Choriocapillaris Flow Deficit as a Biomarker for Diabetic Retinopathy and Diabetic Macular Edema: 3-Year Longitudinal Cohort

PURPOSE: To investigate the relationship between choriocapillaris flow deficit percentage (CC FD%) by swept-source optical coherence tomography angiography (SS-OCTA) and 3-year risk of diabetic retinopathy (DR) progression and diabetic macular edema (DME) development. DESIGN: Prospective, observational cohort study. METHODS: A total of 903 participants with type 2 diabetes mellitus (T2DM) without DR or with mild nonproliferative DR (NPDR) free of DME at baseline were followed up annually for 3 years. All participants underwent standard 7-field fundus photography and spectral-domain OCT. SS-OCTA was used for retinal and choriocapillaris imaging and 3 × 3-mm2 macular CC FD% was quantified. Univariate and multivariate logistic models were used to evaluate the association between CC FD% and 2 or more steps of DR progression and DME development. The additional predictive value of CC FD% for outcome events was assessed using C statistic, net reclassification index (NRI), and integrated discrimination improvement index (IDI). RESULTS: Over 3 years, 295 of 1805 eyes (16.34%) developed DR progression, and 118 eyes (6.54%) developed DME. A higher average CC FD% was correlated with DR progression (odds ratio [OR], 3.41 per SD increase, 95% CI: 2.65-4.39, P < .001) and DME development (OR, 1.37 per SD increase, 95% CI: 1.06-1.77, P = .016) after adjusting for confounders. In the ETDRS regions, increased CC FD% in all fields was associated with DR progression; however, increased CC FD% in the inferior field was associated with DME development. Compared with the models based on established risk factors, the addition of average CC FD% significantly improved the C statistics for DR progression (0.712 to 0.777, P < .001) and DME occurrence (0.743 to 0.773, P = .044). The estimated NRIs and IDIs (all >0) indicated that the addition of CC FD% led to a significant improvement in the discriminative performance for end points. CONCLUSION: CC FD% is independently associated with DR progression and DME development in the Chinese T2DM population and provides incremental predictive value beyond traditional risk factors and retinal microvascular parameters. Further inclusion of CC FD% in DR prediction models helps guide population-based screening and personalized management.