Retinal layer reflectivity and thickness as OCT biomarkers for diagnosis and imaging-based stratification relative to the 4.5-h window in acute central retinal artery occlusion

Purpose To quantify optical coherence tomography (OCT)-derived retinal layer reflectivity and retinal thickness changes in acute central retinal artery occlusion (CRAO) for rapid diagnosis and stratification of ischemic tissue state, referenced to the clinically relevant 4.5-h window for reperfusion therapies. Methods This retrospective multicenter study included 39 patients with unilateral acute non-arteritic CRAO imaged by OCT within 48 h of reliably reported symptom onset. Five horizontal macular B-scans per eye [central (foveal), two superior, two inferior] were analyzed. Inner retinal layers (IRL) and outer retinal layers (ORL) were manually segmented. Mean grayscale reflectivity was extracted from predefined regions of interest and summarized as the within-eye IRL/ORL reflectivity ratio. Retinal thickness was obtained from Early Treatment Diabetic Retinopathy Study (ETDRS) sectors; relative retinal thickness increase (RRTI) was defined as CRAO-eye thickness divided by fellow-eye thickness. Diagnostic performance (CRAO vs. fellow eye) and classification relative to the 4.5-h reference (< 4.5 vs. ≥ 4.5 h) were assessed using receiver operating characteristic (ROC) analyses and non-parametric testing. Results The within-eye IRL/ORL reflectivity ratio reliably distinguished CRAO-affected eyes from fellow eyes across all macular scan locations (AUC = 0.98–0.99; p < 0.001). Retinal thickness increased in CRAO eyes across all ETDRS sectors except the foveal center, with sector-wise diagnostic performance ranging from AUC 0.78–0.99. Both reflectivity- and thickness-based metrics demonstrated time dependence: the IRL/ORL reflectivity ratio increased with time-to-OCT in CRAO eyes ( R 2 0.43–0.54) and discriminated < 4.5 vs. ≥4.5 h (AUC 0.88–0.91). Inter-eye analyses indicated that temporal separation was primarily driven by progressive attenuation of ORL reflectivity, whereas IRL hyperreflectivity was established early and remained comparatively stable. Retinal thickness and RRTI increased with ischemia duration across ETDRS sectors, with strongest temporal discrimination in inferior sectors (S8; optimal RRTI cut-off ≈ 1.20). Conclusion Quantitative OCT biomarkers based on within-eye reflectivity redistribution and retinal edema enable accurate diagnosis of acute CRAO and provide objective imaging-based information on ischemic tissue state, referenced to the 4.5-h therapeutic window. These rapidly obtainable measures may support acute stroke triage and telemedical workflows. Prospective validation, particularly in unknown-onset CRAO, is warranted before clinical implementation.