Early retinal microvascular abnormalities in patients with chronic kidney disease

Abstract Objective To evaluate early retinal microvascular abnormalities in patients with chronic kidney disease (CKD) via optical coherence tomography angiography. Methods A cross‐sectional study. Two hundred patients with CKD stage ≧3 were enrolled in the CKD group, and 50 age‐matched healthy subjects were enrolled in the control group. Main outcome measures were the differences in parafoveal vessel densities in the superficial vascular plexus (SVP) and deep vascular plexus (DVP) between the CKD and control groups. Results The mean ages were 62.7 ± 10.1 in the CKD group and 61.9 ± 9.7 (P = 0.622) in the control group. The CKD group had reduced parafoveal vessel densities in SVP (46.7 ± 4.3 vs 49.7 ± 2.9, P < 0.001) and DVP (50.1 ± 4.1 vs 52. 6 ± 2.9, P < 0.001) when compared to those of the control group. In multiple linear regression models, age, diabetes, estimated glomerular filtration rate, and use of anti‐hypertensive drugs were factors associated with vessel density in SVP, whereas age, diabetes, and smoking were factors associated with vessel density in DVP. Conclusion Patients with CKD had reduced vessel densities in parafoveal SVP and DVP, as compared to that of control subjects. Microvasculature in the different retinal layers may be affected by different systemic factors.

2035 with the anticipated increase in diabetic and older population. 5 Chronic kidney disease has been associated with accelerated atherosclerosis, cognitive impairment, cerebrovascular disease, cardiovascular disease, and mortality. [6][7][8][9] In the eye, patients with CKD have higher risks of cataract, glaucoma, AMD, retinopathies, and visual impairment. [10][11][12] The mechanism behind increased ocular diseases in patient with CKD is still being debated. It may be due to CKD and ocular diseases sharing many common systemic risk factors such as aging, DM, hypertension, smoking, and obesity. 11 Alternatively, it could also be due to mechanisms related to CKD, such as increased oxidative stress by decreased filtration of free radical-generating nitrogenous waste products or increased inflammation by activation of the RAS. 11 Earlier studies have revealed that decreased retinal vessel caliber, smaller fractal dimensions, focal arteriolar narrowing, arteriovenous nicking, and opacification of the arteriolar wall can be found in patients with CKD. [13][14][15] These retinal microvascular changes may be useful biomarkers for predicting cardiovascular diseases, 16 cognitive impairment, 17 and aggravation of renal function in patients with CKD. 18,19 However, there is limited information about microvascular alterations at the capillary level. 20 Although increased intercapillary distance in CKD has been shown through use of scanning laser Doppler flowmetry, 20   In patients with both eyes eligible, the eye with better OCTA quality was used for statistical analysis.

| Optical coherence tomography angiography parameters
AngioVue (Optovue RTVue XR Avanti; Optovue Inc.) was used for acquiring OCTA images for this study. The machine uses an 840-nm diode laser source and has an A-scan rate of 70 kHz. A 3 × 3-mm scan, centered on the fovea, was performed in all eyes. An orthogonal registration algorithm was used to produce a 3-dimensional OCTA image. Then using the machine's AngioVue software (version: A2017,1,0,151), the vascular area was automatically segmented into four layers, that is superficial, deep, outer retina, and choroidal. The default segmentation for the SVP includes vasculature between the internal limiting membrane and 10 μm above the inner plexiform layer. For the DVP, this includes the vasculature between 10 μm above the inner plexiform layer and 10 μm below the outer plexiform layer.
The vessel density is defined as the percentage area occupied by all vessels (including terminal arterioles, venules, and capillaries) in a particular region. The data are provided in an ETDRS grid vessel density map ( Figure 1). The foveal region is a 1-mm-diameter circle, and the parafoveal region is a 1-mm-wide circular annulus. The parafoveal region was further divided into the temporal, superior, nasal, and inferior quadrants. The AngioVue software automatically calculates the vessel density of the SVP and the DVP, respectively.
We also evaluated other foveal parameters provided by the machine software including the FAZ size; FAZ perimeter; FAZ a-circularity index; and FD-300. The foveal parameters were determined from an OCTA image of the inner retina microvasculature, which contained both SVP and DVP ( Figure 1).

| Statistical analysis
To compare the demographic data and clinical characteristics of the CKD group with the control group, Pearson's chi-square test was used for categorical variables and the independent sample t test was used for continuous variables. The independent sample t test was also used to analyze differences in vessel densities and foveal parameters between the two groups. Multiple linear regression models with backward stepwise method were used to determine the potential systemic factors associated with the vessel densities in SVP and DVP of all subjects (Model 1) and of patients with CKD (Model 2). Age, sex, smoking status, BMI, DM, use of anti-hypertensive drugs, systolic BP, diastolic BP, CKD, CKD stage, and eGFR were independent variables entered into the regression models whenever applicable. A two-tailed P value <0.05 was considered as statistically significant. Data were analyzed using SPSS Program Package version 17.0 (SPSS Inc.).

| RE SULTS
There were 200 patients enrolled in the CKD group and 50 healthy subjects enrolled in the control group. The mean age was 62.7, SD (±) 10.1, in the CKD group, and 61.9 ± 9.7 in the control group (P = 0.622).
The demographic data and clinical characteristics are summarized in Table 1. There were no significant differences in age-group, sex, diastolic BP, smoking status, cerebrovascular disease, intraocular pressure, or axial length between two groups. However, the mean BCVA in patients with CKD (logMAR: 0.130 ± 0.151, Snellen equivalent 20/27) was slightly worse than that of the control group (logMAR: 0.069 ± 0.103, Snellen equivalent 20/23) (P = 0.001). Figure 2 shows the mean logMAR BCVA in different stages of CKD. There is a trend toward worse visual acuity with more severe CKD.
Chronic kidney disease group also had significantly higher value in BMI, systolic BP, prevalence of DM, number of patients using antihypertensive drug, and prevalence of cardiovascular disease. The  Table 2. There were 116 (66%) patients who used more than one class of drugs.
The fundus pathologies in the 200 eyes in the CKD group are summarized in Table 3. The most common finding was early AMD (20.5%). Diabetic retinopathy was present in 8% of the eyes. Table 4 compares the parafoveal vessel densities and foveal parameters between the control and CKD groups. Parafoveal vessel density was significantly decreased in the CKD group, in both SVP and DVP. This finding was consistent in all four parafoveal quadrants. Localized

| D ISCUSS I ON
There were two major findings in this study. First, retinal microvascular alterations may occur early in patients with CKD, before the onset of visual symptoms. Secondly, the microvasculature in different retinal layers may respond differently to systemic comorbidities.
Chronic kidney disease has been associated with increased visual impairment and ocular diseases in prior epidemiology studies. [10][11][12] In current study, the BCVA in the CKD group was worse than that of the control group. There was also a trend toward visual acuity decreasing with increased CKD severity (Figure 2). We found a high prevalence of early AMD (20.5%) among patients with CKD. Other retinopathies were not very common in this study because we en- Age and DM are important factors negatively associated with vessel density in both SVP and DVP. A prior OCTA study has demonstrated that aging is associated with decreased vessel density in both the superficial and deep capillary plexus in the normal population. 26 Chronic kidney disease may also contribute to premature aging of microcirculation. 27 Diabetes mellitus was present in 91 (45.5%) patients in the current study. It had been well known that the reduction of vessel density is correlated to the severity of diabetic retinopathy. 28,29 Although most of our patients did not have any       In our study, vessel densities in SVP and DVP were associated with different systemic factors. There are three capillary plexuses over the parafoveal area, namely the superficial, intermediate, and deep capillary plexus. 42 In vivo human study showed that each of these three capillary plexuses may have its own feeding arteriolar supply and draining venules. 43 Each capillary plexus has different anatomical structures and may have its own autoregulation. 43 The plexuses may respond differently to systemic condition alteration, such as changes in BP and oxygenation, or to retinal functional hyperemia evoked by a flickering light stimulus. [44][45][46] In our study, antihypertensive drugs and eGFR were associated with vessel density in SVP, but not in DVP. On the contrary, smoking was associated with vessel density in DVP, but not in SVP. Therefore, our results support the hypothesis that microvasculature in different retinal layers may respond differently to varying systemic factors.
There are several limitations in this study. The study is limited by its small sample size and cross-sectional study design. Longitudinal follow-up data were not available. Furthermore, we enrolled patients without visual symptoms into the CKD group. So, our results may reflect early retinal microvascular alterations rather than latestage retinopathies.
In summary, our study demonstrated that patients with CKD had

PER S PEC TIVE
Optical coherence tomography angiography showed that patients with CKD may have rarefaction and morphological changes of retinal microvasculature in the superficial and DVPs. The microvasculature in different retinal layers may respond differently to various systemic factors.

CO N FLI C T O F I NTE R E S T
No authors have any financial/conflicts of interest to disclose.