Abstract
PURPOSE: Retinal imaging may offer a non-invasive method to detect coronary artery disease (CAD). Geometric perfusion deficits (GPDs), defined as intercapillary areas beyond 30 µm from the nearest vessel, were analyzed for their CAD association and topographical patterns. METHODS: Patients undergoing coronary angiography and healthy controls (mean ages: non-CAD, 58 ± 9 years; CAD, 61 ± 9 years) were assessed. Fovea-centered 6 × 6-mm optical coherence tomography angiography (OCTA) images were acquired, and the superficial capillary plexus was analyzed. GPD density, count, ischemic extent (IE), and top 10 GPD areas (MT10) were extracted. The Gensini score (GS) quantified CAD severity. Multivariate mixed-effects models were calculated for statistical analysis. RESULTS: This study analyzed 237 patients (382 eyes), 141 (225 eyes) with CAD. Significant associations with the GS were found across all metrics after adjustments (density coefficient = 0.050, P = 0.001; IE coefficient = 0.002, P = 0.005; MT10 coefficient = 0.066, P = 0.005; count coefficient = 0.015, P = 0.003). The most profound changes were observed in the parafovea. Significant group differences were found between the non-CAD and moderate CAD groups, as well as between the non-CAD and severe CAD groups in the total image and the parafovea, temporal, and nasal outer segments. Moderate and severe CAD did not display significant differences. GPD metrics showed higher predictive abilities than conventional OCTA metrics. CONCLUSIONS: Four GPD metrics were significantly associated with CAD, with the parafoveal area showing the highest significance and consistency. GPDs differed significantly between CAD and non-CAD patients.