Abstract
OBJECTIVE: To identify pre-treatment optical coherence tomography (OCT) biomarkers predictive of recurrence in eyes with diabetic macular edema (DME) after anti-VEGF therapy. METHODS: This retrospective cohort study included 122 eyes with DME treated with anti-VEGF monotherapy (ranibizumab; Novartis Pharma Schweiz AG, Basel, Switzerland) at Pinghu First People's Hospital (January 2020 - December 2023). The treatment protocol consisted of 3 monthly loading doses followed by a pro re nata (PRN) regimen. Patients were stratified into recurrence (n=54) and non-recurrence (n=68) groups based on predefined criteria during 12 months of follow-up. Baseline OCT parameters were compared, and multivariate logistic regression was used to identify independent predictors. A combined model's predictive performance was assessed using receiver operating characteristic (ROC) curve analysis. RESULTS: The recurrence group had significantly higher baseline central retinal thickness (452.7±84.3 vs 391.5±70.2 μm, P=0.002), subretinal fluid (SRF) prevalence (63.0% vs 36.8%, P=0.008), proportion with ≥5 hyperreflective foci (HRF) (68.5% vs 35.3%, P<0.001), and ellipsoid zone (EZ) disruption rate (57.4% vs 30.9%, P=0.006), and required more injections (6.4±1.3 vs 4.8±1.2, P<0.001). Multivariate analysis confirmed HRF ≥5 (OR=3.52, P<0.001), SRF presence (OR=2.89, P=0.007), and EZ disruption (OR=2.41, P=0.023) as independent risk factors. Their combined model predicted recurrence with an AUC of 0.841. CONCLUSION: HRF, SRF, and EZ integrity are key OCT biomarkers for DME recurrence. A combined model aids risk stratification for personalized management.