Epiretinal membranes in multiple sclerosis: A window into glial pathobiology?

Optical coherence tomography (OCT) provides high-resolution, reproducible images of the optic nerve and retina which provide quantitative measurements that have been established as biomarkers for multiple sclerosis (MS). ¹ In MS, retinal changes, including peripapillary retinal nerve fiber layer (RNFL) and a composite measure of ganglion cell and inner plexiform layer (GCIPL) thinning, have been found to correlate with global central nervous system pathology.^2,3 Epiretinal membranes (ERMs) are common age-dependent fibrocellular proliferations which may include retinal glial cells, hyalocytes, fibroblasts, and extracellular matrix (prevalence 7%–34%). ⁴ Lin et al. present the first systematic investigation of ERMs in patients with MS, ⁵ raising the intriguing possibility that these retinal findings may serve as a window into broader glial pathobiology.

In the current study, Lin and colleagues completed a cross-sectional evaluation of a robust number of participants, including 1120 patients with MS and 144 healthy controls with an extended battery of disability measures including measures of visual function, global disability, walking speed, dexterity, and processing speed, as well as serum neurofilament light chain levels. In this cohort, they found a numerically higher but statistically non-significant difference in the prevalence of ERMs between the patients with MS (18.4%) as compared to the healthy controls (11.8%; p = 0.051). When comparing patients with MS and ERMs (n = 206) to age- and sex-matched patients with MS without ERMs, they found that patients with ERMs had increased disability including elevated expanded disability status scale (EDSS) and reduced low contrast visual acuity (LCVA), as well as reduced retinal thickness measurements as assessed by OCT. Furthermore, ERMs were classified by severity with strong agreement between raters, and the higher ERM severity was associated with higher EDSS scores. The authors propose that MS pathophysiology may be contributing directly to ERM formation and suggest that blood retinal barrier breakdown and Müller glia activation may be potential mechanisms for these findings.

The pathology under study, however, introduces a fundamental challenge to the reliability of the OCT measurements used to characterize it. ⁶ ERMs are inherently known to cause OCT segmentation errors, given that ERMs can cause mechanical traction and changes in the macular contour. Prior studies have found up to a 35.8% segmentation error rate in patients with ERMs and misidentification of RNFL and inner limiting membrane (ILM) boundaries.^7,8 Mechanical disruption from ERMs can result in potential loss of the foveal depression, disruption of the inner nuclear layers, and changes to photoreceptor arrangements which have also been shown to reduce interscan reliability and alter thickness measurements. ⁹ Specifically, ERMs can lead to thickening of the temporal peripapillary RNFL and also thinning of the GCIPL. ⁹ These inherent ERM-derived changes make it more challenging to parse MS-driven neurodegeneration from ERM-driven mechanical distortion in a cross-sectional study. The authors did review studies for segmentation accuracy; however, it is possible that the presence of ERMs may have impacted segmentation despite rigorous control. In addition, it remains unclear whether patients with a history of prior ERM peeling surgery were included in the ERM+ group. Vitrectomy with membrane peeling causes structural retinal changes independent of MS pathology, including RNFL and GCIPL thinning, which would further confound OCT comparisons between groups. It is intriguing to consider that local glial responses in the retina might parallel macrogliosis and microgliosis in the brain, particularly in light of the association of ERM and increased global disability; however, additional longitudinal studies will be needed to further assess this possibility.

The confounding effects of ERMs extend beyond OCT segmentation to the study’s visual and prevalence findings as well. As the authors discuss, the same macular distortion that leads to alterations in OCT segmentation and thickness measurements can lead to an inherent decrease in visual acuity and metamorphopsia. Further longitudinal studies will be needed to determine whether the reduction in visual acuity in patients with ERMs is related to ERM-independent MS pathophysiology. The inclusion of a novel Stage 0 ERM category, defined as minimally visible ERMs spanning fewer than three B-scans, a category absent from the original Govetto staging system, may also inflate the prevalence of ERMs in this study.

Overall, the association of ERMs with global disability and serum neurofilament light chain is hypothesis-generating and cannot be explained by local ERM effects alone. The correlation with OCT thickness and LCVA findings, however, must be interpreted cautiously, as ERMs introduce both mechanical retinal distortion and downstream segmentation artifacts that are difficult to fully exclude even with rigorous quality control. Future studies are needed with a longitudinal design to corroborate these findings in MS. Such studies should also consider the use of artificial intelligence (AI)-corrected segmentation validated for ERM eyes, serum glial fibrillary acidic protein (GFAP) to probe the astrogliosis hypothesis, enrichment for progressive MS patients where glial pathology is most relevant, and exclusion of eyes with prior vitreoretinal surgery. Lin and colleagues should be commended for opening a genuinely novel line of inquiry that, with longitudinal validation, could meaningfully expand the toolkit for understanding disease pathophysiology and monitoring disease burden in MS.