Morphology of Retinal Vessels in Patients With Optic Nerve Head Drusen and Optic Disc Edema

Abstract

Purpose.: We quantitatively investigated the peripapillary vascular morphology in patients with optic nerve head drusen (ONHD) and optic disc edema (ODE).

Methods.: Computer-based fundus analysis was used to investigate peripapillary vascular morphology, including length, branching, and diameter of arteries and veins calibrated by optical coherence tomography.

Results.: Patients with ONHD showed significantly larger diameters of arteries without branching (P = 0.05), and arteries after primary/before secondary branching (P = 0.04) and secondary venous branching started closer to the optic disc (P = 0.03) compared to healthy controls. The ODE patients had significantly reduced number of small peripapillary veins and larger number of veins without branching compared to ONHD and controls (P = 0.02). Anomalous branching with arterial and venous trifurcation presented in the ODE and ONHD groups, with significant higher prevalence in ODE patients for venous trifurcations compared to ONHD and controls (P = 0.02).

Conclusions.: The diameter of vessels in ONHD patients were significantly larger in arteries without branching (P = 0.05), after primary branching (P = 0.04), and venous branching closer to the disc area (P = 0.03) compared to controls. The ODE patients demonstrated widening of the small peripapillary veins measured by a significantly larger number of veins without branching (P = 0.001 and P = 0.02, compared to controls and ONHD, respectively) and less small veins (P = 0.001 and P = 0.04, compared to controls and ONHD, respectively).

Introduction

Optic nerve head drusen (ONHD) is a condition with an incidence from 0.34% to 2.4%. ^1–4 On clinical examination, ONHD may be misdiagnosed as optic disc edema (ODE), and can lead to multiple expensive and time-demanding examinations, such as ultrasound, fluorescein angiography, computer tomography, and/or magnetic resonance imaging. ^5,6 In addition, optical coherence tomography (OCT) can be used to differentiate between ODE and ONHD, as the peripapillary nerve fiber layer (RNFL) has been shown to be thicker in ODE, while ONHD patients often demonstrate RNFL thinning. ^7–9 Moreover, ONHD also can be detected directly using OCT imaging. ¹⁰  

Digital fundus photos and the development of software for image analysis allow objective measurements of retinal vessels, including diameter, branching, and tortuosity. ^11–13 These softwares have been used to assess the condition of retinal vessels in local (retinitis pigmentosa) and general (diabetes, Alzheimer's disease, blood hypertension) pathology. ^14–20  

Retinal vascular abnormalities have been found in ONHD. These studies have been based only on clinical fundus examination and have not been supported by objective measurements of the vessels. Clinical fundus evaluation of ONHD patients indicated mainly abnormal vascular branching, including vessel trifurcations, increased capillarity on the disc, and the presence of relatively large shunt vessels connecting choroidal and retinal circulation. ^21–23 In contrast, in patients with ODE increased cerebrospinal pressure, including the intraorbital optic nerve sheath and axons of the retinal ganglion cells resulting in a swelling of optic nerve head with retinal hemorrhages and venous dilatation, have been described. ^24–26 We aimed to analyze objectively and compare the difference in retinal vessel parameters, including diameter, length, and branching in the peripapillary area in patients with ONHD, ODE patients, and healthy participants using fundus image analysis. 

Subjects and Methods

Participants

The population cohort for this prospective observational study included 25 ONHD patients (12 visible and 13 buried ONHD; 14 females and 11 males; mean age, 32.86; SD ± 22.50) and 24 ODE patients. Two patients with ODE who also showed ONHD on ultrasound examination were excluded, so that data of 22 patients with ODE were analyzed (21 had idiopathic intracranial hypertension and 1 had cavernous sinus thrombosis; 16 females and 6 males; mean age, 33.82; SD ± 13.93). We also studied 25 healthy volunteers (14 females and 11 males; mean age, 35.94; SD ± 14.54). 

The ONHD and ODE participants were recruited prospectively. Healthy participants were recruited from the University of Leicester and University Hospitals Leicester National Health Service (UHL NHS) staff, their relatives, friends, and partners of clinic patients. 

Complete clinical examination was performed on all subjects, including best-corrected visual acuity, IOP measurements, visual field testing (24-2 pattern, Humphrey Field Analyzer; Carl Zeiss Meditec AG, Jena, Germany), and refraction. Additional examinations, including computer tomography and/or magnetic resonance imaging, and lumbar puncture, were performed if necessary to confirm the diagnosis; eye ultrasound and tests for autofluorescence of the optic nerve were done in all ONHD patients. 

Patients with ONHD had at least two of five findings, previously used in other publications investigating ONHD, ⁷ including visible ONHD, autofluorescence, calcification on ocular ultrasound, normal cerebrospinal fluid pressure, and constant disc elevation. 

The ODE patients were examined during the acute stage of the disc swelling. Magnetic resonance imaging was performed in all patients, and indicated signs of cavernous sinus thrombosis in one patient and was normal in the remaining participants. All participants from the ODE group had opening pressure above 25 cm H₂O on lumbar puncture, with normal fluid composition and documented clinical improvement on follow up examinations. 

Healthy volunteers had best-corrected visual acuity of 0.2 logMAR or better, refractive error −3.0 to +3.0 diopters, and no changes on visual field testing. 

Participants from all groups had normal IOP, no other eye conditions or systemic disease, and no previous intraocular or refractive surgery. 

The study was approved by the local ethics committee and adhered to the tenets of the Declaration of Helsinki. Each participant or parents/guardians of participants gave informed consent. 

Fundus Imaging and Analysis

Color optic disc photographs were taken in mydriasis using an FF 450^plus (Carl Zeiss Meditec AG) fundus camera (30°). Image analysis was performed using ARIA software (available in the public domain at http://sourceforge.net/P/aria-vessels/home/Home/) developed in a Matlab environment (MathWorks, Inc., Natick, MA, USA). ²⁷ This study analyzed the data from one eye for each participant using the best quality image. 

To minimize any possible inaccuracies in diameter (arterial/vein ratio) and length measurements due to optic disc protrusion in ODE and ONHD, the margins of the annulus in which the vessel parameters were measured were set between 4.2 and 8.4 mm ring diameters (where no optic disc swelling was observed in any patient). 

The calibration of the fundus images was performed using ultra-high resolution spectral-domain OCT (Copernicus; Optopol Technology S.A., Zawiercie, Poland) with 3-μm theoretical axial resolution (7 × 7 × 2 mm, 75 B-scans, 743 A-scans per B-scan, fixation target set to image the optic disc). The pixel size of the fundus photograph was calculated by matching the position and size of retinal vessels and optic disc of the fundus photograph with the OCT en face view, where vessels are easily recognizable (Fig. 1A). Each OCT scan was 7.0 × 7.0 mm or 1024 × 1024 pixels. We measured and averaged the size of OCT reconstructed fundus matched with fundus photo in each patient and calculated the pixel size (1 pixel = 6.868 μm). The calibration values per pixel were entered directly into the ARIA program, which then produced calibrated measurements of diameters and lengths of retinal vessels. The calibration factor was the same for each fundus photograph due to the same setting being used during OCT imaging and fundus photography. Individual calibration parameters set separately for each image (by matching to the OCT en face view) were highly reproducible with a coefficient of variation of 0.25%. The calibration process was verified by another author and the mean difference was 1.39%. 

Figure 1

 

Calibration and recognition of the retinal vessels on fundus image of a patient with optic nerve drusen. (A) Calibration of fundus image using fundus reconstruction on OCT, matched position of vessels and disc margins; each OCT scan has 7.00 mm = 1024 pixels. The size of the OCT scan was used to calculate the pixel size of the fundus image. The green arrow on OCT fundus recognition indicates that tomograms were performed using a horizontal raster scan. (B) Fundus photography. (C) ARIA recognition of the borders of vessels of the same patient within the analyzed area of an annulus with 4.2 to 8.4 mm diameters (outside of the area of the prominent optic disc). The blue mark on each vessel is placed automatically by the program and later marked manually to define the type of vessel as artery before primary branching (ABB), artery after primary branching (AAB), vein before primary branching (VBB), vein after primary branching (VAB), vein after primary branching and before secondary branching (VAB/BB), vein after secondary branching (vab), vein without branching continuing through analyzed area (V), and small vein without branching ending within analyzed area (SV). The blue arrow shows the automatically measured diameters in one vessel, which currently is under selection (program automatically calculated average vessel diameter).

The CLRIS processor of ARIA software was used for vessel recognition on digital fundus photos as giving the best vessel recognition. The program transforms the color images (Fig. 1B) into black and white images, outlines the vessels borders, and numbers the vessels (Fig. 1C). ²⁷  

In this study, we have not analyzed vessels tortuosity due to the relatively small area of analysis with short vessels and limited almost equal tortuosity. 

At the end of the vessel analysis, ARIA produces a summary table for each patient with mean (±SD) diameter and length of each vessel within the analyzed area. Automatically numbered vessels then were classified in arteries and veins by one investigator based on the fundus photography (Figs. 1B, 1C). The parameters of arteries/veins without branching (vessel did not have branching and continued out of the border of analyzed area) were measured and analyzed, including arteries/veins before primary branching, arteries/veins after primary branching, arteries/veins after primary branching and before secondary branching, arteries/veins after secondary branching, and small veins (vessels that did not have branching within the analyzed area, Fig. 1C). 

Statistical Methods

The sample size was based on measurements from 25 healthy participants in which the mean artery diameter ± SD was 82.31 ± 15.69 μm. To show a change in 20% (i.e., difference of 16.4 μm, power = 90%, α = 0.05) would require 20 participants in each study group. 

Statistical analysis was performed using SPSS software version 16.0 (SPSS, Inc., Chicago, IL, USA). Normality of the data was tested using the Shapiro-Wilk test. An ANOVA was used to compare statistically the three groups with Bonferroni correction used for multiple comparisons. A value of P ≤ 0.05 was considered to be a statistically significant difference. 

Results

The data were distributed normally. There were no differences in age (P = 0.87) and refractive error (P = 0.80) between all groups. 

Peripapillary Retinal Vessels Diameter/Length

Data analysis of the retinal vessel parameters revealed significant differences mainly in arteries without branching, arteries after primary branching, and small veins after secondary branching (Table 1). 

Table 1

 

Mean (±SD) Diameters and Length of the Retinal Vessels (μm) With Significant Differences Between Groups

Parameter	Value, Mean ± SD			P Value
	ODE	ONHD	Controls	ODE vs. Controls	ONHD vs. Controls	ODE vs. ONHD
Diameter of arteries without branching	83.07 ± 14.41	92.62 ± 18.43	82.01 ± 14.25	0.97	0.05*	0.11
Diameter of arteries after primary branching	87.08 ± 14.35	90.57 ± 14.50	79.88 ± 16.16	0.25	0.04*	0.71
Length of the vein after primary/before secondary branching	819.54 ± 117.81	1112.52 ± 324.53	663.06 ± 376.94	0.65	0.03*	0.29
Diameter of veins after secondary branching	69.05 ± 9.54	99.39 ± 20.05	81.18 ± 21.17	0.54	0.13	0.04*

*  Significant difference between groups, P ≤ 0.05.

The ONHD patients showed the largest diameter of the arteries without branching within the analyzed annulus compared to healthy controls (P = 0.05). The diameter of arteries after primary branching also was significantly larger in the ONHD group compared to controls (P = 0.04). Patients with ONHD also had significantly larger diameters of veins after secondary branching compared to the ODE group (P = 0.04). Patients in the ONHD group also had significantly longer length of the veins after primary branching and before secondary branching compared to controls (P = 0.03), whereas the two other groups had secondary vein branching at similar distance from disc margins. 

Parameters of the arteries before primary, after primary/before secondary, and after secondary branching; veins without branching, before and after primary branching, as well as arterial/vein ratio in patients with ODE, ONHD, and healthy controls did not reveal any statistical differences. 

Peripapillary Retinal Vessels Branching and Number of Vessels

The analysis of the vessel branching and number of vessels (Table 2) in three groups demonstrated significant differences in number of vein trifurcations, veins without branching, and small peripapillary veins (Figs. 2, 3). 

Figure 2

 

Fundus photo (A, C, E) and ARIA fundus recognition (B, D, E) of a patient with ONHD, a patient with ODE, and healthy control indicating veins without branching and small peripapillary veins. Red arrow shows small peripapillary vein without branching recognized by ARIA ending within analyzed area. Yellow arrow indicates vein without branching continuing through analyzed area.

Figure 3

 

Fundus photo of patients with vessel trifurcations. (A, B) Fundus image of patients with optic disc edema. (C) Fundus image of patient with optic nerve head drusen. Black arrow indicates venous trifurcation. Blue arrow indicates arterial trifurcation.

Table 2

 

Different Types of Branching and Number of Veins (%) in Analyzed Groups

Parameter	Value, Mean ± SD			P Value
	ODE	ONHD	Controls	ODE vs. Controls	ONHD vs. Controls	ODE vs. ONHD
Arteries without branching	6.55 ± 3.23	5.16 ± 3.13	5.44 ± 2.72	0.65	1.00	0.37
Veins without branching	7.55 ± 2.77	5.72 ± 1.95	4.84 ± 1.82	<0.001*	0.48	0.02*
Small peripapillary veins	0.09 ± 0.43	1.08 ± 1.53	1.72 ± 1.70	<0.001*	0.31	0.04*
Veins without branching and small peripapillary veins	7.64 ± 2.75	6.80 ± 2.40	6.56 ± 2.10	0.40	1.00	0.72
Arterial bifurcations	2.95 ± 1.56	3.88 ± 1.67	3.64 ± 2.00	0.56	1.00	0.23
Vein bifurcations	2.50 ± 1.30	3.44 ± 2.02	3.12 ± 1.99	0.74	1.00	0.25
Arterial trifurcations	0.14 ± 0.35	0.04 ± 0.20	–	0.13	1.00	0.45
Vein trifurcations	0.18 ± 0.39	–	–	0.02*	1.00	0.02*

*  Significant difference between groups, P ≤ 0.05.

Participants in all groups had a similar number of arteries without branching and arterial/venous bifurcations. The number of small peripapillary veins was significantly lower in the ODE group compared to two others (P < 0.05). Patients with ODE had significantly more veins without branching (P < 0.001 and P = 0.04 for healthy controls and ONHD, respectively). The number of small veins and veins without branching added together did not indicate any significant difference in all three groups. 

Arterial trifurcations were found in the ODE and ONHD groups, with more than twice the amount found in ODE patients. However, the difference between ODE and ONHD was not statistically significant. Venous trifurcations were found only in patients with ODE (P = 0.02 for ONHD and controls). 

Discussion

In this study, we compared objectively diameters and lengths of retinal arteries and veins in the peripapillary area of patients with ODE, ONHD, and healthy controls using an image analysis of fundus photography. We found quantitative evidence for significant differences in arterial diameters, length, and number of venous vessels. 

The majority of studies investigating the fundus features of ONHD and ODE have concentrated on optic disc appearance aiming to detect drusen presence using fluorescein angiography or autofluorescence with different color filters. ^2,28 However, little is known about objective parameters of arteries and veins, including diameter, types, and number of branches in the peripapillary area in patients with these pathologies. A few descriptive studies using fundus pictures have observed subjectively features of the peripapillary retinal vessels in patients with ONHD, and found pronounced tortuosity, dilated veins, and abnormal early branching. ²⁹  

Our study showed that patients with ONHD had significantly larger diameter of arteries without branching and after primary branching compared to controls (P = 0.05 and P = 0.04, respectively). A similar pattern was noted in the diameter of veins after secondary branching with significant increase in the ONHD compared to the ODE group (P = 0.048). Patients with ONHD also demonstrated significantly longer veins after primary branching and before secondary branching compared to healthy controls (P = 0.03). Our results showed that in patients with ONHD vein branching starts closer to the disc area compared to controls. These findings possibly may indicate a developmental abnormality as suspected by Antcliff and Spalton. ²⁹  

In ODE, more severe symptoms, including enlargement and increased tortuosity of the veins, peripapillary hemorrhage, and missing venous pulsation as a result of axonal swelling, have been described on fundus examination. ³⁰  

Our results indicated that patients with ODE have a significantly larger number of veins without branching, and less small veins compared to ONHD and controls. Interestingly, in ODE individual diameters of the large veins did not show an increased size in the peripapillary area when measured objectively. There was no difference in the total number of veins without branching (number of small peripapillary veins added to number of veins without branching in the analyzed area). This may indicate that patients with ODE have widening of the small peripapillary veins, which would have been recognized by the ARIA software as veins without branching instead of small peripapillary veins, explaining the reduced number of small peripapillary veins. Therefore, the widened veins are likely to explain the increased number of larger veins without branching measured in our study. So, this means that, while the individual diameters of larger veins were not significantly different compared to ONHD and control groups, the combined venous diameter (for all veins in the peripapillary area) was significantly bigger in the ODE group due to the dilatation of small peripapillary veins. The increased size of small veins (the number of large veins without branching recognized ARIA software) could be a mechanism to help to reduce increased intravenous pressure, which could explain why patients with ODE did not have any difference in diameter of large veins, and had significantly smaller diameters of veins after secondary branching compared to ONHD and healthy control groups. As previous studies have described a widening of large veins, ^24–26 it is possible that veins in ODE patients are wider on the disc “surface” or in the area close to the disc and not within the analyzed area. 

We identified artery and vein trifurcations not only in the ONHD group, as shown in the literature, ^21–23 but also in ODE patients. A surprising finding was that the absolute number of trifurcations was higher in the ODE group compared to ONHD. We cannot completely exclude the possibility that patients with ODE also had buried ONHD. However, ocular ultrasound performed in all ODE patients with anomalous branching did not show any calcium deposits. The number of vein trifurcations was significantly different between the groups (P = 0.02). The absence of trifurcations in healthy controls may support the theory that this anomalous branching is a result of mild chronic (in case of drusen with development of arterial trifurcations), and more acute and severe (in case of edema with development of venous trifurcations) reduced vascular perfusion. 

Limitations of the Study

One limitation of the study is that we have measured vascular anomalies in the peripapillary retina rather than in the disk itself. This is because 2D fundus photography cannot be used to interpret the complex 3D geometrical conformation of vessels in and around prominent optic discs. The OCT and, in particular, Doppler imaging methods may provide a solution to this problem in future studies. Another limitation of the study is that we have not demonstrated whether the observed changes in retinal vasculature have clinical or pathoetiologic significance. It would be important in future studies to understand the mechanisms behind these changes as well as determining whether they have any diagnostic or prognostic value. 

Fundus photography captures the vessel architecture in a single moment in time. However, ODE is a dynamic process depending on the duration, rapidity, and severity of raised CSF pressure. In this study we used only acute stage ODE when raised CSF pressure is more severe. 

In summary, our study objectively describing different peripapillary vascular morphology in patients with ONHD, ODE, and healthy controls showed abnormal vessels architecture in ONHD patients and dilatation of small peripapillary veins in ODE patients. 

Acknowledgments

Supported by the Ulverscroft Foundation, Leicester, United Kingdom. The authors alone are responsible for the content and writing of the paper. 

Disclosure: A.V. Pilat, None; F.A. Proudlock, None; R.J. McLean, None; M.C. Lawden, None; I. Gottlob, None