Surgical prognosis in epiretinal membrane: A 5-year longitudinal study of OCT biomarkers and visual acuity at a high-complexity referral center

Background: To determine the functional and anatomical outcomes at 12 months after pars plana vitrectomy (PPV) for epiretinal membrane (ERM) and the independent prognostic value of preoperative biomarkers in optical coherence tomography (OCT). Methods: Five-year analytical study of 340 eyes with ERM in stages (Govetto G2–4). Best-corrected visual acuity (BCVA) logMAR and OCT were evaluated at baseline, 1, 3, 6, and 12 months. The primary objective was to determine the change in BCVA at 12 months (ΔlogMAR 12 m – baseline). Univariate tests, ANCOVA, longitudinal mixed linear models, and logistic regression were applied for gain ≥0.2 logMAR. The models were adjusted for baseline VA, age, sex, aetiology, surgical technique, and analysis of biomarkers in OCT. Results: Baseline VA was 0.54±0.44 logMAR; distribution according to G2 (n=166), G3 (n=146), G4 (n=28). VA improved significantly at 12 months (Wilcoxon p<0.001), equivalent to 11–12 ETDRS letters. Secondary ERMs showed greater unadjusted gain than idiopathic ones, but aetiology was not an independent predictor after adjustment. The surgical technique was not independently associated with VA at 12 months (β=+0.031 logMAR; 95% CI; p=0.408) or with responder status (OR 1.18; 95% CI; p=0.625). By severity, advanced stages had worse baseline VA and greater absolute gains (Kruskal–Wallis H=7.61; p=0.0223), although with lower final VA compared to lesser stages. Baseline VA was the dominant predictor in all models (β=−0.766; 95% CI; p<0.001). Among preoperative biomarkers, COST line disruption independently increased the probability of response (OR=2.08; 95% CI; p=0.013). DRIL, EZ disruption, and central bouquet alterations did not reach significance after adjustment. Mixed models confirmed early improvement, maintained up to 12 months. Kaplan–Meier showed faster time to response in advanced stages. Conclusions: PPV by ERM achieves significant improvement at 12 months. Baseline VA is the main determinant of prognosis. According to Govetto's classification, the greatest gain occurs in advanced stages, but with lower final VA. The COST line emerges as the only preoperative biomarker with independent prognostic value for achieving ≥0.2 logMAR; the others provide limited utility when baseline VA is considered. Trial registration: Not applicable.