Abstract
Keywords
Introduction
Epiretinal membrane (ERM) is a common vitreoretinal interface disorder affecting approximately 10% of adults and is often associated with visual distortion and reduced visual acuity (VA).1–3 Although frequently idiopathic, ERM may also arise secondary to retinal vascular disease, vitreous hemorrhage, intraocular inflammation, and trauma, among other etiologies.1,2 ERM also represents a commonly recognized sequelae of retinal tears and detachments, developing in up to 40% of eyes following rhegmatogenous retinal detachment (RRD) repair.2,4,5 In addition to glial cells, fibroblasts, and myofibroblasts present in the composition of ERM,6,7 retinal pigment epithelium (RPE) cells also contribute to ERM development following RRD, likely through migration via retinal breaks.7,8
With advancements in retinal imaging modalities such as spectral-domain optical coherence tomography (SD-OCT), ERMs are not only detected more frequently, but novel structural features have also been increasingly identified. However, the clinical significance of many OCT biomarkers remains unclear. Furthermore, comparative studies evaluating imaging differences between idiopathic and secondary ERM are limited in scope and detail.9,10
This study aims to elucidate the distinguishing OCT characteristics of ERM secondary to retinal detachment compared with idiopathic ERM. Improved characterization of the differing morphologies may provide insights into the etiology and the visual and surgical prognosis of eyes with secondary ERM formation following RRD.
Methods
This retrospective study was approved by the Institutional Review Board at Vanderbilt University Medical Center and was conducted in accordance with the tenets of the Declaration of Helsinki. Data were collected, anonymized, and securely stored before analysis.
Two cohorts were compared: (1) eyes with ERM following RRD and (2) eyes with idiopathic ERM. The RRD cohort included eyes that underwent pars plana vitrectomy (PPV) with membrane peeling (MP) for ERM after prior RRD. Eligible patients were identified by chart review using International Classification of Diseases (ICD-9 and ICD-10) codes for ERM, along with confirmed documentation of retinal tear or RRD preceding the development of ERM. All patients underwent PPV with ERM peeling, with or without internal limiting membrane peeling (Current Procedural Terminology codes 67041 and 67042). The control cohort consisted of patients with idiopathic ERM who underwent similar surgical management. Inclusion criteria were age 18 years or older and availability of a preoperative OCT scan (Cirrus; Carl Zeiss Meditec, or Spectralis; Heidelberg Engineering) obtained within 1 month before surgery. Exclusion criteria were recurrent retinal detachment requiring more than 2 surgeries and any prior or concurrent condition associated with secondary ERM formation, including prior vitrectomy for reasons other than RRD, high myopia (axial length >26.5 mm), uveitis, severe nonproliferative or proliferative diabetic retinopathy, diabetic macular edema, retinal vein/artery occlusion, history of globe rupture, history of endophthalmitis, or significant vitreomacular traction. Patients in the idiopathic ERM group were excluded if any of the above conditions were present.
Only patients who underwent surgery were included in each cohort to ensure the selection of sufficiently symptomatic ERMs. Furthermore, PPV allowed for more comprehensive intraoperative exploration of the peripheral retina and the exclusion of previously undetected retinal breaks in patients presumed to have idiopathic ERM. Individuals in whom retinal breaks were identified intraoperatively were excluded from this analysis.
Demographic data, as well as preoperative and postoperative VA, were recorded for all patients. In the RRD group, additional data regarding characteristics of the retinal detachment and the type of repair were also recorded.
Optical Coherence Tomography Grading
The following OCT biomarkers were assessed: ERM stage, central subfield thickness (CST), cystic changes, lamellar hole, a gap between the ERM and the retinal surface, diffuse or focal retinal nerve fiber layer (RNFL) schisis, disorganization of the retinal inner layers (DRIL) grade, ectopic inner foveal layer, foveal herniation, and ellipsoid zone (EZ) disruption. Each parameter was evaluated on both horizontal and vertical foveal cross-sectional scans. All OCT biomarkers were graded as present or absent, except for ERM stage, CST, and DRIL, which were graded according to predefined criteria described below. Grading was independently performed by 2 graders (S.B., A.P.F.); discrepancies were resolved by consensus.
ERM staging was performed according to the classification system proposed by Govetto et al, which incorporates foveal depression, presence of ectopic inner foveal layers, and disruption of retinal layer architecture. EZ disruption was defined as complete loss of the EZ measuring ≥4 μm (equivalent to 1 pixel) on both horizontal and vertical OCT cross-sections, adopted from a study by Coulibaly et al. 11 The OCT scans were required to have adequate image quality with a clearly visible EZ band; attenuation due to apparent shadowing or imaging artifact was not considered true disruption, and discontinuities <4 μm were excluded.
RNFL schisis was described as focal (≤1 disc diameter) or diffuse (>1 disc diameter), following the criteria described by Hussnain et al. 12 DRIL was graded using the system reported by Zur et al 13 in the DREAM study, with 1 point assigned for each of the following features: irregular boundary of the ganglion cell–inner plexiform layer and inner nuclear layer (GCIPL–INL), indistinguishable GCIPL–INL boundary, irregular boundary of the INL and outer plexiform layer (INL–OPL), and indistinguishable INL–OPL boundary. The gap between the ERM and the retinal surface was assessed on horizontal and vertical foveal OCT scans and categorized as absent or present, as described by Murase et al. 14 CST values obtained from Zeiss Cirrus scans were converted to Spectralis-equivalent values using the conversion formula published by Sun et al. 15
Statistical Analysis
The primary outcome was the difference in the frequency of OCT biomarkers between the RRD and idiopathic ERM groups. Categorical variables were compared using the Pearson chi-square test or Fisher exact test, as appropriate. Statistical analysis was performed using R software version 4.5.0 within RStudio (version 2024.12.1+563, Posit). A P value of <.05 was considered statistically significant. Continuous variables were compared using the Wilcoxon rank-sum test.
Results
During the study period from January 2018 to January 2025, 189 patient charts with ERM after RRD were reviewed. Of these, 154 were excluded for the following reasons: 26 eyes did not undergo surgery, 28 eyes lacked OCT imaging in the specified time frame or OCT was unavailable for analysis (eg, performed at satellite clinics), 58 eyes had concurrent conditions as previously defined, and 42 eyes had ICD-9/-10 and Current Procedural Terminology codes indicating retinal detachment and PPV/MP in the contralateral eye.
The RRD cohort consisted of 35 eyes from 35 patients, while the idiopathic ERM cohort consisted of 74 eyes from 74 patients. Demographic features were similar between groups. The mean age at ERM diagnosis was 67.8 years in the RRD cohort and 73.8 years in the idiopathic group (P = .07). In the RRD cohort, 23 patients (65.7%) were male, compared with 41 patients (55.4%) in the idiopathic group (P = .42). Most patients (64.2%) in both groups were pseudophakic: 24 eyes (68.6%) in the RRD group and 46 eyes (62.1%) in the idiopathic group (P = .66).
For eyes with post-RRD ERM, the characteristics of the retinal detachment and treatment methods are summarized in Table 1. The median interval between retinal detachment repair and ERM surgery was 266 days (interquartile range, 455 days). No significant intraoperative or postoperative complications were noted, except for 1 patient in the RRD cohort who developed a recurrent retinal detachment following MP and required reoperation.
Retinal Detachment Type and Treatment Method (RRD Cohort, n = 35).
Abbreviations: PPV, pars plana vitrectomy; RRD, rhegmatogenous retinal detachment.
Mean VA was significantly worse in the RRD cohort compared with the idiopathic groups, both preoperatively (P = .001) and at 6 months postoperatively (P = .002). The mean logMAR VA in the RRD cohort was 0.71 (Snellen equivalent, 20/100) preoperatively and 0.63 (Snellen equivalent, 20/80) postoperatively. The mean logMAR VA in the idiopathic group was 0.44 (20/50) preoperatively and 0.32 (20/40) postoperatively.
ERMs were of higher stage in the RRD group, with 80% classified as stage 3 or 4 compared with 61% in the idiopathic cohort (P = .048). DRIL was both more prevalent and more severe in the RRD group (P = .02). Grade 3 or 4 DRIL was observed in 42% of eyes in the RRD cohort, compared with 15% in the idiopathic cohort. Conversely, mild or absent DRIL was more common in the idiopathic group, with 38% demonstrating no DRIL and 47% demonstrating grade 1 to 2 DRIL, compared with 26% and 31%, respectively, in the RRD group.
Foveal herniation was more common in the RRD group, observed in 14 eyes (40%) compared with only 10 eyes (14%) in the idiopathic group (P = .009). While ectopic inner foveal layer and RNFL schisis did not achieve statistical significance, both were more common in the post-RRD ERM group. Ectopic inner foveal layer was present in 30 eyes (85%) in the RRD cohort compared with 45 eyes (61%) in the idiopathic cohort (P = .07). Similarly, RNFL schisis was identified in 19 eyes (54%) in the RRD group and 23 eyes (31%) in the idiopathic group (P = .07).
The only OCT biomarker significantly more prevalent in the idiopathic ERM cohort compared with the post-RRD ERM cohort was a gap between the ERM and the retinal surface, observed in 46 eyes (62%) and 6 eyes (17%), respectively (P < .001). Figure 1 demonstrates select OCT images from both cohorts, highlighting these differing features.
CST, cystic changes, lamellar hole, and EZ disruption did not differ significantly between the cohorts. A summary of the frequency of OCT biomarkers in the RRD and idiopathic groups is shown in Table 2.
Frequency of Optical Coherence Tomography Biomarkers in RRD and Idiopathic Cohorts.
Abbreviations: DRIL, disorganization of the retinal inner layers; EIFL, ectopic inner foveal layer; ERM, epiretinal membrane; OCT, optical coherence tomography; RNFL, retinal nerve fiber layer; RRD, rhegmatogenous retinal detachment.
Representative optical coherence tomography (OCT) images of epiretinal membrane (ERM) features. (A) Stage 1 idiopathic ERM. (B) Stage 4 post-rhegmatogenous retinal detachment (RRD) ERM with foveal herniation (arrowheads) and retinal nerve fiber layer (RNFL) schisis (arrow). (C) Idiopathic ERM with grade 1 disorganization of the retinal inner layers (DRIL) (bracket). (D) Post-RRD ERM with grade 4 DRIL (bracket), foveal herniation (arrow), and RNFL schisis (arrowhead). (E) Idiopathic ERM demonstrating a gap between the ERM and the retinal surface (asterisks). (F) Post-RRD ERM without a gap between the ERM and the retinal surface.
Conclusions
The current study comparing OCT biomarkers of ERMs secondary to RRD with idiopathic ERM reveals several key distinguishing features. ERMs following RRD were of a higher stage, showed more severe DRIL, and showed an increased prevalence of RNFL schisis. Conversely, idiopathic ERMs showed a broader distribution of stages and a higher prevalence of a gap between the ERM and the retinal surface, which may facilitate MP. These characteristic differences between post-RRD and idiopathic ERMs are clinically relevant, given their potential implications for visual outcomes and surgical complexity, as discussed below.
The higher-grade ERMs and increased frequency of ectopic inner foveal layer observed in the post-RRD group compared with the idiopathic ERM group are consistent with findings reported by Israilevich et al. 16 These differences may reflect the contribution of RPE cells, as well as inflammatory mediators and cytokines, to the formation of ERM after RRD.17,18 The greater severity of ERM and increased presence of ectopic inner foveal layer in the RRD cohort likely contribute to worse VA and may limit postoperative visual potential following MP compared with idiopathic ERM.
DRIL, a relatively novel OCT biomarker that has been more heavily studied in diabetic macular edema and is associated with visual prognosis,19,20 was more prevalent in the RRD-associated ERM group compared with idiopathic ERM. In addition, DRIL was more severe on preoperative OCT in the post-RRD cohort. DRIL has been proposed to reflect retinal stress, inflammation, and possible ischemia, as supported by its association with greater severity of diabetic retinopathy and poorer VA.21–23 This association may partially explain the worse visual outcomes in the RRD group, beyond differences in baseline VA, and may be useful in counseling patients regarding their postoperative visual prognosis.
Although not statistically significant, RNFL schisis tended to be more common in post-RRD ERM. Sigler et al 9 described fingerlike projections extending from the ERM, a finding that likely corresponds to the current designation of RNFL schisis in eyes with ERM secondary to retinal breaks. Their findings, along with our current findings, suggest that the presence of these structural changes may be associated with poorer visual prognosis in this population.
The current analysis demonstrated that a gap between the ERM and the inner retinal surface was more frequently observed on OCT in idiopathic ERM compared with post-RRD ERM. This gap has been proposed as an indicator of the degree of retinal contracture.14,24 The slowly developing nature of idiopathic ERM may allow more time for retinal contracture compared with the more rapid development of ERM after RRD. This aligns with the findings of Mori et al, 10 who reported that primary ERMs tend to be more globally adherent, whereas secondary ERMs exhibit more focal adherence. The presence of a gap between the ERM and the retinal surface may aid in ERM peeling by providing a potential plane of separation between the retinal surface and the ERM. Conversely, its relative absence in post-RRD ERM may suggest increased surgical difficulty and potentially worse surgical outcomes.25,26
This study has several strengths. It includes a large number of OCT biomarkers, representing, to our knowledge, a broader range than that evaluated in previously published studies. The inclusion of both established and emerging biomarkers further contributes to the limited body of literature on secondary ERM. In addition, the inclusion of visual outcomes enhances the clinical relevance of this work and provides information that may assist surgeons in preoperative planning and patient counseling regarding visual prognosis.
Several limitations of this study must also be acknowledged. Its retrospective design introduces the potential for selection bias, particularly toward more severe ERMs warranting surgical intervention. In addition, the relatively small sample size may limit generalizability and reduce statistical power to detect certain differences between groups.
Certain variables were assessed using binary or categorical classifications (eg, presence vs absence of a gap between the ERM and the retinal surface; or diffuse vs focal RNFL schisis) rather than quantitative measures, which may limit the ability to detect more subtle distinctions between cohorts. Furthermore, the duration of ERM was challenging to ascertain in the idiopathic group, as many patients were referred with a preexisting diagnosis. As a result, the impact of disease duration on OCT morphology and visual outcomes could not be studied.
OCT biomarker analysis demonstrates that ERMs secondary to RRD are characterized by higher stage, increased prevalence of ectopic inner foveal layer and foveal herniation, and more severe DRIL and RNFL schisis. Collectively, these features are associated with a less favorable visual prognosis compared with idiopathic ERM.
In contrast, a gap between the ERM and the retinal surface was more frequently observed in idiopathic ERM and may indicate a more favorable surgical outcome, as the membrane may be less adherent to the retinal surface. These findings suggest that earlier surgical intervention for post-RRD ERM may be worth considering; however, changes to clinical practice should be guided by future prospective studies.
Recognition of these differences is important for surgical planning and for counseling patients regarding expected visual outcomes.
Footnotes
Ethical Considerations
This retrospective study was approved by the Institutional Review Board at Vanderbilt University Medical Center and was conducted in accordance with the tenets of the Declaration of Helsinki. The collection and evaluation of all protected patient health information was performed in a Health Insurance Portability and Accountability Act-compliant manner.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported in part by an unrestricted grant from Research to Prevent Blindness to the Vanderbilt Eye Institute. The funding organization had no role in the design or conduct of the study.
Declaration of Conflicting Interests
The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Dr. Finn is a consultant with AbbVie and Genentech; serves on the advisory boards of Apellis Pharmaceuticals, Bausch + Lomb, EyePoint Pharmaceuticals, and Iveric Bio; and is a consultant with and serves on the advisory board of Bausch + Lomb. None of the other authors declared potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Statement of Informed Consent
Informed consent was waived owing to the retrospective nature of the study. All data were collected, de-identified, and securely stored.