Idiopathic epiretinal membranes (ERMs) are avascular fibrous layers affixed along the surface of the macula. They are principally the consequence of physiological posterior vitreous detachment and therefore mainly affect patients beyond the fifth decade. Based on color fundus photography, studies report a prevalence ranging from 6% to 19% in elderly people. Moreover, their real frequency is probably underestimated, since the new spectral-domain optical coherence tomography (SD OCT) devices provide better detection capability.

More than two thirds of ERMs are nonprogressive, but in some cases expansion and further retraction lead to tangential tractions with a variable amount of macular distortion and thickening. Although only 2% are responsible for visual acuity below 20/200, visual disturbances even with preserved distance acuity impact these patients’ quality of life (QOL). Several authors have investigated the relationship between functional disturbances and SD OCT findings in various retinal diseases including ERM. It appears that integrity of external retinal layers and foveal thickness are closely correlated with visual parameters.

Since the first 6 cases of successful ERM removal were reported by Machemer in 1978, the surgical procedure has steadily improved. ERM peeling is now an efficient, safe, and standardized technique with good anatomic and functional outcomes for at least 75% of patients and visual improvement that extends up to 1 year. Since cataract arises in most phakic patients within 2 years after ERM removal, it has been demonstrated that the combination of these 2 procedures allows equivalent outcomes to consecutive surgery with lower costs and fewer constraints. Nevertheless, in some cases and despite successful peeling without complications, functional outcome may be disappointing. It is therefore necessary to identify preoperative prognostic factors for good postoperative visual recovery. Some clinical factors have long been known, notably preoperative visual acuity and duration of visual impairment. Recent publications have suggested that preoperative SD OCT findings such as foveal thickness and integrity of the photoreceptor (PR) layers may also be reliable prognostic factors.

The purpose of this study was to determine the clinical and anatomic preoperative factors associated with complete visual recovery (20/20 or more) 1 year after a combined cataract and idiopathic ERM surgery in a large retrospective cohort. Secondly, we designed a preoperative predictive score for full visual recovery.

Methods

Population

This retrospective interventional case series study reviewed 279 consecutive eyes that underwent combined cataract and ERM surgery between April 1, 2009 and September 28, 2012 at the Dijon University Hospital, Dijon, France. All patients signed informed consent for the surgical procedure. The protocol was in accordance with the tenets of the Declaration of Helsinki. Owing to its retrospective design, the local institutional review board (Comité de Protection des Personnes, région Est) ruled that approval was not required for this study. Patients were excluded if they presented any cause of secondary ERM (including diabetic retinopathy, vein occlusion, ocular inflammation, retinal detachment, or previously vitrectomized eye), high myopia (axial length ≥26 mm or spherical equivalent ≥6 diopters), severe media opacities with weak SD OCT signal strengths, or any macular or optic nerve disease. Patients with a follow-up of less than 12 months were excluded and only 1 eye was included in those with bilateral ERM.

Clinical Assessments

Data concerning age, sex, and duration of symptoms were gathered from the patients’ preoperative charts. A complete ophthalmologic examination was performed before and at 1, 6, and 12 months after the surgery, including best-corrected visual acuity (BCVA) with Snellen charts converted into the logarithm of the minimal angle of resolution (logMAR), slit-lamp biomicroscopy, indirect fundus, and macular SD OCT.

Optical Coherence Tomography Parameters

Grayscale SD OCT scans were obtained using a Cirrus high-definition device (Carl Zeiss Meditec AG, Jena, Germany). The acquisition protocol covered a 6 × 6 × 2-mm macular volume composed of 1 native vertical and 128 horizontal B-scans, each composed of 512 axial A-scans. A signal strength greater than or equal to 5/10 was required and data were analyzed using the inbuilt Cirrus software version 6.5. Images were reviewed by 2 independent readers (Y.K., J.C.R.), who were not aware of the patients’ visual outcomes. In the event of discordant readings, the case was discussed to determine the most appropriate interpretation. Figure 1 describes SD OCT parameters used for statistical analysis and Supplemental Figure 1 (Supplemental Material available at AJO.com ) displays examples of abnormal preoperative findings. Macular volume was automatically segmented by the software with delineation of internal limiting membrane (ILM) and retinal pigment epithelium (RPE), allowing reliable retinal thickness assessment. We evaluated mean foveolar thickness (MFT) as the mean distance between the ILM and the RPE in the central 1-mm area automatically provided by the software. To assess the level of impairment of the PR layer, we studied 4 SD OCT parameters: the outer limiting membrane (OLM); the inner and outer segment (IS/OS) junction; the Verhoeff membrane, also called the cone outer segment tip (COST) line; and the PR outer segment (PROS) length. Further qualitative SD OCT parameters were assessed (foveolar profile, macular wrinkling, retinal hyporeflective inner plexiform cysts, foveal bulge along the IS/OS, vitreomacular traction, foveal herniation, macular pseudohole, and lamellar pseudohole).

Optical coherence tomography parameters used for pre- and postoperative morphologic analyses in patients undergoing combined cataract and epiretinal membrane surgery. (Top, left and right) Mean foveolar thickness. This parameter corresponds to the mean distance between the internal limiting membrane and the retinal pigment epithelium (RPE) in the central 1-mm area automatically provided by the software. (Bottom left) Outer limiting membrane (OLM) and inner and outer segment (IS/OS) junction. The OLM (a) and IS/OS (b) lines were assessed over a 500-μm area centered on the fovea. An intact line was defined as a regular continuous line (grade 0). Irregularity or a focal disruption of less than 200 μm was graded 1 and a gap exceeding 200 μm was graded 2. The disruption score totaled the 2 grades evaluated on both vertical and horizontal native B-scans passing through the foveola. (Bottom right) Cone outer segment tip (COST) line and photoreceptor outer segment (PROS) length. The COST line was assessed as normal (c) or damaged (d) when it was replaced by hyporeflectivity within the 500-μm central area. PROS length (e) was defined as the distance between the IS/OS junction and the RPE. The mean of 3 measurements (1 central and 2 others at 100 μm on either side of the first one) on the horizontal central B-scan was used for statistical analyses.

Surgical Procedure

All patients underwent a combined cataract and ERM surgery owing to visual impairment or disabling metamorphopsia. A 23 or 25 gauge 3-port pars plana core vitrectomy was performed. The ERM and ILM were peeled using a microforceps over an area covering 4–6 mm centered on the fovea. ILM dye was used at the surgeon’s discretion (Brilliant Peel; Fluoron GmbH, Ulm, Germany, or ILM-Blue; DORC, Zuidland, the Netherlands). Cataract surgery was performed by phacoemulsification (Stellaris PC system; Bausch & Lomb, Rochester, New York, USA) with a 3-piece intraocular lens (Acrysof MA 50 BM; Alcon Surgical, Inc, Fort Worth, Texas, USA), either before or after the vitrectomy.

Statistical Analysis

Patients were split into 2 groups according to their postoperative BCVA 12 months after the surgery. The first group was composed of those with optimal visual recovery (≥20/20) and the second group comprised patients with a poorer visual outcome (<20/20). Statistical analysis was performed using STATA/SE 10.0 (StataCorp LP, College Station, Texas, USA). Since some of the variables did not meet the assumptions necessary for parametric statistical tests, the comparisons were carried out with the Mann-Whitney, Wilcoxon matched-pairs signed rank, χ ² , or Fisher exact tests. Further Spearman rank correlation coefficients with their associated linear curve fits were provided. A multivariate analysis was conducted with the BCVA group set as the dependent variable in a logistic regression model of the potential prognosis factors (namely age, duration of symptoms, initial BCVA, MFT, and IS/OS disruption score). Finally, the preoperative predictive score was built by analyzing a multivariate multiple regression model of independent prognostic factors converted into binary variables, with the final BCVA group set as the dependent variable. Several analyses were performed to determine the most appropriate thresholds for each factor, taking into account its median (age, duration of symptoms, and IS/OS disruption score) or the final visual outcomes in the subgroups of the preliminary analysis (initial BCVA). In the final model, each regression coefficient was weighted relative to the least of them to obtain a 10-point score. Further specificity, sensitivity, and the receiver operating characteristic (ROC) curve were calculated to determine the most accurate cutoff for this overall score. All these analyses were 2-tailed and the significance threshold was fixed at .05.