nevertheless, the natural course of IRDs after ILM peeling and its impact on visual function in other macular diseases remain uncertain.

In the present study, we evaluated IRDs and RS in different sectors using optical coherence tomography angiography (OCTA), spectral-domain optical coherence tomography (SD–OCT), and MP-3 at multiple postoperative timepoints in patients with iMH, iERM, or mFS undergoing pars plana vitrectomy (PPV) combined with ILM peeling. We aimed to investigate the pattern of postoperative IRDs and its impact on visual function.

In the present study, we found that IRDs increased gradually at ILM-peeled area with time in iMH, iERM, and mFS. Most IRDs distributed in temporal sector in both perifovea and parafovea, interrupting the ganglion cell layer. The progress of IRDs may influence postoperative RS only in iMH and iERM. Then we identified that baseline axial length was a protective factor for IRDs after ILM peeling.

Our results showed that IRDs occurred at superficial layer of inner retina at 1 M and gradually went through GCL. ILM is composed of vitreous-facing endfoot membranes of Müller cells and basal lamina [16]. Histopathological studies [17, 18] showed that cellular fragments belonging to Müller cells appeared on the retinal side of peeled ILM. In vitro, Taylor et al. [19] emphasized the importance of physical support by ILM in maintaining the structure of inner retina. In vivo, the endfoot of Müller cells contain enormous mechanosensitive ion channels [20] and the processes of Müller cells penetrate nearly whole layers of retina as “bridges” with photoreceptor’s axons [21]. Once biomechanical changes were detected, mechanosensitive channels may transfer certain signals to the entire retina. The loss of inner retina mechanic support might influence the layer near RNFL initially and then expand along the processes of Müller cells to the entire retina layers. Thus, we suspected that the lack of support in inner retina, stimulation of stretch force, and damage to the endfoot of Müller cells might contribute to the formation of IRDs. Nevertheless, we found that IRDs did not reach IPL, which might be because the supracellular alignment of tissue redistributes mechanical energy before it reaches the nucleus [22].Previous electron microscopic findings showed [23] that the perikaryons of Müller cells are located in the inner nuclear layer. To note, the follow-up period of our study is 12 months. Future study with prolonged follow-up is needed to investigate whether IRDs would expand along the processes of Müller cells in the long run.

Most IRDs were found in temporal sector in both perifovea and parafovea in our study, which might be explained by the following reasons. First, histological studies showed that the Henle fiber layer (HFL), which was composed of the processes of Müller cells, horizontally extends the most in the temporal retina and least in nasal retina [16], making the processes of Müller cells more prone to contraction and degeneration. Besides, Veronese et al. [24] found that the nasal side of the fovea was more resistant to traction because nerve fibers and retinal ganglion cells (RGCs) were bound together closely in the papilla–macular bundle, which may make it difficult for dimples at the nasal subfield to “sink” widely. In addition, the retina thinning was mostly seen in the temporal subfield after ILM peeling [25‐27], because of foveal displacement toward optic disc. The thinner retina in temporal field might make it more prone to contracting damage, thus forming more IRDs.

Next, we investigated the change of visual function after ILM peeling. Great improvement in BCVA was observed in all patients. However, RS was improved only in parafovea of iMH and mFS. No improvement in RS was observed in the perifovea, particularly in the temporal sector of iMH (RS actually declined with time after ILM peeling; Fig. 9). Ultrastructure research [16, 28] showed that Müller cells in HFL supported structural stability through increasing compliance and resistance by its “z-shape” at the parafovea. In contrast, at the perifovea, shorter Müller cells aligned in a vertical pattern as pillars, through which it was easier to transmit mechanical stress directly to photoreceptors without a “z-shaped” “spring buffer.” Furthermore, the densities of photoreceptor and RGCs were less in the temporal sector compared with the nasal retina [29], resulting in the insufficient compensation of light path in this area.

This study illustrated a negative correlation between IRDs and RS beyond central fovea in iMH or iERM, which might explain why some patients experienced scotoma and metamorphopsia after ILM peeling. IRDs were possibly secondary to damage in Müller cells. Being interposed by neurons and photoreceptors, Müller cells participate in light transmitting through endfeet to main process as “optical fiber” and lend support to surrounding microenvironment functionally and metabolically [16, 21, 30]. ILM peeling might damage the endfoot of Müller cells, hampering the microenvironment of RGCs, causing anoikis and eventually leading to the loss of visual function [31‐33]. To support this, Akino et al. [34] confirmed that visual field defects were associated with GCL thinning in iMH after ILM peeling. However, no significant correlation between IRDs and RS was found in mFS, which might be due to the low IRDs count after ILM peeling.

We also explored the relationship between preoperative factors and IRDs, finding a significant negative correlation between axial length and IRDs at 1 M and 12 M. Also, patients with mFS presented the least IRDs across all the timepoints, which lent further support of the effect of axial length on postoperative IRDs. Long-term elongated axial length would enhance Müller cell gliosis [35]. The overstretch of Müller cell processes along with HFL may compromise the capacity of photoreceptors synapses [28]. Hence, we hypothesized that the retina in mFS might have already adapted to low metabolic demands under abnormal Müller cells, downregulating the mechanical influence caused by tearing the glial endfoot of Müller cells.

To our knowledge, this is the first study to quantitatively compare inner retina defect after ILM peeling combined with PPV among iMH, iERM, and MFS. Based on the findings of our study, we consider that ophthalmologist may be more conservative in the temporal sector during ILM peeling, especially in patients with normal axial length and adherent ILM–ERM. Future studies may pay attention to the biomechanical importance of ILM and extracellular matrix components of the inner retina to gain further insights into the pathogenesis of IRDs after ILM peeling.

Our study has several limitations, the first being that is a single-center retrospective study with relatively small sample size. Multicenter studies with larger sample sizes are warranted in the future. Second, all the procedures were completed by the same surgeon in our study, which limits the generalizability of our results. However, this assured the homogeneity of procedures. Additionally, it is reported that ICG may have adverse effects on retina. Therefore, we cannot be sure that the formation of IRDs is purely the result of mechanical stretch, since the dye only stays within eyes for less than 30 s. Future studies need to investigate IRDs after ILM peeling using a dye other than ICG.

Our study suggests that ophthalmologists may avoid the temporal sector of parafovea and perifovea, especially in eyes with normal axial length or strong ILM–ERM adherence during ILM peeling.