The epiretinal membrane (ERM) is a fibrocellular overgrowth at the interface between the vitreous and the retina, situated above the inner limiting membrane (ILM) (1). The incidence of this condition rises with age and has been reported to reach up to 28.9% in the United States population (2). This fibrotic process is driven by the accumulation of collagen and the transformation of retinal Müller cells, retinal pigmented epithelium (RPE) cells, and hyalocytes into myofibroblasts (3, 4). Ultimately, this transformation leads to contraction and subsequent puckering of the macular area. As a result, this membrane exerts traction on the retina, ultimately causing metamorphopsias and reduced visual acuity (5). Currently, surgical peeling is the only therapeutic option (6).

The evolution of these changes are not currently directly quantified by physicians with an objective measure, but is rather assessed by clinicians through visual inspection of series of images of the retina (7). Consequently, surgical timing remains uncertain and there is a wide disparity in approaches among surgeons (7). Quantitative assessments of retinal changes have become more widespread with the introduction of Optical Coherence Tomography (OCT) imaging, which allows precise measurements of retinal morphology (8).

Because the development of ERMs is often accompanied by changes in retinal thickness, this parameter is often used as a surrogate measure of change (9). However, this alone does not directly quantify many important features of ERMs. In a previous publication, we introduced the relaxation index (RI), a measure to quantify the three-dimensional shift in position of vascular landmarks before and after ERM peeling surgery (10), showing that this measurement was linked to post-operative best corrected visual acuity (BCVA) recovery. In this manuscript, we want to instead focus on using this measurement to track the natural evolution of ERMs in OCT scans, measuring the progression of traction from a baseline assessment up to the immediate post-operation period.

The study analyzed 9 eyes from 9 patients (average age: 81 ± 4 years). The mean time between the first recorded visit (T1) and surgery was 12 ± 9 months. At T2 (preoperative), the average BCVA was 0.54 ± 0.18.

We present an updated method of processing images to obtain the previously proposed RI (10). This method has demonstrated effectiveness in monitoring traction from the initial follow-up, through the pre-operative phase, to the release of traction after surgery. Unlike the previous method, we no longer rely on OCTA images for tracking vessel crossings. Instead, these are manually marked on the infrared (IR) image, significantly expediting the imaging processing speed. It’s worth noting that in this version, we did not prioritize assessing functional outcomes. There was a statistically significant improvement in the BCVA after surgery. However, we did not identify a statistically significant correlation between the RI and BCVA), which could be attributed to the limited size of the cohort analyzed.

In our previous study, we examined patients one week before surgery, and at 1, 3, and 6 months post-surgery (10). We observed that the RI stabilized at its final state one month after the operation. Therefore, in this analysis, we concentrated on the progression of traction in the preoperative period from the initial visit to one month after surgery. We observed a statistically significant difference in the RI across all the timepoints. We studied the progression of tangential traction from the first appointment, to one week before surgery and at one month after surgery. Even if this is a small cohort there is a clear trend that shows how the traction increases and is then released after surgery. This method is consistent in allowing quantitative analysis of tangential traction which is a key feature in epiretinal membrane pathophysiology. Objective measurement of tangential traction should in our opinion guide clinical decisions.

Despite recent advancements in OCT analysis tools, the classification of ERM remains predominantly qualitative (7). Various classification systems have been proposed by different authors. Hwang et al. introduced a system that relies on foveal characteristics, validated with multifocal electroretinography (mERG), to delineate functional distinctions between stages. Konidaris et al. suggested a categorization based on vitreous state and structure, but this classification lacks validation, rendering its clinical utility uncertain (12). Stevenson et al. advocated considering central foveal thickness and inner segment ellipsoid band integrity as critical morphological parameters (13). Additionally, Govetto et al. established a four-stage classification system, centered on the presence of the foveal pit and the integrity of outer and inner retinal layers (7).

Another study underscored the significance of traction as a pivotal prognostic factor, particularly focusing on intra-retinal changes and measuring traction depth (14). Quantifying intraretinal morphologic changes remains a challenge. Furthermore, several investigations have explored the relationship between the morphology of retinal layers and BCVA, revealing that the morphology of the inner segment/outer segment (IS/OS) layer can serve as a predictive indicator for functional outcomes. Finally, there is an increasing interest at evaluation the effect of traction on ectopic inner foveal layers and its correlation with functional outcomes. Zhang Z. et al. found that stereoscopic foveal deviations significantly correlated with the baseline and postoperative visual acuity (15).

The introduction of the RI is aimed at directing scientific attention towards the macular region as a three-dimensional entity. The effect of the ERM traction on the retina indeed, as the alteration of the foveal morphology and retinal layers disorganization, can be caused both by the vertical traction and horizontal tangential forces.

While a three-dimensional evaluation is achievable with en-Face OCT, the study of tractional forces is uniquely possible with the RI. In our view, the RI represents a powerful index that now requires augmentation with extensive clinical data to fully demonstrate its clinical utility.

The current version of the RI incorporates several noteworthy updates, as previously discussed in the methods section. However, there are certain limitations that currently constrain the applicability of this method. Primarily, the manual marking of vascular crossings remains a significant bottleneck, in our view. Addressing this limitation is crucial to enable the widespread application of this method, as it is currently time-consuming. In this regard, our forthcoming work is dedicated to leveraging deep learning algorithms to automate vessel detection. Additionally, we applied stringent inclusion criteria and only assessed this on a small cohort of patients, which limits the robustness of our findings. Visual acuity was recorded in decimals and not in LogMAR. Lastly, we did not find any correlation with the functional method, which restricts the practical implications of this data to a pure study of the tractional forces without immediate clinical relevance.

The Relaxation Index emerges as a comprehensive and straightforward parameter for objectively measuring and tracking ERM traction in three dimensions across a wide area of the posterior pole, both pre- and post-surgery. Further studies involving larger cohorts are necessary, and potential integration with built-in software may establish it as a critical parameter for making objective decisions regarding surgical timing.