Background
Diabetic retinopathy, a chronic and specific complication of diabetes1, is the main cause of blindness in the working-age population worldwide2. The alteration that most commonly causes visual loss in this disease is macular edema, a thickening at the area of maximum vision in the retina, which damages photoreceptors and reduces visual acuity3; its timely treatment allows to reduce thickening and, in some cases, improve visual acuity4.
A feature that can prevent visual recovery after treatment is the damage to the capillary network surrounding the fovea, the zone with the highest resolution in the macula where there are no blood vessels5; this region, known as the foveal avascular zone, enlarges in the presence of capillary closure and ischemia6, a condition that affects up to 7% of people with diabetic retinopathy7. The foveal avascular zone can be outlined by a fluorescein retinal angiography8, and currently with optical coherence tomography devices, which generate what is known as optical coherence tomography angiography (OCTA); this study provides other data such as parafoveal vessel density, the perimeter of the foveal avascular zone, and its circularity9.
The circularity is the similarity that the shape of the foveal avascular zone has with a circle, as calculated with the formula: circularity = 4πA/P2, where A is the area and P the perimeter of the foveal avascular zone; it becomes irregular when capillary closure exists, which reduces the circularity. The value of this variable comes along with the report of parafoveal vessel density, and it could be useful to identify an early reduction of this, particularly in center-involving macular edema, because retinal thickening changes the measurement level and limits the quantification of vessel density.
In our population, the characteristics of the foveal avascular zone circularity are unknown, which limits its application to detect cases with capillary loss. A study was conducted to determine the distribution of the foveal avascular zone circularity and its correlation with parafoveal vessel density, in subjects with and without diabetes.
Methods
This was an observational, descriptive, cross-sectional, and prospective study in subjects treated at a federal hospital in Mexico City. The Institutional Review Board of the hospital authorized the study, which adhered to the tenets of the Declaration of Helsinki; all the participants signed an informed consent before their evaluation.
Inclusion criteria were: subjects of any gender, aged 40-70 years, without diabetes or with Type 2 diabetes, without retinopathy or with mild or moderate diabetic retinopathy. Subjects with any other retinal diseases, previous intraocular surgeries, lens opacities, or any condition that prevented the acquisition of an angiotomographic image of adequate quality were excluded; elimination criteria were withdrawal of the informed consent and the presence of macular ischemia, which was operationally defined as an area of the foveal avascular zone that exceeded by two standard deviations (SD), the mean reported in eyes with non-proliferative diabetic retinopathy (0.62 mm2)10.
The sample was divided into three groups: subjects without diabetes were assigned to Group 1, subjects with diabetes without retinopathy to Group 2, and people with diabetic retinopathy to Group 3; a researcher recorded diabetes duration, treatments received, and the presence of systemic arterial hypertension, and measured the best corrected visual acuity in all the subjects, using Snellen equivalents. Another researcher obtained an eye fundus photograph with a non-mydriatic camera (Canon, Japan), which a third researcher (a retina specialist) evaluated to grade retinopathy according to the early treatment diabetic retinopathy study criteria. A fourth researcher acquired optical coherence angiotomography images, using the Cirrus 5000a device (Zeiss, Meditec, Dublin, CA), with the following characteristics: a 6 mm retinal thickness macular cube and a 3 × 3 mm angiogram of the superficial capillary plexus, obtained through the automatic segmentation algorithm of the equipment; image quality controls included a correct foveal centering, the absence of artifacts (eyelashes, movement, or any other artifact) that blocked the OCTA signal, and a signal strength >7.
Data of the parafoveal vessel density were obtained in each subject and measured by regions as follows: at the periphery of the millimeter concentric to the fovea (central vessel density), within 0.5-1.5 mm from the foveal center (inner vessel density), and in the entire 3 mm diameter region concentric to the fovea (complete vessel density). The equipment measured the area, perimeter, diameter, and circularity of the foveal avascular zone as well; the equipment calculated all variables automatically, including circularity.
Statistical analysis
Means of the angiotomographic variables were compared between the groups using Kruskal-Wallis's test and Mann-Whitney's U-test; correlations between circularity, foveal avascular zone, and parafoveal vessel density were obtained by Spearman's Rho test. p < 0.05 was considered significant; all the data were stored and analyzed with the SPSS version 22 software, for Windows.
Results
We evaluated 77 eyes of 52 subjects with an aged man mean 55.90, SD ± 9.0, 45 eyes were from females (58.4%); 57 eyes were from subjects with diabetes, which had a duration of 1 month-25 years (10.43 ± 73.8), 22 of them (73.3%) received oral hypoglycemic treatments. 14 eyes were of subjects with systemic arterial hypertension (36.8%); Group 1 had 20 eyes, Group 2 had 34 eyes, and Group 3 had 23 eyes. Table 1 shows the comparison of the general variables between groups.
Variables | Group 1 (n = 20) | Group 2 (n = 34) | Group 3 (n = 23) | p |
---|---|---|---|---|
Age | 55.70±12.79 | 55.67±6.03 | 56.65±9.18 | 0.86* |
Female gender | 40% | 58.8% | 73.9% | 0.07** |
Diabetes duration | - | 7.08±5.22 | 14.95±7.57 | 0.002*** |
Oral hypoglycemic treatments | - | 94.2% | 76.9% | 0.38** |
Systemic arterial hypertension | - | 20.6% | 30.4% | 0.41** |
*Kruskal-Wallis,
**χ2,
***Mann-Whitney's U-test.
Diabetic retinopathy in Group 3 was mild in 13 cases and moderate in 10; central field thickness in the sample was 212-298 µm (252.51 ± 19.89), macular volume was 9.10-10.90 mm3 (10.05 ± 0.44), and the foveal avascular zone area was 0.05-0.54 mm2 (0.29 ± 0.11). Table 2 presents the comparison of retinal thickness between groups.
Variable | Group 1 (n = 20) | Group 2 (n = 34) | Group 3 (n = 23) | p* |
---|---|---|---|---|
CFT (field 1) (µm) | 260.55±17.52 | 253.97±16.88 | 243.35±22.93 | 0.01 |
Field 2 (µm) | 316.75±16.68 | 322.62±14.61 | 317.39±16.74 | 0.36 |
Field3 (µm) | 305.00±14.27 | 308.85±15.37 | 302.78±21.21 | 0.62 |
Field 4 (µm) | 314.55±17.05 | 316.03±15.31 | 311.78±17.99 | 0.66 |
Field 5 (µm) | 320.00±18.40 | 325.00±15.25 | 316.30±19.15 | 0.26 |
Field 6 (µm) | 274.15±15.54 | 280.68±12.28 | 284.30±10.87 | 0.08 |
Field 7 (µm) | 261.70±13.91 | 263.62±13.24 | 270.69±14.03 | 0.10 |
Field 8 (µm) | 260.00±14.49 | 265.68±13.16 | 267.26±11.80 | 0.32 |
Field 9 (µm) | 288.75±17.09 | 297.18±15.02 | 296.35±13.28 | 0.21 |
Macular volume (mm3) | 9.91±0.50 | 10.09±0.44 | 10.10±0.39 | 0.46 |
*Kruskal-Wallis, CFT: center field thickness.
The central vessel density mean was higher in Group 1 than in Group 2 and higher in Group 2 than in Group 3; the inner and complete vessel densities were also higher in Group 2 than in Group 3. The foveal avascular zone area mean was lower in Group 1 than in Group 2, and in Group 2 the means of the area, perimeter, and diameter of the foveal avascular zone were lower than in Group 3; the circularity showed no difference between the groups (Table 3).
Variable | Group 1 (n = 20) | Group 2 (n = 34) | Group 3 (n = 23) | p |
---|---|---|---|---|
Central vessel density | 11.69±2.60 | 9.32±2.46 | 7.00±2.07 | <0.001 |
Inner vessel density | 21.11±2.29 | 20.78±1.52 | 19.90±1.80 | 0.06 |
Complete vessel density | 20.06±2.11 | 19.49±1.53 | 18.45±1.73 | 0.01 |
Foveal avascular zone | ||||
Area | 0.22±0.10 | 0.28±0.09 | 0.38±0.10 | <0.001 |
Perimeter | 2.03±0.48 | 2.29±0.39 | 2.77±0.47 | <0.001 |
Diameter | 0.65±0.15 | 0.73±0.13 | 0.88±0.15 | <0.001 |
Circularity | 0.65±0.06 | 0.65±0.10 | 0.61±0.10 | 0.27 |
*Kruskal-Wallis.
In the sample, the circularity had a low positive correlation with the inner vessel density (Rho = 0.35, p = 0.002) and with the complete vessel density (Rho = 0.33, p = 0.004). These correlations did not appear either in Group 1 (Rho = 0.42, p = 0.37) or in Group 2 (Rho = 0.03, p = 0.23); in Group 3, the circularity had a correlation with the central vessel density (Rho = 0.45, p = 0.03), with the inner vessel density (Rho = 0.56, p = 0.005), and with the complete vessel density (Rho = 0.53, p = 0.009) (Figs. 1 and 2). There was no correlation between the circularity and the area, the perimeter or the diameter of the foveal avascular zone, either in the sample or in each group.
Discussion
The mean circularity of the foveal avascular zone in subjects without diabetes was 0.65, which did not differ from that found in subjects with diabetes, with and without retinopathy. The circularity and parafoveal vessel density had no correlation either in subjects without diabetes or in subjects with diabetes with no retinopathy; these variables had a low correlation in people with diabetic retinopathy.
Circularity in subjects without diabetes was lower than that reported with other OCTA devices, which produced 0.81 ± 0.0711, and did not differ from that found with the same equipment in other populations. Kim et al. had reported that there was no difference between mean circularity of subjects without diabetes and people with diabetes without retinopathy and that the mean in subjects with retinopathy was lower12.
Krawitz et al. and his group13 reported that circularity means did not change between subjects with diabetes without retinopathy and people with retinopathy; Lee et al.14, also found no difference between these groups and reported a higher circularity means in subjects with mild non-proliferative retinopathy compared with subjects without diabetes. The circularity means did not change between groups in our study, probably because it only evaluated cases whose foveal avascular zone area did not suggest ischemia.
Fluorescein angiography is more sensitive to detect early retinopathy, but it is a time-consuming invasive method, which is not appropriate as a test to find preclinical changes15; one study found that acircularity measurements (another method of evaluating circularity) were comparable between optical coherence angiotomography and fluorescein angiography with scanning ophthalmoscopy16.
Although the foveal avascular zone did not present characteristics of ischemia, its size increased in subjects with retinopathy, which indicates damage to the parafoveal capillary network. Parafoveal vessel density in the central region decreased in subjects with diabetes compared with subjects without diabetes, and this decrease was more marked in subjects with retinopathy. Even though vessel density decreased, it was not enough to alter the circularity in subjects with diabetic retinopathy.
The correlation between the circularity and the inner vessel density in people with retinopathy was low, and the regression analysis identified a better fit with a cubic model. A steep circularity change was observed with a low vessel density, but the plateau effect that occurred in a range of densities suggests that the most common stages of diabetic retinopathy do not modify the circularity; this variable had a higher positive correlation with the foveal avascular zone area in people with retinopathy than in the other groups, which would indicate that its alterations appear with more advanced stages of retinopathy, where ischemia exists.
One strength of the study is that we only evaluated subjects without foveal ischemic features, which reduced the dispersion of the variables in people with retinopathy; another one was working with the superficial capillary plexus, whose vascular supply is mainly arterial17, has a standardized measurement, and is better to reveal ischemia data than the deep capillary plexus18. In addition, retinal thickness was normal, which prevented vessel density values from distorting due to measuring at different levels, as it can occur in eyes with a retinal thickening. One potential limitation is that the low number of subjects in each retinopathy level restricted a regression analysis between circularity and parafoveal vessel density; the correlation between these variables had not been described, and it will be convenient to analyze it in further studies, to identify a cutoff point for changes of circularity in eyes with retinopathy.
Alterations in the circularity, according to the findings of this study, appear later than the reduction of parafoveal vessel density; in diabetic people, without retinopathy, a normal circularity, as well as a foveal avascular area of normal size, would not indicate that the parafoveal capillary network is healthy. The circularity of the foveal avascular zone does not add information in eyes with mild and moderate diabetic retinopathy when the parafoveal vessel density can be identified nor could it anticipate by itself a decrease in vessel density, when this variable cannot be appropriately measured.