Retinal Vessel Phenotype in Patients with a History of Retinal Vein Occlusion Open Access

Retinal vein occlusions (RVO) have prevalence rates ranging from 0.1 to 1.1% of the aging population, and bilateralization occurs in 5% of the cases over a 1- to 3-year period. Virchow’s triad includes the 3 broad categories of factors that are thought to contribute to thrombosis, as in central retinal vein occlusion (CRVO): hypercoagulability, hemodynamic changes (stasis and turbulence), and endothelial injury/dysfunction. Two main risk factors have been identified: systemic hypertension and open-angle glaucoma [1]. The pathogenesis of branch retinal vein occlusion (BRVO) is multifactorial in origin and not completely defined, with a possible combination of mechanical compression, degenerative changes in vessel walls, and/or hypercoagulable factors [1]. The arteriosclerotic changes (at the sites of arteriovenous crossing) are believed to result in venule occlusion through endothelial cell damage, thrombosis, and focal venous narrowing at sites of arteriovenous nicking [2].

The study of vessel caliber and tortuosity merits further consideration since it is not known whether a patient’s vessel network promotes the onset of RVO, independently of association with cardiovascular risk factors. We hypothesize that the retinal phenotype (vasculature morphometry) may be associated with the occurrence of RVO. The analysis of fundus camera images offers a noninvasive measurement method to study the vascular retinal network. The vascular morphological phenotype (including tortuosity and fractal dimension) provides information on the architecture and geometry of the vessel network, which determines the efficiency of blood circulation. We used the VAMPIRE (Vessel Assessment and Measurement Platform for Images of the Retina, Universities of Edinburgh and Dundee) software, which enables a detailed quantitative analysis of the vascular morphometry [3]. VAMPIRE has been used in several studies on retina biomarkers for lacunar stroke, cognition, dementia, hypertension, and cardiovascular diseases among others [4]. More recently, VAMPIRE has been used to investigate the retinal vessel phenotype in ocular disease, such as nonarteritic ischemic optic neuropathy and glaucoma [5, 6].

There is evidence that vascular parameters such as central retinal vein equivalent (CRVE), central retinal artery equivalent (CRAE), arterio-venule ratio (AVR), and tortuosity are considered to be correlated between right and left eyes. We studied the vascular phenotype of the unaffected contralateral eyes of patients with RVO and compared it to a control group. Previous studies evaluated the contralateral eyes of RVO and showed the subtle abnormalities in eyes with normal appearance: late peripheral retinal leakage using peripheral fluorescein angiography [7], nonperfused foveal capillaries using adaptive optics scanning light ophthalmoscope [8], and decreased retinal nerve fiber thickness in the inferior and superior temporal quadrants using optical coherence tomography [9]. Therefore, we aimed to characterize the phenotype of retinal vessels using CRAE, CRVE, tortuosity, and fractal dimension in the nonaffected contralateral eye of patients with a history of CRVO and BRVO and to compare this phenotype with that of a control group matched for age, sex, systemic hypertension, and diabetes.

Thirty-four patients suffering from CRVO and 15 patients from BRVO were included in this multicentric case-control study in 2016 and 2017 (17 in the Ophthalmology Department of Grenoble Hospital and 22 in Dijon University Hospital). The inclusion criteria of patients were age over 18 years, diagnosis of CRVO or BRVO, and patients with ametropia (spherical equivalent) ≤3 diopters (D). Exclusion criteria for the contralateral eye of the RVO eye were the presence of additional ocular diseases and fundus pictures with low quality (n = 14) due to cataract or poor fixation.

Forty-nine controlled matched subjects (1:1) were prospectively included at the Ophthalmology Department of Grenoble University Hospital. The inclusion criteria were age over 18 years, patients without any ophthalmologic medical history, and patients with ametropia ≤3 D. Exclusion criteria were adults under guardianship or unable to consent, patients with ametropia >3 D, and ocular disease. This control population was matched to the RVO population for age (5-year interval), sex, and systemic hypertension and diabetes.

Thirty- or 45-degree fundus camera images of the right eye were acquired, centered on the optic nerve and the macula, using a nonmydriatic camera: Visucam 200 (Carl Zeiss Meditec^TM, France, resolution of 2,124 × 2,056 pixels) or CR2 (Canon^TM Europa, Amstelveen, The Netherlands, resolution of 4,752 × 3,168 pixels).

VAMPIRE measures semiautomatically morphological parameters of the retinal vessels. First, the optic disc (OD) contour and the macula center are located. This enabled the definition of the usual retinal coordinates (x axis through OD and macula centers and origin in the OD center) and circular zones around the OD, namely, zone A (between OD center and 0.5 optic disc diameter [ODD]), zone B (between 0.5 and 1 ODD), and zone C (between 0.5 and 2 ODD), shown in Figure 1. Manual correction can be performed efficiently when the OD or fovea has been incorrectly identified automatically. Vessels are subsequently detected and labeled as arterioles or venules semiautomatically (Fig. 1).

Here, we used CRAE, the mean of the widths of the 6 largest arteries (using the revised Knudtson formulas) [10]; CRVE, as above but for the 6 largest venules [9]; AVR (=CRAE/CRVE); FD of the vascular network in zone C, a measure of geometric complexity of the pattern of the vessels in zone C, including the degree of branching complexity and vascular tortuosity. AVR, CRAE, and CRVE were computed in zone B and fractal measure and vascular tortuosity analysis in zone C. Raw measurements of CRAE and CRVE were in pixels. Tortuosity may be associated with high blood flow, angiogenesis, and blood vessel congestion.

Following previously reported studies, the pixel-to-mm conversion factor was obtained by dividing the average vertical ODD (over all images, acquired with the same camera at the same resolution) by the assumed average of the disc diameter in microns (1,850 μm) [10-12]. Excellent intra- and interoperator reproducibility (above 0.82 and 0.91, respectively) for the 2 operators (R.S. and O.G.) participating in the study was obtained with 100 fundus images from healthy subjects (n = 30) or patients suffering from glaucoma or contralateral eyes from RVO (n = 70) twice by 2 different operators (R.S. and O.G.).

Statistical analysis was performed using the Statistical Package for the Social Sciences program (SPSS 17.0 for Windows, Chicago, IL, USA). Quantitative data were expressed as mean and standard deviation, after checking the normality of data. The intraclass correlation coefficient with a confidence interval of 95% and a Bland-Altman plot were done for the reproducibility study. Mean comparisons were studied using the 2-tailed t test. Statistical significance was set at p < 0.05.

The inter- and intraoperator reproducibility (ICC) was calculated from a series of 100 fundus images (Table 1) and was considered as excellent. Interoperator ICC ranged from 0.918 to 0.992 and intraoperator ICC from 0.823 to 0.98.

This population included 22 males and 41% had systemic hypertension. The vascular parameters of the contralateral eye were compared with the control eyes (Tables 2, 3). No significant difference was found between the control group and the CRVO group for age (p = 0.75), refraction (p = 0.08), CRAE (p = 0.06), CRVE (p = 0.3), and AVR (p = 0.6). Contralateral eyes of CRVO exhibited a significantly higher arterial and minimum arterial tortuosity values (p = 0.012). There was no statistically significant difference between the 2 groups for fractal dimension analysis (p = 0.26).

This population included 8 males and 35% had systemic hypertension. The vascular parameters of the contralateral eye of 15 patients were compared with a group of 15 matched control eyes (Tables 4, 5). Contralateral eyes of patients with a history of BRVO had a higher CRAE (p = 0.02), AVR (p = 0.006), and minimal arterial tortuosity (p = 0.05). Fractal analysis showed that contralateral eyes of BRVO had significantly higher values of fractal dimension, such as D0a (p = 0.005), D1a (p = 0.008), D2a (p = 0.008), D0tot (p = 0.01), D1tot (p = 0.01), and D2tot (p = 0.01).

This original study showed that contralateral eyes exhibit (a) in cases of CRVO, a greater arterial tortuosity values, and (b) in cases of BRVO, a greater CRAE, AVR, arterial tortuosity, and higher values of fractal parameters, as compared with the control group matched for age, sex, systemic hypertension, and diabetes. These exploratory results strongly suggest that the morphology of the retinal vasculature may be associated with the occurrence of RVO, independent of known risk factors.

Epidemiological studies have shown that contralateral eyes of patients with a history of CRVO in 1 eye have (a) a significantly increased risk of developing RVO compared with the general population (7% probability within 4 years) [13, 14] and (b) an increased number of nonperfused capillaries near the foveal avascular zone and decreased perfused foveal microvascular density [8]. Therefore, studying contralateral eyes may elucidate the early pathological changes signaling a future occlusive event.

The significant finding in our CRVO population was the abnormal higher arterial tortuosity in patients with a history of CRVO. This abnormality could increase the blood resistance and modify the blood rheology in some retinal vessels and may indicate a predisposition for occurrence of RVO. In the literature, reduced arterial tortuosity is usually reported with age, hypertension, and BMI, whereas increased venular tortuosity is associated with a younger age, higher BP, and lower HDL [15].

The other important finding was the absence of changes of CRAE, CRVE, AVR, and fractal dimension in patients as compared with the matched control group. Notwithstanding the pilot size of our cohort, this could be a major finding since these parameters are correlated with age, BMI, and history of hypertension [16], which were all taken into account with matching populations.

Patients with a history of BRVO had higher CRAE and AVR, whereas CRVE was not significantly different when compared with the control group. A larger arteriolar caliber has been previously reported with current cigarette smoking [17], the activity of plasma GPx-3, a major player in oxidative stress regulation [17], calcium channel blockers, combined use of aspirin and antihypertensive agents [18], and higher levels of serum glucose (glycated hemoglobin HbA1c) [18]. These factors should be therefore considered in future studies. In the present study, only 6 patients out of 49 took calcium channel blockers (4 CRVO and 2 BRVO).

The increase of AVR is probably due to the significant increase in CRAE, whereas CRVE is comparable in both groups. On the other hand, matching both groups with the presence or absence of hypertension may explain why we did not measure the well-known decrease in CRAE associated with hypertension [19, 20]. One previous study investigated the vascular phenotype of eyes (n = 25) before development of BRVO and found that AVR was significantly smaller in eyes, whereas CRAE and CRVE did not vary significantly as compared with the fellow eyes which did not develop BRVO. AVR of the contralateral eye of BRVO in our study (0.76 ± 0.05) was close to that found (0.73) in fellow eyes in the study of Kawasaki et al. [21]. On the other hand, severe arteriovenous nicking, isolated retinopathy, and a smaller angle at the crossing site were more prevalent in eyes which developed BRVO compared with fellow eyes. Our study is complementary to this latter study since we studied only fellow eyes at the time of BRVO, which is different from a longitudinal study. We also compared the fellow eyes to a control group matched for hypertension, age, and diabetes.

We found that contralateral eyes of BRVO have higher fractal dimension of the vasculature network. The FD index quantifies the complexity and density of the vessel branching pattern visible in a fundus image. FD variations are an indicator of deviations from the normal or optimized network and, when reduced, a potential marker of cardiovascular (coronary disease) or neurologic (stroke) disease. Our results suggest that these patients maintain a high level of complexity even if hypertension and age are considered.

We did not find an overall increase in tortuosity, whereas the minimal arterial tortuosity was increased significantly in these patients. This suggests a change in the distribution of arterial tortuosity among arterioles within unaffected contralateral eyes in both CRVO and BRVO populations. This difference is not likely associated with diabetes and systemic hypertension since these factors were well balanced with the control group. Factors known to have an impact on tortuosity were not significantly involved in our series (refraction and race). The significance of this result should be confirmed in further studies and other associated factors should be investigated, such as cardiovascular diseases (such as ischemic heart disease [16], obesity [15], dyslipidemia [22], and renal dysfunction [23]).

Strengths of this study were (a) the excellent inter- and intraobserver agreement (above 0.9 and 0.8, respectively) using VAMPIRE, consistent with that described in other studies using this software [24, 25], and (b) matching of the control group on potential confusion factors (hypertension, age, and diabetes). This control group was matched not only for well-known factors influencing vascular parameters (age) [15] and systemic hypertension but also for risk factors of RVO (diabetes and systemic hypertension) [1]. No RVO subject in our cohort had glaucoma.

We acknowledge several limitations of this pilot study. First, the modest (pilot size) size of the cohort. Second, the absence of measurements of blood pressure or the assessment of the severity of hypertension. Many studies reported association between lower monofractal dimensions and higher mean arterial blood pressure [26, 27]. Association between lower fractal dimensions and diastolic blood pressure was also reported [26, 27]. Other clinical variables such as history of cardiovascular or neurovascular disease, BMI, and dyslipidemia may alter the retinal vascular phenotype in the RVO and control groups [28].Third, the use of images taken at different resolutions. To partially correct for these magnification differences between cameras, a calibration factor was obtained by measuring the average vertical height of the OD (in pixels) in all retinal images with the same resolution and the same camera and dividing the assumed average of the disc diameter (1,850 μm, average disc diameter measured for the Caucasian subjects) with the average value ODD measurements in pixels in the same series. This technique has been extensively used in different studies using SIVA [12] and IVAN [29]. Fourth, associations between smaller retinal vessel diameters and longer axial length and more myopic refraction were reported previously [30, 31]. The effect of refraction was limited in our study since the spherical equivalent was 0.5 ± 1.5 D in the RVO group and 0.5 ± 1.5 D in the control group. Ideally, it would be useful to measure axial length in order to convert pixels into microns using Bennet’s formula [32].

Fifth and finally, we assumed that contralateral eyes reflected the phenotype of both eyes. We based this hypothesis on previous studies showing good correlations for vascular parameters between right and left eyes. Using VAMPIRE (Kirin, Mirna: Genetic Analysis of Retinal Traits. University of Edinburgh, Ph Thesis, 2013, https://era.ed.ac.uk/handle/1842/9619?show=full), monofractal and multifractal dimensions of the retinal vasculature were also tested for left and right eye correlation (n = 130 images of healthy eyes). Correlation values were higher compared to the branching parameters. Left-right eye correlations were 0.78 for monofractal dimensions and 0.50 for multifractal dimensions. We note however that the degree of symmetry of the morphology and its quantitative characterization are the object of current discussion [33].

This pilot study in patients with a history of CVRO or BRVO suggests that both diseases are not associated with the same retinal vascular phenotypes in the contralateral eye. Contralateral eyes of CRVO eyes exhibited a higher tortuosity, whereas contralateral eyes of BRVO eyes had higher CRAE, AVR, and fractal dimension. These factors should be studied further in a longitudinal, larger study to better understand the potential risk factors associated with the retinal phenotype.

Written informed consent was obtained from the subjects after explanation of the experiment. The study followed the Declaration of Helsinki guidelines for research involving human subjects and was approved by the local Institutional Review Board (IRB# 5921).

C. Chiquet: Thea, Allergan, and Horus. A. Bron: Aerie, Allergan, Bausch Lomb, Santen, and Théa. F. Aptel: Aerie, Allergan, BaushLomb, Glaukos, Horus, Quantel, Santen, and Thea. C. Creuzot-Garcher: Allergan, Bayer, Alcon, Novartis, Roche, Théa, and Bausch and Lomb. R. Semecas, L. Arnould, O. Gavard, T. Mautuit, T. MacGillivray, S. Hogg, and E. Trucco: none.

The study was funded by the Association de Recherche et de Formation en Ophtalmologie (ARFO) and Fondation de France (Berthe Fouassier grant).

All authors made substantial contributions to the conception or design of the work; or the acquisition, analysis, or interpretation of data for the work; drafting the work or revising it critically for important intellectual content; final approval of the version to be published; they agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.