A total of 139 eyes from 139 DR patients were included in this study. They were divided into four groups according to the EDTRS: diabetic patients without DR (n = 29 eyes, 20%), patients with M-NPDR (n = 36 eyes, 26%), patients with S-NPDR (n = 37 eyes, 27%) and patients with PDR (n = 37 eyes, 27%). The relevant characteristics of all the study participants are listed in Table 1. There were no significant differences between the groups in terms of sex, eye type or hypertension (P < 0.05). They can be pairwise comparisons. In terms of age and duration, there were significant differences among groups with the progression of DR (P = 0.004, P < 0.001). (Table 1).
PSV, EDV and RI of the OA, PCA and CRA were measured and analysed in retrobulbar blood vessels of DR patients in different stages. The results showed that with increasing DR severity, the PSV of OA, PCA and CRA decreased from NDR to S-NPDR, remained the lowest in S-NPDR and then increased in PDR. The EDVs of OA and CRA also increased first and then decreased and remained the lowest in the S-NPDR group. The RI of OA decreased gradually as DR progressed, while no such change was observed in the other two vessels (Fig. 1). OAPSV, PCAEDV, PCARI, CRAPSV and CRAEDV were significantly different among the different groups (P = 0.008, P = 0.017, P = 0.007, P < 0.001, P < 0.001). CRAPSV and CRAEDV were significantly higher in the NDR group than in the S-NPDR group, and the difference was statistically significant (P < 0.001, P = 0.002); PCAPSV, PCAEDV, CRAPSV and CRAEDV were significantly lower in the NDR group than in the M-NPDR group (P = 0.026, P = 0.026, P = 0.012, P = 0.002). In contrast, the PCARI in the S-NPDR group was significantly higher than that in the M-NPDR group (P = 0.026). In addition, CRAPSV and CRAEDV in the PDR group were significantly higher than those in the S-NPDR group (both P < 0.001) (Table 2).
Table 3 summarizes the diagnostic sensitivity of the hemodynamic indexes of CRA, PCA and OA between M-NPDR and S-NPDR by ROC curve. Among all the hemodynamic parameters, the area under the curve (AUC) of CRAPSV was the highest (0.781), but the sensitivity and specificity were lower (80% and 30%, respectively). PCAPSV had relatively high sensitivity (94%) and the highest specificity (68%) among these indexes. The sensitivity and specificity of PCAEDV were 97% and 65%, respectively. CRAEDV and OAEDV had the same sensitivity (89%) in this diagnostic prediction model, but the specificity of CRAEDV (49%) was slightly lower than OAEDV specificity (65%) (Table 3 and Figure 2).
Furthermore, retinal and choroid blood flow data of 74 eyes in 74 patients were analysed and divided into three groups, including 20 eyes in the NDR group, 32 eyes in the M-NPDR group, and 22 eyes in the S-NPDR/PDR group. Retinal and choroid blood flow parameters in different groups were summarized and analysed. Some statistically significant data are shown (Supplementary Table 1 and Table 2).
The correlation analysis results of ocular vascularity in SMI and UWF-OCTA parameters are presented below. The retrobulbar parameters in SMI were significantly correlated with retinal and choroidal parameters in different degrees of DR.
For the CRA, in the NDR group, CRAPSV was positively correlated with nasal, superior and temporal SVD (15–18 mm) (r = 0.760, r = 0,673, r = 0.788; P < 0.05); CRAPSV was positively correlated with inferotemporal ChCVD, supratemporal and inferotemporal ChVVD (r = 0.663, r = 0.644, r = 0.634, P < 0.05); CRAEDV was positively correlated with temporal SVD (3–6 mm), superior SVD (15–18 mm), DVD (0–1 mm), nasal DVD (1–3 mm), superior DVD (3–6 mm), temporal and inferior DVD (6–9 mm), nasal DVD (15–18 mm), supratemporal and inferotemporal ChVVD, the coefficients of association ranged from 0.649 to 0.782 (all P < 0.05); CRARI was negatively associated with temporal SVD (3–6 mm), DVD (0–1 mm), nasal, superior and inferior DVD (1–3 mm), nasal, superior and inferior DVD (3–6 mm), temporal and inferior DVD (6–9 mm), and nasal DVD (15–18 mm), the coefficients of association ranged from − 0.635 to -0.899 (all P < 0.05). No significant correlation was found between CRARI and choroidal parameters (Table 4). In the M-NPDR group, no correlation was found between the hemodynamic parameters of CRA and retinal parameters. However, CRAPSV was positively correlated with temporal ChCVD (r = 0.543, P < 0.05), CRAEDV was positively correlated with nasal ChVVD (r = 0.498, P < 0.05), and no correlation was found between CRARI and choroidal parameters. (Table 5). In the S-NPDR/PDR group, there was no correlation between CRAEDV and CRARI and retinal and choroidal parameters. A significant correlation was found between CRAPSV and temporal SVD (6–9 mm) (r = 0.635, P < 0.05) (Table 6).
For the PCA, in the NDR group, PCAPSV was positively correlated with SVD (15–18 mm), DVD (0–1 mm) and inferotemporal CVV (r = 0.794, P < 0.01; r = 0.710, r = 0.636, P < 0.05); PCAEDV was correlated with SVD (15–18 mm), DVD (0–1 mm), nasal DVD (15–18 mm), supratemporal, supratemporal and central ChVVD and inferotemporal CVV, and the coefficients of association ranged from 0.636 to 0.842 (all P < 0.05); PCARI was negatively correlated with superior and temporal SVD (1–3 mm) and supratemporal, supranasal and inferior ChCVD, and the coefficients of association ranged from − 0.668 to -0.735 (all P < 0.05) (Table 4). In the M-NPDR group, PCAPSV was significantly positively correlated with temporal SVD (1–3 mm), temporal and superior SVD (9–12 mm), temporal and inferior SVD (12–15 mm), temporal and inferior SVD (15–18 mm), temporal and inferior DVD (12–15 mm), and temporal and inferior DVD (15–18 mm), and the coefficients of association ranged from 0.499 to 0.792 (all P < 0.05). No association was found between PCAPSV and choroidal parameters. PCAEDV was significantly correlated with SVD (0–1 mm); nasal, superior and temporal SVD (1–3 mm); temporal SVD (12–15 mm); temporal and inferior SVD (15–18 mm); and nasal, superior and temporal DVD (15–18 mm). The coefficients of association ranged from 0.498 to 0.721 (all P < 0.05). In the correlation analysis with choroid parameters, PCAEDV was positively associated with nine quadrants of CVV and CVI (r = 0.579 ~ 0.729, all P < 0.05). There was a negative correlation between PCARI and supratemporal, superior, supranasal, temporal, central and nasal ChCVD (r= -0.570 ~ -0.757, all P < 0.05) (Table 5). In S-NPDR/PDR group, PCAPSV was positively correlated with ChVVD; PCAEDV was positively correlated with SVD (9–12 mm), SVD (12–15 mm), SVD (15–18 mm) and DVD (0–1 mm) (r = 0.729, P < 0, 05. r = 0.758, P < 0.01; r = 0.654, r = 0.645, P < 0.05); PCARI was negatively correlated with nasal and superior SVD (9–12 mm), nasal SVD (12–15 mm), nasal SVD (15–18 mm), nasal, superior and inferior DVD (1–3 mm), nasal DVD (6–9 mm), and nasal DVD (9–12 mm) (r= -0.618 ~-0.924, all P < 0.05). PCAEDV and PCARI were correlated with choroidal parameters (Table 6).
For the OA, in the NDR group, OAPSV was negatively correlated with nasal SVD (12–15 mm) and inferotemporal and inferonasal ChVVD (r=-0.643, r=-0.618, r= -0.671, P < 0.05). No correlation was found between OAEDV and OARI and choroidal and retinal parameters. (Table 4). In the M-NPDR group, OAPSV was negatively correlated with inferior SVD (3–6 mm), inferior SVD (6–9 mm) and inferonasal ChVVD (r=-0.510, r=-0.575, r= -0.511, P < 0.05); OAEDV was negatively correlated with inferior SVD (3–6 mm) and superior, temporal and inferior DVD (9–12 mm).(r=-0.528, r=-0.556, P < 0.05; r= -0.631, P < 0.01; r= -0.582, P < 0.05); OARI was negatively correlated with inferior, supratemporal, supranasal, temporal and nasal ChCVD (r=-0.594, r=-0.584, r=-0.514, r=-0.561, r=-0.579, P < 0.05) (Table 5). In the S-NPDR/PDR group, only OAEDV was negatively correlated with superior ChVVD (r=-0.631, P < 0.05) (Table 6).