The current study used a commercially available SS-OCTA to investigate the retinal and CC blood perfusion in diabetic eyes without clinically visible DR and compared their quantitative indices with those of nondiabetic controls. The diabetic eyes included in this study were from patients recently diagnosed with DM (between 1 and 5 years), and the quantitative assessments demonstrated that the CC flow reduction may precede the retinal flow changes in the macula. This finding may suggest that flow deficits in the choriocapillaris might be an earlier preclinical marker of microvascular dysfunction than retinal microvasculature in diabetic eye disease.
Using OCTA to assess the retinal vascular density in diabetic eyes without clinically visible DR, researchers have drawn mixed conclusions. For example, Dimitrova et al.
16 and Hwang et al.
29 showed significantly reduced parafoveal VAD in both the superficial and deep retinal capillary plexuses in diabetic eyes without retinopathy compared to controls, while Simonett et al.
30 and Carnevali et al.
15 reported significantly reduced parafoveal VAD in the deep but not in the superficial plexus in diabetic eyes without retinopathy. In contrast, other groups demonstrated that no significant differences were found in the superficial, deep, or whole parafoveal VAD between the two groups.
13,14 Compared with VAD, VSD is a more sensitive metric to measure perfusion changes at the capillary level.
24 However, VSD has not been compared between diabetic patients without retinopathy and nondiabetic controls. Although OCT and histology studies have confirmed the trilaminar capillary layout in the parafovea and perifovea,
20,21,31 most of the published studies investigating vessel density separated the retinal vascular system into two major plexuses. Moreover, recent studies also showed that additional segmentation and evaluation of the intermediate retinal layer from the superficial and deep retinal layers may enhance the ability of OCTA to detect early microvascular changes in diabetic eyes.
29,32,33 In this study, VAD and VSD were measured on the SRL, IRL, and DRL using our validated semiautomated segmentation algorithm. Our study showed that no significant difference was found in VSD or VAD in diabetic eyes without retinopathy compared to controls within the three vascular plexuses for a 6 × 6-mm scanning protocol.
The FAZ area assessed by OCTA has also been well investigated in diabetic eyes without retinopathy. Some groups reported a small but significant enlargement of the FAZ area in diabetic eyes without retinopathy compared to controls,
16,34 while others reported no significant difference of FAZ between the two groups.
13–15,35–37 All of the aforementioned measurements were conducted using the built-in default settings of commercial systems with automatic segmentation into two slabs (superficial and deep). Because of inherent errors in automatic segmentation of the superficial and deep capillary plexuses within the central fovea, FAZ measurements using default settings are likely to be biased.
38 We measured the FAZ area by utilizing the full-thickness retinal slab at the edge of the FAZ where the retinal vascular plexuses merge,
22 rather than attempting to divide it artificially. Our study showed no statistically significant difference in the FAZ area between the two groups, which was consistent with recent reports.
29,36 However, another report found a small enlargement of the FAZ measured on the full-thickness slab in diabetic eyes without retinopathy.
39 Further studies with larger sample sizes are needed to evaluate whether FAZ measurements would be a useful diagnostic tool for early diabetic microvascular dysfunction.
Retinal nonperfusion area (NPA) measurements were reported as a sensitive OCTA quantitative metric and could distinguish diabetic eyes without retinopathy from normal eyes.
29,36 We also conducted a detailed investigation into the NPA; however, in our study, no statistical difference of the NPA was observed in the parafoveal or perifoveal region between the two groups in different retinal layers. One possibility for the discrepancy in the results is that in contrast to the earlier study, our diabetic cohort had a short duration from diagnosis of DM (2.1 ± 1.2 years), which may have allowed us to study the microvascular changes in diabetic eyes at a very early stage. Furthermore, although the NPA was reported to be less age dependent,
40 we nevertheless selected age-matched individuals as controls in this study since prior investigations have shown that vascular density,
41–43 FAZ size,
41,43 and CC FDs tended to be influenced by age.
44,45
Several groups also investigated CC perfusion in diabetic eyes. Using an investigational SS-OCT system, Choi et al.
46 described focal or diffuse CC flow impairment in diabetic eyes without quantitative assessments. Nesper et al.
13 reported increased percent area of nonperfusion in CC in a 3 × 3-mm angiogram in diabetic eyes without retinopathy using a commercially available SD-OCT system. On the other hand, other groups reported no significant difference in CC vessel density between diabetic eyes without retinopathy and normal controls with SD-OCTA.
14–16 However, the vessel density metric may not be a good choice for CC quantification. As we know, CC vasculature is extremely dense in the posterior pole with small intercapillary distances (5–20 µm) that are smaller than the OCT system's lateral resolution (15–20 µm)
47; therefore, individual capillaries of CC cannot be clearly resolved with current commercial OCT systems. Instead of quantifying the CC vasculature directly, many researchers have chosen flow deficits to analyze CC perfusion.
12,44,45 The CC FD represents the area where there is a lack of CC flow or CC flow below the detectable threshold of the OCT system. To improve the robust assessment of the CC FDs, we segmented FDs with a size larger than normal intercapillary spacing (24 μm in diameter) for quantification, which is within the capability of the OCT system to resolve.
28 Moreover, we quantitated and compared the CC FDs in different macular regions since the CC FDs presented regional distributions in the macula.
44,45 We observed significantly increased FD% and enlarged average size of FDs in CC in diabetic eyes compared with controls within all quantified regions.
Vascular abnormalities in the choriocapillaris have also been demonstrated in diabetic eyes without retinopathy in histopathologic studies. Using alkaline phosphatase activity as a marker for viable CC endothelial cells, McLeod and Lutty
48 found that CC dropout was generally much more pronounced and involved larger areas in postmortem subjects with diabetes even without DR than those without diabetes. Interestingly, in a mouse model of DM, reduced choroidal perfusion was noted to occur prior to alterations of retinal perfusion and visual function.
49 Impaired visual function preceding clinically visible DR has also been observed in some population-based studies.
50,51 In the present study, we demonstrated with noninvasive SS-OCTA that CC perfusion reduction may precede retinal vascular changes in the macula of diabetic eyes. Although the outer retina receives most of its blood supply from the CC,
52 whether reduced perfusion contributes to abnormal visual function in diabetic patients before overt retinopathy still requires further investigation.
We acknowledge several limitations in this study. First, our study included a relatively small number of patients. This was mainly related to the strict exclusion criteria we employed. Larger cohort studies are necessary to confirm these preliminary findings. Second, this is a cross-sectional analysis with a short duration of diabetes. With the increase of duration, choroidal and retinal microvascular alterations may be more obvious and present different characteristics. Further longitudinal studies are needed to elucidate these microvascular alterations over time with the progression of diabetes. Third, no significant changes in retinal perfusion metrics were observed between the two groups. However, this does not necessarily mean that early retinal vascular alterations are not actually present. The development of more sensitive OCTA metrics may help to detect retinal perfusion alterations in early diabetes. Fourth, imaging of the deep large choroidal vessels may also provide additional information on the pathogenesis and progression of diabetic eye disease, which warrants a proper investigation. Lastly, we did not correct image magnification in lateral measurements due to the variation of axial length.
53 The magnification variation may affect the ability of the quantification metrics of FAZ, NPA, and CC FD sizes, although it has a negligible effect on the density or percentage measurements (e.g., VAD, VSD, and CC FD%). In the current study, only the patients with a refraction error less myopic than –6.0 diopters were included for the analyses. This inclusion criterion would limit the magnification variation to a relatively small range. This study was retrospective in its nature, and axial length measurements were not available for all the patients. Nevertheless, we ran a test on the FAZ, NPA, and mean size of FDs by considering the magnification variation and assuming the axial length artificially at the extreme cases of 26.4 mm (–6.0 diopters) and found that this did not change our final conclusions. However, we would suggest in future larger cohort and longitudinal studies that this magnification factor is considered for more accurate analyses to draw more definitive conclusions, particularly in the cases of myopic/hyperopic eyes.