Epiretinal membrane peeling in eyes with pseudophakia: prognostic and morphological factors of visual outcomes Authors: Tzu-Ying Lee, Shih-Chou Chen, Chui-Lien Tsen, Tsung-Tien Wu Published: 2025 ABSTRACT PurposeIn patients with symptomatic epiretinal membrane (ERM), pars plana vitrectomy (PPV) and ERM peeling, followed by internal limiting membrane (ILM) peeling is the most common treatment. In this study, spectral domain optical coherence tomography (SD-OCT) was used to investigate the predictive factors of visual outcomes in patients with pseudophakic ERM following surgery.MethodsThe records of 61 pseudophakic eyes with ERM undergoing PPV, ERM peeling, and ILM peeling were retrospectively reviewed. OCT features such as central foveal thickness (CFT), cotton ball sign, ERM foveoschisis, ellipsoid zone (EZ) disruption, external limiting membrane (ELM) disruption, ectopic inner foveal layer (EFIL), disorganization of the retinal inner layers (DRIL), dissociated optic nerve fiber layer (DONFL), and microcystic macular edema (MME), as well as surgical factors, were compared before and after surgery. The correlations between the OCT and surgical factors and change of best-corrected visual acuity (BCVA) before and after surgery were examined.ResultsThe mean follow-up time was 6 months. BCVA (P < 0.01) and CFT (P < 0.01) showed significant improvements after ERM removal. McNemar test results indicated that only the cotton ball sign (P = 0.002) and DRIL (P < 0.001) exhibited statistically significant improvement after surgery. Multiple linear regression analyses showed that ERM-foveoschisis (P = 0.015) and MME (P = 0.009) were significantly correlated with less improvement of visual acuity, while better visual acuity before surgery was significantly associated with better improvement of visual acuity (P < 0.01).ConclusionsBetter baseline visual acuity is correlated with better visual acuity improvement after surgery, while ERM foveoschisis and MME identified preoperatively on OCT are associated with relatively less improvement of visual acuity. Introduction Epiretinal membrane (ERM), also referred to as macular pucker, is a sight-threatening disorder. The formation of an ERM is related to abnormality of the vitreoretinal interface combined with posterior vitreous detachment, causing a pathological fibrocellular layer that proliferates over the internal limiting membrane. Patients may present with decreased or distorted central vision, metamorphopsia, or may be asymptomatic. An idiopathic epiretinal membrane (iERM) accounts for most cases of ERM, while ocular trauma, uveitis, or retinal detachment can result in secondary ERM. Symptomatic ERM is typically treated with pars plana vitrectomy (PPV), ERM peeling, followed by internal limiting membrane (ILM) peeling [1]. Spectral domain optical coherence tomography (SD-OCT) has become the standard method for diagnosing ERM. The method identifies microstructural abnormalities of the retinal layers to determine ERM severity before and after surgery. Recent evidence indicates that SD-OCT detected anatomical changes can help predict postoperative visual acuity [2]. The study by Raquel et al. further supports using OCT-based staging and structural markers to guide prognosis after idiopathic ERM surgery [3]. A review of published studies found that few studies have listed cataracts as an exclusion criteria; however, we believe it is necessary to exclude patients with cataracts to avoid interference with the analysis of outcomes. In addition to confounding baseline visual acuity, the gradual progression of cataracts may affect the efficacy of ERM surgery, and the prediction of postoperative visual acuity using SD-OCT findings. Some surgeons may successively conduct ERM peeling and phacoemulsification-vitrectomy; however, when this is done it makes it difficult to determine if improvement of vision is a result of ERM removal or the pseudophakia state [4]. Recent studies have shown that specific OCT parameters are linked to visual prognosis after ERM surgery [5, 6]. Kim et al. examined the correlations between individual OCT parameters and visual outcomes in pseudophakia ERM patients [7]. However, separate analysis of each OCT parameter without controlling for other factors and patient characteristics may not provide conclusive evidence as to which parameters are related to improved visual acuity. Thus, the purpose of this study was to identify which OCT parameters may be independent predictors of visual outcomes in pseudophakia patients with ERM who undergo surgery. Our review of the literature indicated that no prior study has combined all OCT parameters into a multivariate regression model focused on pseudophakia patients with ERM. We selected nine outer and inner retinal parameters that have been discussed in prior studies and can easily be evaluated by SD-OCT. The use of these parameters allows a simplified and efficient approach for each clinical evaluation. Methods Patients The records of patients who underwent successful surgery for iERM between July 2017 and July 2022 at Kaohsiung Veterans General Hospital, Taiwan, were retrospectively reviewed. ERM diagnosis and surgical procedures were performed by 2 retina surgeons. This study was approved by the hospital Review Board and Ethics Committee. Because of the retrospective nature of the study, the requirement of informed consent was waived. Inclusion criteria for the study were: (1) Diagnosis of ERM; (2) Diagnosis of pseudophakia before ERM surgery; (3) Treated with PPV, ERM peeling, and ILM peeling; (4) Attended follow-up visits for at least 3 months after surgery. Exclusion criteria were: (1) Eyes with high myopia defined by an axial length ≥ 26 mm, roughly equivalent to < − 6.0 D in our clinical setting; (2) Lamellar or full-thickness macular hole on preoperative SD-OCT examination; (3) History of prior surgery, intravitreal or subretinal injection, or laser treatment; (4) Laser treatment during ERM surgery; (5) History of, or concomitant ophthalmological disease that affects visual acuity or causes maculopathy, including glaucoma, uveitis, optic neuropathy, ocular trauma, diabetic retinopathy, choroidal neovascularization, retinal vein occlusion, and retinal tear or detachment. To minimize the influence of cataract surgery-related lenticular changes on retinal structure and visual recovery, all patients required a minimum interval of one month between cataract surgery and ERM surgery. Surgical treatment All surgeries were performed by 2 surgeons at Kaohsiung Veterans General Hospital, Taiwan. First, a 3-port 23- or 25-gauge microincision vitrectomy surgery was performed using the Alcon Constellation Vision System (Alcon. Laboratories, Inc., Fort Worth, TX, USA) under visualization with the OPMI Lumera 700 surgical microscope (Carl Zeiss Meditec AG, Jena, Germany). After the vitrectomy procedure, posterior vitreous detachment (PVD) was performed in patients without PVD. ERM peeling was then performed with intraocular forceps. Following removal of the ERM, indocyanine green (0.25% diluted in a 1-to-1 ratio with 20% glucose water) was applied to identify ILM for less than 10 s, and to then perform ILM peeling. Peeling of the ERM and the ILM was performed eccentrically starting 1.5-to-2-disc diameters from the fovea, and the membranes were gradually removed in layers. No intraoperative tamponade or alternative dyes were used during the surgery. Air-fluid exchange was conducted based on the surgeon’s discretion. Statistical analysis Descriptive statistics were summarized as mean ± standard deviation, and interquartile range (IQR). The best-corrected visual acuity (BCVA) was converted into the logarithm of the minimum angle of resolution (logMAR). Only eyes with complete baseline and postoperative (≥ 3-month) BCVA and OCT data were included in the regression analysis; no imputation was performed for missing values. The mean values of continuous variables were compared using the paired t-test. The McNamar test was used to compare categorical variables. Multivariate linear regression analysis was performed to identify the OCT parameters that were most associated with visual acuity. Values of P < 0.05 were regarded as statistically significant. Multivariable linear regression identified OCT predictors of postoperative BCVA. Effect sizes (R², adjusted R², standardized betas) were reported. Multicollinearity was checked using VIF (variance inflation factors) (VIF > 5 indicating concern), residual normality with Q–Q plots and the Shapiro–Wilk test, and homoscedasticity with residual–fitted plots and the Breusch–Pagan test. Models were revised when assumptions were not met. Statistical analyses were performed using SPSS version 26.0 software (SPSS Inc., Chicago, IL, USA) and R Project for Statistical Computing version 4.3.1 for Mac. In R, multivariable linear regression was performed using the base lm function in the stats package, and diagnostic plots were generated using ggplot2. Clinical assessment and SD-OCT imaging Every patient underwent a thorough standard ophthalmologic examination that included measurement of intraocular pressure, slit-lamp biomicroscopy, color fundus photography, and SD-OCT (Optovue, Inc., Fremont, CA) before and after surgery. The SD-OCT scans parameters were evaluated by a single experienced retinal specialist who was unaware of the patient history. Because no second grader was available, inter-rater reliability could not be assessed. Instead, the grader re-evaluated a random subset of scans after a 4-week washout period, and intra-rater agreement was measured using Cohen’s kappa. Central foveal thickness (CFT) was determined by measuring the mean thickness of the retina within a circular region of 1 mm diameter around the fovea. Central foveal bouquet (CB) abnormalities were evaluated according to previously described criteria. The cotton ball sign was defined as a localized, fuzzy, encircled hyper-reflective lesion in the outer foveal retina, representing focal traction on the photoreceptors and CB [8] (Fig. 1). Other CB alterations have been described in ERM, including foveolar detachment and pseudovitelliform changes [9]. In our cohort, we focused on the cotton ball sign as an easily recognizable manifestation of central foveal traction. Microcystic macular edema (MME), also known as inner nuclear layer microcysts (INLC), is defined as small hyporeflective lesions with no clear cyst wall, located in the inner nuclear layer, within or outside the perimacular rim [10] (Fig. 2). Fig. 1Imaging of a patient with the cotton ball sign. (A) Baseline color fundus imaging. (B) OCT before ERM surgery showed the presence of ERM and the cotton ball sign, characterized as a dome-shaped, hyperreflective area (white arrow). (C) OCT after ERM surgery showed disappearance of the cotton ball sign with restoration of the ellipsoid zone and retinal pigment epithelium Imaging of a patient with the cotton ball sign. (A) Baseline color fundus imaging. (B) OCT before ERM surgery showed the presence of ERM and the cotton ball sign, characterized as a dome-shaped, hyperreflective area (white arrow). (C) OCT after ERM surgery showed disappearance of the cotton ball sign with restoration of the ellipsoid zone and retinal pigment epithelium Fig. 2Imaging of a patient with MME. (A) Baseline color fundus imaging. (B) OCT before ERM surgery showed the presence of ERM, a cotton ball sign (white arrow), and a small, ellipsoidal-like area of MME (white arrowhead). (C) OCT after ERM surgery showed that the MME subsided with restoration of the ellipsoid zone at the fovea Imaging of a patient with MME. (A) Baseline color fundus imaging. (B) OCT before ERM surgery showed the presence of ERM, a cotton ball sign (white arrow), and a small, ellipsoidal-like area of MME (white arrowhead). (C) OCT after ERM surgery showed that the MME subsided with restoration of the ellipsoid zone at the fovea An ectopic inner foveal layer (EIFL), as described by Govetto et al., involves a continuous hyporeflective or hyperreflective band extending from the inner nuclear layer (INL) and inner plexiform layer (IPL) across the foveal region, and is visible on all OCT scans centered on the fovea [11]. When the horizontal extent in microns for which any boundaries between the ganglion cell–inner plexiform layer complex (GCIPL), inner nuclear layer (INL), and outer plexiform layer (OPL) could not be identified, it was defined as disorganization of retinal inner layers (DRIL) [12]. In addition, SD-OCT scans were evaluated for external limiting membrane (ELM) disruption and ellipsoid zone (EZ) disruption, based on the presence of a continuous hyperreflective line appearing in the macula. A newly proposed consensus definition for ERM foveoschisis based on OCT (Fig. 3) includes 2 mandatory criteria: a contractile ERM and foveoschisis at the level of Henle’s fiber layer. Optional criteria are the presence of microcystoid spaces in the inner nuclear layer, retinal thickening, and retinal wrinkling [13]. Additionally, a distinct post-operative appearance of dark arcuate striae along the course of the optic nerve fibers, resulting from small dimples in the regions of the retina where the ILM has been removed, was defined as a dissociated optic nerve fiber layer (DONFL) [14]. The aforementioned characteristics were reviewed on the pre- and post-operative OCT images of all patients. Fig. 3Imaging of a patient with ERM-foveoschisis. (A) Baseline color fundus imaging. (B) OCT before ERM surgery showed the presence of ERM, and ERM-foveoschisis (white arrowhead). (C) OCT after ERM surgery showed restoration of the foveal microstructure Imaging of a patient with ERM-foveoschisis. (A) Baseline color fundus imaging. (B) OCT before ERM surgery showed the presence of ERM, and ERM-foveoschisis (white arrowhead). (C) OCT after ERM surgery showed restoration of the foveal microstructure Results A total of 61 patients with ERM (61 pseudophakia eyes) including 29 males and 32 females with a mean age of 71 ± 3.5 years were included in the study. Characteristics of the patients are summarized in Table 1. Baseline logMAR BCVA was 0.49 ± 0.27 and the mean CFT was 439 ± 57.5 μm. After the surgery, the BCVA of 42 eyes improved, the BCVA of 9 eyes worsened, and the BCVA of 10 eyes remained the same. Table 1Demographic and clinical data of epiretinal membrane patients with pseudophakia Table 1Demographic and clinical data of epiretinal membrane patients with pseudophakia Demographic and clinical data of epiretinal membrane patients with pseudophakia BCVA and SD-OCT changes after ERM surgery are shown in Table 2. The CFT was significantly improved from 439 ± 57.5 μm to 358 ± 28.5 μm (P < 0.001), and the BCVA was improved from 0.49 ± 0.27 to 0.28 ± 0.21 (P < 0.001). OCT findings of cotton ball sign and DRIL were significantly decreased after surgery (both, P < 0.05). Table 2Summary of OCT characteristics before and after epiretinal membrane (ERM) peeling Table 2Summary of OCT characteristics before and after epiretinal membrane (ERM) peeling Summary of OCT characteristics before and after epiretinal membrane (ERM) peeling The results of the multivariate analyses are shown in Table 3, and the predictive values of the OCT parameters for improvement of BCVA are shown in Fig. 4. Controlling for other variables, eyes with ERM-foveoschisis and eyes with MME before the surgery showed an improvement in BCVA after surgery, but this improvement was relatively less compared to eyes without these conditions (regression estimate = -0.181 logMAR, 95% confidence interval [CI]: -0.38 – -0.04, P = 0.015; regression estimate = -0.183 logMAR, 95% CI: -0.48–0.17, P = 0.009, respectively). On the other hand, patients who had better baseline visual acuity tended to have more improvement in visual outcomes (regression estimate = 0.835 logMAR, 95% CI: 0.93–1.25, P < 0.001). Table 3Results of multivariable analysis of factors associated with the longitudional change in best-corrested visual acuity (BCVA) after epiretinal membrane (ERM) removal in affected eyes Table 3Results of multivariable analysis of factors associated with the longitudional change in best-corrested visual acuity (BCVA) after epiretinal membrane (ERM) removal in affected eyes Results of multivariable analysis of factors associated with the longitudional change in best-corrested visual acuity (BCVA) after epiretinal membrane (ERM) removal in affected eyes Fig. 4Predict value of BCVA (logMAR) improvement after ERM surgery in eyes with ERM-foveoschisis and MME, and the baseline visual acuity. Black lines in A and B, and gray background in C represent estimate under the 95% confidence interval Predict value of BCVA (logMAR) improvement after ERM surgery in eyes with ERM-foveoschisis and MME, and the baseline visual acuity. Black lines in A and B, and gray background in C represent estimate under the 95% confidence interval Discussion Epiretinal membrane surgery is widely regarded as one of the most frequently conducted procedures by retinal surgeons; however, a prevailing challenge for surgeons is identifying the right time for surgical intervention. Even with successful ERM removal, the visual outcome is not always satisfactory. Currently, SD-OCT is the gold standard for evaluating macular pathology and considerable research has been dedicated to the prediction of surgical and anatomical outcomes following ERM surgery based on SD-OCT findings. Raquel et al. show that SD-OCT features—including ERM stage, macular edema, and ellipsoid zone status—effectively predict visual outcomes after iERM surgery [3]. Attention was initially focused on outer retinal findings in ERM due to deterioration of the outer retinal layers, and it was suggested that persistent impairment of photoreceptors thereby was correlated with an unfavorable visual prognosis [15]. However, since traction on the inner retinal layer has been shown to be associated with electrophysiological dysfunction and structural changes in patients with ERM, more attention has been given to inner retinal findings [16]. Due to a lack of consensus, outer and inner retina layer SD-OCT findings are currently used for evaluation before ERM surgery and to predict outcomes [17]. Our findings showed a significant improvement in CFT and BCVA following ERM removal. The result is likely because early membrane removal prevents further progression of photoreceptor damage in patients with ERM [18]. Other studies have examined the relation between the CFT and visual outcomes in patients with ERM; however, results are inconsistent. Some studies have shown a correlation between a postoperative decrease in CFT and an increase in BCVA [18, 19]. On the other hand, 2 multivariate analyses conducted by Hosoda et al. and Kim et al. found that CFT was not a reliable predictor for visual outcome after ERM surgery [20, 21]. In our multivariate linear regression model, no significant association between the degree of BCVA improvement and the degree of CFT resolution was found. The varying results may indicate that CFT is just one of several variables affecting visual acuity, and the relations are not fully understood [20]. Our results showed that DRIL was significantly reduced after surgery, and similar results were reported in prior studies [22, 23]. Prior research has indicated that severe DRIL at baseline, defined as GCIPL, INL, and OPL all being indistinguishable, is correlated with limited improvement in visual acuity after surgery [17, 23]. A multicenter study by Dinah et al. concluded that DRIL grading predicts surgical outcomes in idiopathic ERM, with severe DRIL indicating limited benefit from surgery [24]. The study by Karasavvidou et al. in which a linear regression model was developed did not find a correlation between mild DRIL and BCVA outcome after surgery [17]. Our results showed no significant link between DRIL and poor visual outcomes, which we speculate is due to lower baseline DRIL severity in our patients. The cotton-ball sign is a round or diffuse highly reflective region at the center of the fovea caused by continuous inward traction from the ERM. Our results revealed a significant decrease in the cotton ball sign after ERM removal, which showed the release of inward traction on the retinal layer after surgery. Tsunoda et al. [25] pointed out that the disappearance of the cotton ball sign after ERM surgery does not necessarily indicate an increase or decrease in visual acuity after surgery. González-Saldivar et al. considered the cotton ball sign to be a possible prognostic factor of poor visual acuity after ERM surgery, but not statistically significant [8]. Our multivariable analysis did not find a significant relation between the cotton ball sign and an improvement in BCVA after ERM surgery. As reported in prior publications, our results showed that better baseline visual acuity was predictive of a better improvement in postoperative BCVA [16, 26]. We speculated that in most of our patients, ERM is present for a relatively short period of time, causing less functional disturbance. Clinically, these findings support the concept that eyes with less severe macular disruption at presentation possess a more favorable anatomical substrate for postoperative recovery, even though some eyes with poorer baseline BCVA may exhibit greater numerical improvements in logMAR. Recently, a new OCT-based definition of idiopathic epiretinal membrane foveoschisis (ERM-foveoschisis) was proposed by an international panel of vitreoretinal experts [13]. In our study, although not statistically significant, ERM-foveoschisis was shown to be resolved postoperatively (P = 0.063), which is consistent with the results of previous studies [27, 28]. A reasonable explanation is that release of traction allows the axons of Müller cells to return to their original morphology, which restores their function in the visual transmission pathway [29]. However, a persistence of intraretinal hyporeflective spaces in the foveal region was noted by Figueroa et al. in the early postoperative period, although they gradually subsided during follow-up period [29]. The delayed healing progress observed in patients with ERM-foveoschisis might indicate chronic inflammation and damage. This could explain our findings that patients with preoperative ERM-foveoschisis showed comparatively less visual improvement. MME is another predictor of postoperative visual acuity after ERM surgery proposed recently. Lee et al. noted less improvement in BCVA after ERM surgery in patients with preoperative MME during a follow-up of 3 months [30]. Yang et al. observed similar results in a study with a follow-up period of 24 months [31]. Our results showed that the postoperative improvement of BCVA was less in patients with preoperative MME than patients without MME. Müller cell damage and retrograde trans-synaptic degeneration to bipolar cells are the two most prevalent hypotheses for this finding, and it is speculated that breakdown of the integrity of the blood–retinal barrier and dysfunction of bipolar cells affect intraretinal fluid hemostasis and visual pathway transmission [30–32]. These findings suggest that MME might be associated with chronic inflammation, leading to challenges in achieving greater visual improvement after surgery. Recent evidence also shows that ERM-related MME is more closely linked to Müller cell disruption and surgical stress rather than retrograde maculopathy, further supporting a structural rather than neurodegenerative basis for its impact on postoperative visual recovery [9]. Foveoschisis and MME were associated with reduced postoperative visual gains, likely reflecting Müller cell dysfunction and altered gliosis. Govetto et al. showed that central foveal bouquet abnormalities, including cotton ball sign, foveolar detachment, and vitelliform lesions, form a progressive spectrum mediated by Müller cells, with more severe changes linked to worse vision [33]. DONFL is not seen in our cases on preoperative OCT but appears in 2 cases postoperatively. Recent evidence indicates that ERMs with prominent inner retinal thickening or associated macular holes often have a stickier ILM during peeling, which strongly predicts the development of DONFL and is associated with greater inner retinal thinning and slower functional recovery [34]. This supports careful ILM handling in eyes with these high-risk features. Because the 9 eyes with worsened BCVA showed heterogeneous baseline OCT characteristics and the sample size was small, a formal subgroup analysis was not performed. Further work in larger cohorts is warranted. The mean follow-up in our study was 6 months, with a minimum of 3 months. While most visual recovery after ERM surgery occurs within 3–6 months, some studies show continued improvement up to 12–24 months. Therefore, our OCT-based model reflects early to mid-term outcomes, and the predictive value of certain OCT markers may differ with longer follow-up. Recent Artificial intelligence (AI)-based (artificial intelligence) work also highlights the importance of specific preoperative OCT features, including outer nuclear layer (ONL), combined photoreceptor and retinal pigment epithelium complex (PR + RPE), and inner retinal thickness, with outer retinal thinning and intraretinal fluid associated with poorer recovery [35]. These observations support the relevance of structural OCT biomarkers, and larger, longer-term studies are needed to validate them. The use of a single-grader OCT protocol in our study limited the ability to assess inter-rater reliability. To address this, the grader repeated a subset of scans after a washout period, and intra-rater reliability was evaluated using Cohen’s kappa. Although the limitations of our study include its retrospective nature and relatively small sample size, there are several highlights in our study. All eyes in our study were pseudophakic, eliminating the influence of postoperative lens changes on functional assessment. Previous studies, including Momenaei et al., show that mild ERM rarely progresses after cataract surgery, while higher-stage ERM carries greater risk [36]. By studying only pseudophakic eyes, we removed cataract surgery as a variable, allowing clearer evaluation of ERM-related prognostic and morphological factors. Additionally, ERM-foveoschisis and MME are relatively newly defined OCT parameters. To the best of our knowledge, this is the first multivariable analysis that suggests these two parameters are indicators of limited visual outcomes in patients with ERM and pseudophakia after surgery. Conclusions In conclusion, in this study we determined prognostic factors for visual outcomes of ERM surgery by eliminating confounding factors such as BCVA improvement following combined cataract surgery or BCVA deterioration due to cataract progression. Multiple linear regression analyses showed that better baseline visual acuity is correlated with better visual improvement after surgery (P < 0.01). ERM-foveoschisis (P = 0.015) and MME (P = 0.009) were significantly correlated with less improvement of visual acuity. These results help to predict the visual prognosis of ERM patients receiving ERM peeling surgery.