Abstract:AIM: To observe and quantitatively analyze the thickness of macular ganglion cell inner plexiform layer(GCIPL)and the characteristics of superficial retinal capillaries vessel density in different stages of diabetic retinopathy(DR)by optical coherence tomography(OCT)and optical coherence tomography angiography(OCTA).METHODS: A retrospective case-control study. Thirty-three patients with diabetic(54 eyes)were selected as the DR group from December 2019 to May 2020. Among them, six patients(8 eyes)as non-diabetic retinopathy(NDR)group, eighteen patients(28 eyes)as non-proliferative diabetic retinopathy(NPDR)group and nine patients(18 eyes)as proliferative diabetic retinopathy(PDR)group according to fundus conditions. Eighteen healthy volunteers(26 eyes)without eye disease were selected as the normal group. The macular GCIPL thickness and the values of vascular linear density(vascular density, VD)and density of vascular perfusion(perfusion density, PD)in the superficial retinal capillaries vessels in each quadrant of macular region were observed and quantitatively analyzed in DR patients with different stages. RESULTS: The VD, PD and minimum thickness of GCIPL in each quadrant of DR group was lower than that of the healthy control group(P<0.05). The minimum thickness of GCIPL in macular area and the VD of superficial retinal capillaries in each quadrant decreased significantly in patients with different stages of diabetic retinopathy(P<0.01). The inferior VD of superficial retinal capillaries vessels had the highest diagnostic value for DR(AUC=0.807, optimal diagnostic threshold value of 18.60 mm-1, sensitivity of 0.923, specificity of 0.648). The minimum thickness of GCIPL in macular area of DR patients was positively correlated with VD of superficial retinal capillaries vessels in each quadrant(r=0.342, 0.480, 0.384, 0.342, all P<0.05). CONCLUSION: OCT combined with OCTA can provide repeatable and quantifiable detection methods and monitoring indicators for early assessment and regular follow-up of DR progress.