Changes of blood flow in macular zone of patients with diabetic retinopathy at different stages evaluated by optical coherence tomography angiography.

PURPOSE

To evaluate changes in macular blood flow in patients with various stages of diabetic retinopathy (DR) by optical coherence tomography angiography (OCTA).

MATERIALS AND METHODS

The data of 94 patients (122 eyes) with type 2 diabetes mellitus admitted from January 2018 to May 2021 were recorded, including 31 patients (42 eyes) with no DR (NDR group), 33 patients (40 eyes) with mild non-proliferative DR (mild NPDR group) and 30 patients (40 eyes) with moderate NPDR (moderate NPDR group). Forty healthy patients (40 eyes) who were examined during the same period were selected as control group. The blood flow densities at the superficial capillary plexus (SCP), deep capillary plexus (DCP) and 300μm-wide region near the foveal avascular zone (FAZ) (FD300), FAZ area, FAZ perimeter and acircularity index (AI) were compared. Correlations of blood flow densities at the SCP, DCP and FD300, FAZ area, FAZ perimeter and AI with the severity of DR were investigated by multivariate regression analysis. A receiver operating characteristic (ROC) curve was plotted to analyze the predictive efficacy of the OCTA indices for early DR.

RESULTS

Blood flow densities at the SCP, DCP and FD300 significantly decreased, while FAZ perimeter and AI increased in the NDR and DR groups compared with those of the control group (P<0.05). Blood flow densities at the SCP, DCP and FD300 of the DR group were lower than those of the NDR group, but the FAZ perimeter and AI of the former were larger (P<0.05). Blood flow densities at the SCP, DCP and FD300 in the macular zone of DR patients declined with increasing DR stage, while FAZ perimeter and AI increased (P<0.05), without collinearity among the variables. The area under the curve of FD300 for predicting early DR was highest (0.804), with a sensitivity and specificity of 84.13% and 67.19%, respectively. AI with a cut-off value of 1.05 showed the optimal specificity (87.84%) and sensitivity (73.62%) for predicting early DR.

CONCLUSIONS

The flow density at FD300 declines, while FAZ perimeter and AI increase in the macular zone of patients with early DR. These three OCTA indicators show close associations with DR stage, and can function as predictors of early DR.