Three-Dimensional Optical Coherence Tomography Evaluation of Vascular Changes at Arteriovenous Crossings

Abstract

Purpose.: To study the three-dimensional morphologic features of retinal arteriovenous crossings with optical coherence tomography (OCT) and elucidate the vascular changes associated with crossing phenomena as seen on fundus photographs.

Methods.: We examined 150 consecutive eyes with no ocular disease. In each eye, fundus photographs were taken, and one randomly selected arteriovenous crossing was examined by OCT. The OCT analysis was performed by using sequential thin sections along and perpendicular to the retinal vessels.

Results.: The OCT analysis of these arteriovenous crossings showed that the veins abruptly changed their directions to pass the artery and frequently displayed focal luminal narrowing with no compression or flattening. The OCT measurements revealed that the veins narrowed by 21.0% ± 12.9% at the crossings. The degree of narrowing correlated positively with the diameter of the crossing arteries (r = 0.419, P < 0.001). On fundus photographs, crossing phenomena were observed in 103 of the 150 selected crossings. Venous narrowing measured by OCT was more severe in eyes with crossing phenomena on fundus photographs (P < 0.001). Four types of crossing phenomena were observed: concealment, tapering, deflection, and humping. Venous narrowing rates were similar among all four types. Although the subjects with deflection or humping phenomena were more likely to suffer from hypertension, the mean venous narrowing rate at such crossings was similar to that observed with the other crossing phenomena.

Conclusions.: Arteriovenous crossings exhibited focal narrowing of the venous lumen with no compression or flattening. Increased venous narrowing and larger arteries were observed at crossings with crossing phenomena.

Introduction

Owing to the transparency of the ocular media, the retinal vasculature can be observed directly and noninvasively. As early as the late 19th century, Gunn ^1,2 suggested that retinal vascular abnormalities seen during the fundoscopic examination reflect the pathology associated with systemic vascular diseases and hypertension. Classically, retinal vascular changes (i.e., arteriolar narrowing, caliber irregularity, light reflex alterations, and arteriovenous crossing phenomena) are categorized according to the Keith-Wagener-Barker and Scheie classification systems. Such changes have been used as prognostic markers of cardiovascular disease. ^3–5  

Crossing phenomena are based on features of the arteriovenous crossing, including concealment, tapering, deflection, and humping. These features are markers for arteriosclerosis and hypertension. On fundus photographs that capture severe crossing phenomena, venous blood flow sometimes appears to be extremely narrow or even to be interrupted at arteriovenous crossings. Although previous histologic studies have investigated the pathogenesis of crossing phenomena, ⁶ the morphologic and functional changes associated with each crossing phenomenon are still unclear. Furthermore, the qualitative and subjective criteria for the evaluation of arteriovenous changes detract from the validity of intersubject comparisons. ⁷  

In the past decade, advances in optical coherence tomography (OCT) image resolution and acquisition speed have allowed for more detailed observations of retinal architecture. In contrast to fundus photography or angiography, OCT allows the researcher to image the retina and vasculature in perpendicular sections. Moreover, sequential thin sections allow for three-dimensional evaluation of the retinal vessels. ^8,9 Few previous studies have examined the morphology of the retinal vasculature at the crossing sites in living human eyes. The purpose of the current study was to study arteriovenous crossings by using sequential thin OCT sectioning. The results elucidated the actual vascular changes associated with crossing phenomena as they appear on fundus photographs. 

Methods

The Ethics Committee at Kyoto University Graduate School of Medicine approved this prospective study, which was conducted in accordance with the tenets of the Declaration of Helsinki. Written informed consent was obtained from each subject before the start of the study. 

This prospective study consisted of 150 eyes with no ocular disease from 150 consecutive subjects (70 men and 80 women) aged 50 years or older. The retinal vessels were examined by using OCT at the Department of Ophthalmology of Kyoto University Hospital between June 2012 and December 2012. Eyes with a history of intraocular surgery except for cataract surgery and those with any co-existing ocular disease (e.g., glaucoma, diabetic retinopathy, retinal vein occlusion, age-related macular degeneration, or a senile cataract resulting in poor-quality OCT images) were excluded from the current study. 

After a comprehensive medical interview, refractive error measurements and keratometry were performed, and 45° digital fundus photographs were obtained (3216 × 2136 pixels, TRC-50LX; Topcon, Tokyo, Japan) after pupil dilation. In each eye, one arteriovenous crossing within 1 to 3 disc diameters from the optic disc center was selected. The selected crossing was examined with Spectralis HRA+OCT (Heidelberg Engineering, Heidelberg, Germany). Longitudinal and cross-sectional images of major retinal blood vessels were obtained (minimally 25 sections of 5°, respectively). Each sequential section was obtained by averaging the information from at least 25 scans of the same location. 

As reported previously, ^9,10 cross-sectional OCT images depict major retinal vessels as round, hyporeflective configurations with two distinctive hyperreflective lines and two circular hyperreflectivities in a line (Fig. 1). The innermost and outermost hyperreflective lines are derived from the vessel walls. The arterial walls generally have higher reflectivity than the venous walls. Blood flow is depicted as intravascular paired circular reflections, which are frequently hourglass shaped. In longitudinal OCT sections of retinal vessels, these four reflections formed two lines and two bands. The innermost and outermost lines are derived from the vessel walls, and the two intermediate bands are derived from the bloodstream (Fig. 1). Intravascular paired reflections are explained as resulting from the shear rate of blood flow, erythrocyte arrangement, and OCT fringe washout. ¹⁰ Intravascular reflectivity was reported to decrease with increasingly oblique angles of observation and to be absent when blood flow was parallel to the line of sight. 

Figure 1

 

Cross and longitudinal sections of retinal vessels obtained with OCT. (A) Optical coherence tomography sections of retinal vessels were obtained along the arrows. (B) A cross-sectional OCT image taken along the arrow (b) shows major retinal vessels as round hyporeflective configurations with two distinctive hyperreflective lines and two circular hyperreflectivities in a line. The innermost and outermost hyperreflective lines are derived from vessel walls. Blood flow is seen as intravascular paired circular reflections, which are frequently hourglass shaped. (C) Optical coherence tomography image in (B) with indications of vascular walls and outlines. (D) In a longitudinal OCT section of a retinal vessel (d), four hyperreflectivities form two lines and two bands. (E) Optical coherence tomography image in (D) with indications of vascular walls. Red lines indicate the hyperreflectivities of the arterial wall, and blue lines indicate the hyperreflectivities of the venous wall. White dotted lines outline the vessel.

In the current study, the inner or outer diameters of the retinal vessels were determined by measuring the perpendicular distance between the inner or outer edges of the two hyperreflective lines from the vessel wall, respectively. The vessel wall thickness was calculated according to the following formula: Wall Thickness = (Outer Vessel Diameter − Inner Vessel Diameter)/2 (Fig. 2). ⁹ At the crossings, inner and outer arterial diameters were measured. Owing to the difficulty of imaging venous walls at crossings, only the inner diameter was measured. 

Figure 2

 

Calculation of arterial wall thickness and the degree of venous narrowing. (A) On OCT sections, the inner and outer diameters of the retinal vessels (a1 and a2) were determined by measuring the distances between the inner and outer edges of the two hyperreflective lines from the vessel wall. Thickness of the retinal arterial wall was calculated by using the following formula: Arterial Wall Thickness = (a2 − a1)/2. (B) At the arteriovenous crossing site, the inner diameters of the vein were measured at the same distance (300–500 μm) proximal (v2) or distal (v3) from the crossing site. Presumed inner venous diameter (v4) at the crossing site was calculated by using the following formula: v4 = (v2 + v3)/2. The degree of venous narrowing at the crossing was calculated according to the following formula: Venous Narrowing Rate = (v4 − v1)/v4 × 100 (%). v1, actual inner venous diameter.

Focal narrowing of the venous lumen was observed at arteriovenous crossings. The inner diameters of the vein were measured at the same distance (300–500 μm) proximal or distal from the crossing. The presumed inner venous diameter at the crossing was calculated by averaging these two measurements. The degree of venous narrowing at the crossing was calculated according to the following formula: Venous Narrowing Rate = (Presumed Inner Venous Diameter − Actual Inner Venous Diameter at Crossing Site)/Presumed Inner Venous Diameter × 100 (%) (Fig. 2). A single, blinded retinal specialist (KK) obtained these measurements. Presumed inner arterial diameter was calculated in the same manner. 

The presence or absence of crossing phenomena was assessed at the selected crossing by using fundus photographs obtained at the same visit. At artery-over-vein (AV) crossings, the crossing phenomenon was determined to be positive when concealment, tapering, or deflection of the vein was observed. At vein-over-artery (VA) crossings, the crossing phenomenon was determined to be positive when venous humping was observed. The concealment phenomenon denotes a vein appearing to stop abruptly on either side of the artery. The tapering phenomenon means that the vein appears to taper down either side of the artery. The deflection phenomenon at AV crossings and humping phenomenon at VA crossings indicate the vertical deviation of a vein from its normal course at crossings, which has been referred to as the Salus sign. Two retinal specialists (KK, YM) independently assessed crossing phenomena in a masked fashion. In cases of disagreement, the third retinal specialist (AT) made the final decision. 

In this study, we investigated the association of vascular parameters and systemic disease with crossing phenomena. Hypertension, diabetes mellitus, and dyslipidemia were defined by use of the corresponding medications. We also focused on each type of crossing phenomenon and investigated the differences among them. Statistical analyses were performed by using PASW statistics software (version 18.0; SPSS, Chicago, IL). The unpaired t-test, one-way analysis of variance, and χ² test were performed to compare parameters among subject groups. Multiple comparisons were adjusted by using the Bonferroni correction. A P value of less than 0.05 was considered to be statistically significant. 

Results

Table 1 presents the demographic information for the subjects included in the study. The mean age was 69.6 ± 8.7 years (range, 51–92 years). In the current study, no eye had signs of ocular disease. In each eye, one crossing was randomly selected and examined; 118 were AV crossings, and 32 were VA crossings. 

Table 1

 

Demographic and Ocular Characteristics of Subjects Included in the Current Study

Subject no.	150
Sex, male/female	70/80
Age, y	69.6 ± 8.7
Spherical equivalence, diopters	0.0 ± 1.9
Hypertension, ±	70/80
Diabetic mellitus, ±	15/135
Dyslipidemia, ±	39/111

Morphologic Features of Arteriovenous Crossings Visualized on OCT Images

Sequential thin-section OCT clearly depicted the three-dimensional morphologic features of the retinal vasculature. Major retinal arteries and veins ran within the retinal nerve fiber layer, occasionally in the inner plexiform layer. At the crossings, retinal arteries ran straight within the inner retina, while veins abruptly changed their direction and passed under or over the retinal artery (Supplementary Videos S1, S2). Retinal veins sometimes appeared to run deep nearby the junction of the inner segment and outer segment line (Fig. 3). At the VA crossings, some veins seemed to protrude into the vitreous cavity to pass over the arteries (Fig. 3). 

Figure 3

 

Vascular architecture at arteriovenous crossings as examined with OCT. (A, F) Fundus photographs of the AV and VA crossings (arrows). (B, G) Optical coherence tomography sections at the crossing, obtained along the retinal veins. (D, I) Optical coherence tomography sections at the crossing, obtained along the retinal arteries. (C, E, H, J) Optical coherence tomography images with indications of vascular outlines from the upper OCT images. (A–E) Artery-over-vein crossing. A retinal artery runs straight within the inner retina. At the crossing, the retinal vein abruptly changes direction and passes under the retinal artery. This retinal vein runs deep nearby the junction of the inner segment and outer segment line. (F–J) Vein-over-artery crossing. A retinal artery runs straight within the inner retina. At the crossing, the retinal vein seems to protrude into the vitreous cavity to pass over the arteries. The red lines indicate arterial outlines and the blue lines indicate venous outlines.

There was no change in arterial lumen caliber at these crossings. Mean presumed inner arterial diameter and actual inner arterial diameter were 80.4 ± 18.6 μm and 81.7 ± 18.9 μm, respectively. The coefficient of correlation was 0.966 (P < 0.001). However, focal narrowing of the venous lumen was observed at most crossings; inner venous diameter was reduced by 21.0% ± 12.9% (range, 0%–54.2%) of the presumed inner venous diameter. However, no compression or flattening was seen in the veins at these crossings (Fig. 4). Some eyes showed a discrepancy between the venous width depicted on the fundus photograph and the venous lumen as visualized on OCT sections. On some fundus photographs of eyes with severe crossing phenomena, the bloodstream seemed to be extremely narrow or even to be interrupted. The OCT sections of such eyes revealed that the actual venous lumen was round and larger than depicted in the fundus photograph (Fig. 4). 

Figure 4

 

Venous narrowing rather than compression at the arteriovenous crossings. (A, F) Fundus photographs of an AV crossing, showing crossing phenomena (arrows). (B, G) Optical coherence tomography sections obtained along the retinal veins at crossings. (D, I) Optical coherence tomography sections obtained at the crossing, perpendicular to the retinal veins. (C, E, H, J) Optical coherence tomography images with indications of vascular outlines from the upper OCT images. (A–E) An AV crossing that shows concealment. The retinal vein shows focal narrowing of the lumen at the crossing. However, the vein does not exhibit signs of compression or flattening. The venous lumen is round, even just under the artery. (F–J) An AV crossing that shows severe tapering. On fundus photographs, the bloodstream seems to be extremely narrow in the area of the crossing site. Optical coherence tomography sections reveal that the actual venous lumen is larger than portrayed by fundus photograph and maintains a round shape. Red lines indicate arterial outlines, and blue lines indicate venous outlines.

Vascular Parameters in Subjects With Versus Without Crossing Phenomena

Of 150 crossing sites, 47 were judged to lack crossing phenomena, and 103 were considered to involve crossing phenomena, as determined from the fundus photographs. Table 2 presents a comparison of both groups. Mean outer and inner arterial diameters at the crossings were significantly larger in the positive group than the negative group (P < 0.001). Although presumed inner venous diameters were similar in both groups (P = 0.582), the actual inner venous diameters at the crossings were significantly smaller in the group with crossing phenomena (86.3 ± 17.8 μm versus 104.8 ± 14.0 μm, P < 0.001). Accordingly, venous narrowing rate was larger in those with crossing phenomena (25.6% ± 12.1% vs. 10.9% ± 8.1%, P < 0.001). The venous narrowing rate correlated with arterial diameter (Fig. 5A, r = 0.419, P < 0.001). Arterial wall thickness was also larger in eyes with crossing phenomena (P < 0.001). Systolic blood pressure and the prevalence of systemic disease were similar in both groups (P > 0.05). Intraocular pressure was lower in the crossing phenomenon–positive groups (P = 0.047). Women showed crossing phenomena more commonly than men (P = 0.036). 

Figure 5

 

(A) Scattergram for inner arterial diameters and the degree of venous narrowing at crossings. The venous narrowing rate shows a significant positive correlation with inner arterial diameter (r = 0.419, P < 0.001). Presumed inner venous diameter (B), inner arterial diameter (C), and the degree of venous narrowing (D) measured at the arteriovenous crossings with or without crossing phenomena. *P < 0.01, #P < 0.05, compared with the crossings that did not show crossing phenomena. The red line shows the mean value for each parameter.

Table 2

 

Comparison of Vascular Parameters and General Factors at Crossings With Versus Without Crossing Phenomena

	Crossing Phenomena
	Negative	Positive	P Value
Subject no.	47	103
Age, y	68.9 ± 9.0	69.9 ± 8.6	n.s.
Distance from optic disc center, μm	3639 ± 985	3571 ± 1009	n.s.
Outer arterial diameter, μm	95.7 ± 18.0	119.9 ± 18.7	<0.001
Actual inner arterial diameter, μm	68.5 ± 18.0	87.9 ± 16.6	<0.001
Presumed inner arterial diameter, μm	68.5 ± 17.3	85.9 ± 16.6	<0.001
Arterial wall thickness, μm	13.6 ± 2.1	16.0 ± 2.7	<0.001
Actual inner venous diameter, μm	104.8 ± 14.0	86.3 ± 17.8	<0.001
Presumed inner venous diameter, μm	118.3 ± 16.3	116.5 ± 18.5	n.s.
Venous narrowing rate, %	10.9 ± 8.1	25.6 ± 12.1	<0.001
Intraocular pressure, mm Hg	13.8 ± 2.6	12.8 ± 2.9	0.047
Systolic blood pressure, mm Hg	135.6 ± 20.4	138.0 ± 19.3	n.s.
Sex, male/female	28/19	42/61	0.036
Hypertension, ±	20/27	50/53	n.s.
Diabetic mellitus, ±	6/41	9/94	n.s.
Dyslipidemia, ±	9/38	30/73	n.s.

 Unpaired t-test and the χ² test were performed for comparisons. n.s., not statistically significant.

Vascular Parameters Associated With Crossing Phenomena

Of 103 crossing sites with crossing phenomena, concealment, tapering, deflection, and humping were seen at 40, 46, 21, and 16 crossings, respectively. Six crossings exhibited concealment and deflection phenomena simultaneously, and 14 crossings exhibited tapering and deflection phenomena. Table 3 shows the vascular parameters associated with each type of crossing phenomenon. All types of crossing phenomena were associated with larger arterial diameter and more severe narrowing of the venous lumen. Arterial and venous diameters as well as the venous narrowing rate did not vary significantly among phenomena (P > 0.05, Fig. 5). 

Table 3

 

Comparison of Vascular Parameters at Crossings Showing Each Type of Crossing Phenomenon

	No Crossing Phenomena	Concealment	Tapering	Deflection	Humping
Subject no.	47	40	46	21	16
Age, y	68.9 ± 9.0	67.5 ± 7.4	71.5 ± 9.4	74.3 ± 7.4	71.3 ± 8.0
Distance from optic disc, μm	3639 ± 985	3731 ± 994	3588 ± 960	3881 ± 1103	3153 ± 1151
Outer arterial diameter, μm	95.7 ± 18.0	123.4 ± 18.3*	116.3 ± 18.6*	116.4 ± 19.3*	123.7 ± 16.2*
Inner arterial diameter, μm	68.5 ± 18.0	90.3 ± 15.2*	84.7 ± 17.5*	85.0 ± 17.5†	93.1 ± 14.9*
Arterial wall thickness, μm	13.6 ± 2.1	16.6 ± 2.8*	15.8 ± 2.5*	15.7 ± 2.6†	15.3 ± 2.3
Actual inner venous diameter, μm	104.8 ± 14.0	90.2 ± 18.4*	84.2 ± 13.4*	87.1 ± 14.8*	80.6 ± 23.9†
Presumed inner venous diameter, μm	118.3 ± 16.3	116.1 ± 18.9	117.9 ± 18.1	116.0 ± 21.8	112.4 ± 19.3
Venous narrowing rate, %	10.9 ± 8.1	22.0 ± 12.4*	28.1 ± 8.4*	24.4 ± 7.6*	28.3 ± 17.4†

 Six crossings exhibited concealment and deflection phenomena simultaneously and 14 crossings exhibited tapering and deflection phenomena. Unpaired t-test and Bonferroni correction were used for comparison.

*   P < 0.01, compared with crossings that lacked associated crossing phenomena.

†   P < 0.05, compared with crossings that lacked associated crossing phenomena.

Almost all crossings that showed tapering, deflection, or humping phenomena on fundus photographs showed focal venous narrowing greater than 14.0% in magnitude. However, at crossings associated with venous concealment, the focal venous narrowing rate varied widely, from 0.0% to 54.3%. 

On fundus photographs, deflection and humping phenomena reflect changes in venous running direction (deviations), while concealment and tapering represent changes in venous lumen appearance. We created a “deviation group,” which included crossings with deflection or humping phenomena. Among 150 crossing sites, 47 were judged to lack crossing phenomena (group N); crossing phenomena without deviations (group D−) were seen at 66 crossings; deviations were seen at 37 crossings (group D+). Table 4 shows the comparison of vascular or general factors among these three groups. Age (P = 0.019), prevalence of hypertension (P = 0.012), and vascular parameters associated with crossing phenomena (P > 0.001) differed significantly among these three groups. Mean subject age was greater in group D+ than in group D− (P = 0.005). Groups D− and D+ exhibited similar vascular parameters (P > 0.05). However, the prevalence of hypertension was significantly higher in group D+ than group D− (P = 0.004). 

Table 4

 

Comparison of Vascular Parameters and General Factors With or Without Venous Deviation at the Crossing

	No Crossing Phenomenon, Group N	No Venous Deviation, Group D−	Venous Deviation, Group D+	P Value
Subject no.	47	66	37
Age, y	68.9 ± 9.0*	68.1 ± 8.6†	73.0 ± 7.7	0.019
Distance from optic disc center, μm	3639 ± 985	3574 ± 919	3567 ± 1167	n.s.
Outer arterial diameter, μm	95.7 ± 18.0‡	120.1 ± 19.1	119.5 ± 18.3	<0.001
Inner arterial diameter, μm	68.5 ± 18.0‡	87.6 ± 16.6	88.5 ± 16.7	<0.001
Arterial wall thickness, μm	13.6 ± 2.1‡	16.3 ± 2.7	15.5 ± 2.4	<0.001
Actual inner venous diameter, μm	104.8 ± 14.0‡	87.4 ± 17.0	84.3 ± 19.2	<0.001
Presumed inner venous diameter, μm	118.3 ± 16.3	117.7 ± 17.3	114.4 ± 20.6	n.s.
Venous narrowing rate, %	10.9 ± 8.1‡	25.4 ± 11.8	26.1 ± 12.7	<0.001
Sex, male/female	28/19	29/37	13/24	n.s.
Hypertension, ±	20/27*	25/41†	25/12	0.012
Diabetes mellitus, ±	6/41	7/59	2/35	n.s.
Dyslipidemia, ±	9/38	16/50	14/23	n.s.

 One-way analysis of variance and the χ² test were used for comparisons among three groups. Pairwise difference analysis for posttests was performed by using the unpaired t-test and χ² test.

*   P < 0.05, compared with the venous-deviation positive group.

†   P < 0.01, compared with the venous-deviation positive group.

‡   P < 0.001, compared with the venous-deviation positive group.

Discussion

This is the first study that clarified the three-dimensional morphologic features of arteriovenous crossings in living human eyes. Retinal arteries and veins run primarily straight within the inner retina, but veins are likely to change direction abruptly and pass under or over arteries at crossings. Rigid arteries are generally thought to compress veins at crossings in the retina. ^11–13 However, sequential thin OCT sections showed no actual compression or flattening of the venous lumens. Rather, retinal veins frequently exhibited luminal narrowing at the crossings. Notably, the lumens retained a circular shape in cross-section. 

In the current study, the degree of venous narrowing at the crossing correlated positively with the diameter of the crossing artery. Although the reason for this association is unclear, venous remodeling might be involved. At the crossings with larger arteries, veins must change course markedly to pass the crossing arteries, which results in turbulent blood flow. Turbulent blood flow causes chronic damage to the venous endothelial cells, leading to endothelial cell proliferation and venous wall remodeling. Previous histologic studies would support our hypothesis. Yu et al. ¹⁴ have reported that venous endothelial cells, especially those in older eyes, became elongated at arteriovenous crossings owing to the proliferation of stress fibers. Jefferies et al. ¹⁵ have reported focal stratification of the venous basement membrane at the crossing, at the point diametrically opposed to the point of contact with the artery. 

As the venous lumen narrows at the crossings, venous flow velocity and pressure increase, which damages the venous endothelium further. Such endothelial dysfunction contributes to venous thrombosis, ^16,17 which is why most branch retinal vein occlusions (BRVOs) occur at arteriovenous crossings. ^15,18–23 Muraoka et al. ⁸ recently have shown that the venous lumen is not compressed at crossings, which caused BRVO, and that intravenous thrombi are seen downstream from the affected crossing. The authors conclude that vein compression is not the primary cause of BRVO. Our research suggests such occlusions are due primarily to focal narrowing of the venous lumen. Larger arteries represented a risk factor for venous narrowing at the crossing site. We therefore assumed that the risk of developing BRVO may be influenced by the distance of crossing sites from the optic disc or the number of major retinal vessels in the eye. 

Previous researchers have reported the various types of crossing phenomena. ^24,25 In the current study, the presence of crossing phenomena was significantly associated with reduced venous diameter and increased arterial diameter, arterial wall thickness, and venous narrowing. None of these parameters varied among the various types of crossing phenomena. The various types of crossing phenomena can be clearly differentiated on fundus photographs, but these differences are not associated with any identifiable microvascular abnormalities. 

The degree of venous narrowing varied widely (range, 0.0%–54.3%) at crossings associated with venous concealment, although the bloodstream seemed to be interrupted on fundus photographs. The vessels may have been obscured by adventitia sheath thickening. Seitz ⁶ has reported proliferation of the adventitia and glia, forming a single sheath around the crossing, in eyes with crossing phenomena. Vascular sheath hypertrophy obscures the associated vascular changes from being viewed on fundus photographs. 

On the basis of histologic findings, several hypotheses have been proposed to explain the nature of crossing phenomena: arterial compression of the vein, narrowing of the venous lumen, sclerosis of the venous wall, or proliferation of the adventitia and glia surrounding the artery and vein. ⁶ The current OCT examinations clearly showed no signs of arterial compression of the vein at crossings. Rather, our findings indicate that focal narrowing of the venous lumen is a common feature associated with crossing phenomena. Unfortunately, OCT does not allow us to evaluate the vascular sheath at crossings. Nonetheless, sclerosis of the venous wall or thickened adventitial sheaths may be involved in focal narrowing of the venous lumen at crossing sites. 

Crossing phenomena are some of the earliest vascular changes associated with arteriosclerosis or hypertension. ^3–5 Various large epidemiologic studies ^7,26,27 have shown that crossing phenomena are strongly associated with systemic hypertension. In this study, however, the prevalence of hypertension was similar in groups with and without crossing phenomena. This may reflect the small size of the study population. 

In the current study, however, we found the prevalence of hypertension to be higher in subjects with venous deviations (deflection and humping). In such eyes, the mean venous narrowing rate was similar to that in eyes with crossing phenomena that were not considered deviations. Venous deviation may reflect the microvascular changes associated with hypertension more sensitively than other crossing phenomena. Further study is needed to assess the relationship between venous narrowing at crossings and hypertension. 

This study has various limitations, including a relatively small sample size and a cross-sectional study design. In the current study, we recruited subjects from the patients who visited our hospital for eye examinations. This may not be a representative general population and could have introduced selection bias. In addition, OCT examination cannot be used to visualize perivascular tissue. In the current study, we studied three-dimensional features of a single, randomly selected crossing in each eye. Optical coherence tomography examination enables us to assess the venous changes at crossings with a view that is not obscured by retinal tissue. In contrast to fundus photography, OCT allows for the measurement of objective and quantitative parameters. We found that crossings typically exhibited venous lumen narrowing without compression or flattening. Increased venous narrowing and larger arteries were more likely to be seen at crossings that exhibited crossing phenomena. If the presence of larger arteries is an independent risk factor for crossing phenomena, the severity of crossing phenomena could represent a marker of arteriosclerosis, after adjustment for arterial diameter. Optical coherence tomography–based studies of retinal vessels are useful and provide quantitative parameters that enable us to identify the association of retinal vascular changes with ocular and systemic risk factors with greater precision and detail. Further studies with larger sample sizes would be necessary to fully explain the pathogenesis and risk factors associated with vascular changes at arteriovenous crossing sites. 

Supplementary Materials

pdf S3 

mov S1 

mov S2 

Acknowledgments

Supported, in part, by the Japan Society for the Promotion of Science, Tokyo, Japan (Grant-in-Aid for Scientific Research, No. 21592256). 

Disclosure: K. Kumagai, None; A. Tsujikawa, Pfizer (F); Y. Muraoka, None; Y. Akagi-Kurashige, None; T. Murakami, None; K. Miyamoto, None; R. Yamada, None; N. Yoshimura, Topcon Corporation (F), Nidek (F, C), Canon (F)