Baseline Deep Capillary Plexus Ischemia Predicts Earlier Diabetic Retinopathy Complications Than Foveal Avascular...

To evaluate whether diabetic macular ischemia metrics-deep capillary plexus baseline geometric perfusion deficit (GPDd) and foveal avascular zone enlargement (FAZe)-predict diabetic retinopathy (DR) complication risk and timing over a 2-year period.

Prospective observational cohort study.

265 eyes (175 patients) across the DR severity spectrum.

Patients underwent baseline and follow-up ophthalmic evaluations. GPDd was quantified as the percentage of retinal tissue ≥30μm from the nearest perfused capillary on 3 × 3mm optical coherence tomography angiography (OCTA) images of the deep capillary plexus. FAZe was calculated as the difference between structural FAZ area measured on en face OCT and functional FAZ area measured on OCTA. Time-to-complication analysis was conducted using Cox regression models, adjusted for age, sex, HbA1c, diabetes duration, and hypertension; because of its collinearity with OCTA metrics, DR severity was excluded from the primary analysis. Restricted mean survival time (RMST) quantified differences in complication-free survival time (days). Kaplan-Meier curves compared survival profiles between high- and low-biomarker groups, defined using the optimal Youden's J-index cutoff for referable DR.

Baseline GPDd and FAZe, occurrence of DR complications, and time to first complication. DR complications were defined as loss of ≥10-letter BCVA, DR severity progression, vitreous hemorrhage, initiation of anti-VEGF therapy, or panretinal photocoagulation.

Higher baseline GPDd was associated with both increased hazard and earlier onset of DR complications. In RMST analysis, a +1 SD increase in GPDd corresponded to a 70-day reduction in complication-free survival (95% CI, -103.9 to -36.0; p<0.001). When dichotomized into high and low GPDd groups, eyes with high GPDd developed complications 117.2 days (∼4 months) earlier than eyes with low GPDd (95% CI, -187.7 to -46.8; p=0.001). In multivariable Cox regression, each one standard deviation (SD) increase in GPDd (3.62%) was associated with 77% higher hazard of complications (p<0.001). By contrast, FAZe was not significantly associated with increased hazard or earlier complication onset after adjustment for GPDd.

Elevated GPDd, but not FAZe, independently predicted increased complication risk and shorter complication-free survival. GPDd is a robust, time-sensitive biomarker for patients who might benefit from closer monitoring.