Surgeons Remove Intact Retinal Glial Cells More Often in Advanced Epiretinal Membrane

When surgeons peel the internal limiting membrane (ILM) from the retina during vitreoretinal surgery, they may be removing more than just the membrane itself. A new retrospective clinicopathological study from University Hospital Ulm in Germany reports that intact retinal glial cells, with ultrastructural features consistent with Müller cells, were found on the retinal side of surgically excised ILM specimens only in eyes with idiopathic epiretinal membrane (iERM) and never in eyes with full-thickness macular hole (FTMH). The three eyes in which such cells were identified all had Stage 3 epiretinal membranes, ectopic inner foveal layers, and central bouquet abnormalities on optical coherence tomography (OCT), along with significantly greater central macular thickness than eyes without the cells. The findings, published in Eye and Vision, suggest that the strength and complexity of the interaction between the retina and the ILM differ between these two common vitreomacular interface disorders.

The study included 42 eyes of 42 patients who underwent 23- or 25-gauge pars plana vitrectomy with epiretinal membrane and ILM peeling at the Department of Ophthalmology, University Hospital Ulm, between April 2024 and May 2025. Twenty-two eyes had iERM and 20 had FTMH. All excised specimens were prepared as flat mounts for light microscopy and then processed into serial ultrathin sections for transmission electron microscopy (TEM). Clinical data and OCT findings were analyzed alongside the ultrastructural results, allowing the researchers to correlate what surgeons removed with what imaging had shown before and after surgery. The Ethics Committee of Ulm University approved the surgical removal and histopathological analysis of the patient specimens.

Transmission electron microscopy confirmed the presence of the ILM in all 42 specimens. The membrane appeared as a thin, acellular, lamellar structure characteristically rolled toward the vitreous side. Well-defined ILM thinning was observed in 14 of the 42 specimens, affecting 7 of the 22 iERM specimens and 7 of the 20 FTMH specimens. The thinning was confined to the retinal side of the membrane, while the vitreal surface remained smooth, and the thinned regions were clearly distinguishable from preparation artifacts. Notably, TEM identified no ILM breakthroughs of the kind described as ILM pores in earlier work.

Retinal cell fragments, defined as cellular debris without a visible nucleus, were far more common than intact cells. Fragments were detected in 39 of 42 specimens overall, or 93 percent, and occurred in both disease groups: in 21 of 22 iERM specimens and 18 of 20 FTMH specimens. The fragments were categorized by diameter as small, medium, or large, and most were round and frequently located within regions of ILM thinning. In specimens containing large fragments, these appeared to protrude from the retinal side into the thinned ILM region. Because fragments were present in both conditions, the authors conclude that ILM peeling cannot be considered free of effect on the underlying retina in either disease.

The most striking difference between the two groups involved intact, nucleated cells. Cellular profiles adherent to the retinal surface of the peeled ILM were identified in 3 of 22 iERM eyes, or 14 percent, but in no FTMH eye. These cells extended into the ILM and exhibited ultrastructural characteristics of glial cells. One profile showed a prominent, irregularly shaped electron-dense nucleus with surrounding cytoplasm containing membrane-bound organelles including mitochondria. Another had an elongated, irregular nucleus with heterogeneous electron density and condensed chromatin. A third contained a large oval-to-irregular nucleus with heterogeneous chromatin and a distinct nuclear envelope. The close association of these profiles with the ILM and their processes extending into the membrane were morphologically consistent with a Müller cell-like phenotype, and the cells lacked the neuritic processes and polarization typical of retinal neurons.

The location of these cells was not random. Retinal cells were frequently found within well-defined regions of ILM thinning, an association that was statistically significant. The three eyes with iERM and retinal cells had a mean central macular thickness of 619 plus or minus 57 micrometers, significantly greater than that of iERM eyes without retinal cells. All three had Stage 3 epiretinal membranes with ectopic inner foveal layers and central bouquet abnormalities, features regarded in the literature as markers of chronicity and progressive tractional remodeling. Mean best-corrected visual acuity in these three eyes was 0.60 plus or minus 0.36 logMAR at baseline and did not differ significantly from that of eyes without retinal cells. The operating surgeon rated ILM peeling as difficult in two of the three eyes.

The imaging analysis also revealed broader patterns in each disease group. In the iERM group, Stage 3 membranes occurred in 12 eyes and Stage 2 in 10, and higher membrane stage correlated significantly with worse preoperative visual acuity. Mean preoperative central macular thickness was 510.7 plus or minus 81.9 micrometers and decreased significantly over follow-up. In the FTMH group, the mean baseline minimum linear diameter of the hole was 395.4 plus or minus 200.9 micrometers, and larger hole size was significantly associated with worse preoperative visual acuity. The macular hole was closed in all eyes with available follow-up data. Among FTMH eyes, those with large retinal cell fragments had significantly worse postoperative visual acuity than those with small or medium fragments, although OCT findings did not differ significantly by fragment size overall.

The authors interpret the difference between the two conditions as reflecting distinct pathophysiological mechanisms at the vitreoretinal interface. Idiopathic epiretinal membrane is characterized by chronic tangential traction from a firmly adherent epiretinal membrane and ILM complex, which may impose sustained mechanical stress on the underlying retinal surface and promote close adhesion between the ILM and superficial retinal cellular elements, predisposing retinal cells to displacement or removal during peeling. Full-thickness macular hole formation, by contrast, is driven primarily by focal anteroposterior traction with a lesser tangential component, and because many FTMH cases lack an epiretinal membrane, a tightly adherent membrane complex is less likely to develop. The researchers also note that mechanical stress from vitreous traction may induce Müller cell contraction and migration, and that ILM separation may be more difficult in regions of Müller cell migration across the membrane.

The study has limitations that the authors describe explicitly. The relatively small number of specimens may limit generalizability, and because of the limited sample size and small subgroup numbers, the analyses were restricted to univariate comparisons without multivariable modeling, which the authors state could produce overfitting and unstable estimates. TEM itself has inherent sampling limitations, since only selected ultrathin sections can be examined, so the absence of retinal cells from the analyzed sections does not definitively exclude their presence elsewhere in the tissue, and the observed associations should be considered exploratory. Cytological and immunocytochemical analyses were not performed, partly because no single immunocytochemical marker specifically identifies Müller cells and because specimens were fixed in high-concentration glutaraldehyde to optimize ultrastructural preservation, which precluded later immunophenotypic analyses. Incomplete 6-month follow-up data and the retrospective design, which prevented precise control of confounders such as disease duration and traction chronicity, are additional caveats.

From a surgical perspective, the findings underscore that the tissue encountered during ILM peeling differs between diseases. The authors suggest that advanced OCT-based evaluation of the ILM and the retinal-ILM interface may help predict adherent retinal cells and guide surgical planning, and that intraoperative OCT could provide real-time guidance to minimize retinal interaction during peeling. They caution, however, that whether these differences should alter surgical strategy, and whether modified peeling techniques affect the extent of cellular removal, remain unclear and require further study. Larger longitudinal studies with functional outcome measures, including postoperative visual acuity and retinal sensitivity, are needed to clarify the clinical significance of the retinal cell adhesion patterns observed in this exploratory analysis.

Subject of Research: Ultrastructural analysis of retinal cells and fragments removed with the internal limiting membrane during vitreomacular interface surgery

Article Title: Retinal glial cell removing during internal limiting membrane peeling is associated with higher-stage idiopathic epiretinal membrane

Article References: Vogt, D., Schmitz, C., Dammak, A., Govetto, A., Schumann, R. G., Romano, M. R., Parlak, M., & Wolf, A. (2026). Retinal glial cell removing during internal limiting membrane peeling is associated with higher-stage idiopathic epiretinal membrane. Eye and Vision, 13(1), Article 42. https://doi.org/10.1186/s40662-026-00514-z

Image Credits: AI Generated

DOI: 10.1186/s40662-026-00514-z

Keywords: epiretinal membrane, internal limiting membrane, macular hole, Müller cells, transmission electron microscopy, optical coherence tomography, vitreoretinal surgery, retinal glial cells, ILM peeling, vitreomacular interface, pars plana vitrectomy, retina