Introduction

Eyes with branch retinal vein occlusion (BRVO) are clinically divided into two forms, perfused (non-ischemic) and non-perfused (ischemic) BRVO. The latter is defined as eyes having a retinal capillary non-perfusion area of 5 disk diameters or larger in a fluorescein angiogram (FA) [1]. Eyes with non-perfused BRVO tend to develop neovascularization from the residual retinal vasculature into the vitreous cavity leading to vitreous hemorrhage, one of the most severe late complications of BRVO [1].

A recent study using spectral-domain optical coherence tomography (SD-OCT) reveals that the thickness of the retinal nerve fiber layer (RNFL) and the ganglion cell inner plexiform layer (GC-IPL) in eyes with non-perfused BRVO recovering from macular edema, were reduced in the affected ischemic retinal area [2]. Thinning of these inner retinal layers is usually found in eyes with glaucoma, whose principal pathology is loss of retinal ganglion cells without ischemia of the retinal circulation. The principal finding of glaucoma is a scotomatous visual field defect, which represents a regional reduction in light sensitivity due to disruption of the intra-retinal transduction of the visual signal from photoreceptors to the optic nerve via the retinal ganglion cells.

Thus, a similar visual dysfunction with reduction of light sensitivity may also be observed in eyes with non-perfused BRVO exhibiting thinning of the inner retina, including the retinal ganglion cell layers. However, to date, there have been few reports of perimetric examinations in eyes with BRVO.

To confirm the relationship between the reduction of light sensitivity demonstrated in perimetry and the retinal areas of impaired perfusion in eyes with BRVO, I investigated FAs and the results of Humphrey perimetry program 30-2 in eyes with chronic BRVO recovering from retinal edema and hemorrhage.

Methods

This retrospective study of observational case series was conducted in an institutional setting. The study design was approved by the Institutional Review Board of the University of Yamanashi Hospital. The research methods and analyses described herein adhered to the tenets of the Declaration of Helsinki.

I studied the medical records of 43 eyes from 42 patients with BRVO whose perfusion status was assessed by FA images obtained at least 6 months after disease onset. I excluded eyes with glaucoma, diabetic retinopathy, and other ocular diseases that could cause visual field abnormality. All patients had undergone an ophthalmic examination, including measurement of best-corrected visual acuity, intraocular pressure measurement, indirect ophthalmoscopy, fundus photography, and SD-OCT (Cirrus HD-OCT system, Carl Zeiss Meditec, Dublin, CA, USA) at the initial examination and at the follow-up visits. The FA images in the current study were taken using a fundus camera (TRC-50DX Mydriatic Retinal Camera, Topcon medical systems, Tokyo, Japan; or HRA-2 Spectralis, Heidelberg Engineering, Dossenheim, Germany) at the follow-up visit at least 6 months after disease onset, when retinal hemorrhage and edema had been resolved. Disease onset was assessed by the patient’s self-report of visual symptoms. Most eyes received one or more intravitreal injections of anti-vascular endothelial growth factor (VEGF) agents before FA testing. Macular edema was spontaneously resolved without any treatment in other eyes. No eyes received laser photocoagulation prior to FA testing. I analyzed the results from the 30-2 SITA-Fast program of a Humphrey Visual Field Analyzer II (Carl Zeiss Meditec) with good reliability of fixation loss scores ≤ 20% and false positive or false negative errors ≤ 33% obtained on the same day as the FA or within 6 months afterwards.

Correlation between light sensitivity and perfusion status

I studied the FA images within the retinal squares between 12° nasal and 18° temporal, and between 0° and 18° in either vertical direction on the BRVO affected side. This area was divided into 15 retinal squares by 4 vertical and 2 horizontal lines as shown in the FA image in Fig. 1.

Fig. 1
figure 1

Humphrey 30-2 results and the corresponding fluorescein angiogram. The light sensitivity at the 15 test points in the total deviation display (lower left) are superimposed onto the corresponding squares in the FA image (right) of the eye with BRVO of the superotemporal branch. The framework of yellow grid lines was magnified, as the center of the shaded square corresponding to the blind spot is located on the vertical line passing through the center of the optic disc. Upper left is the grey scale display of the Humphrey 30-2. BRVO branch retinal vein occlusion

Full size image

The framework of grid lines was magnified to locate the center of the shaded square in Fig. 1 at 15° nasal, at the center of the optic nerve head. The corresponding light sensitivity of each of the 15 unit squares was obtained from the total deviation display at the 15 intersection points of the 5 vertical arrays (3° and 9° temporal and 3°, 9°, and 15° nasal) and the 3 horizontal rows (3°, 9°, and 15° superior for eyes with inferotemporal BRVO or 3°, 9°, and 15° inferior for eyes with superotemporal BRVO). Each of the retinal squares in the FA and the corresponding light sensitivity from the total deviation display were located in mirror images across the horizontal midline. Each value of light sensitivity from the total deviation display of the Humphrey 30-2 was superimposed onto the grid of 15 squares in the FA image (Fig. 1).

Grading of perfusion status in each retinal square

The perfusion status of each of the 15 retinal squares in the FA was graded as perfused, partially perfused or non-perfused according to the capillary-free area within the square. Perfused samples were defined when the capillary-free area was < 1/3 of each square, and non-perfused when > 2/3 of each square. Partially perfused was defined when the capillary-free area was between 1/3 and 2/3 for each square (Fig. 2).

Fig. 2
figure 2

Grading of perfusion status in each retinal square. Fifteen retinal squares corresponding to the test points in the total deviation display of the Humphrey 30-2 between 9° temporal and 15° nasal and between 3° and 15° vertical were graded according to the area of capillary-free retina. The non-perfused square contains a capillary-free retinal area larger than 2/3 of the square, the partially perfused square shows a capillary-free area between 1/3 and 2/3 of the square, and the perfused square shows a capillary-free area smaller than 1/3 of the square

Full size image

Three ophthalmologists evaluated independently the perfusion status of a total of 645 retinal squares consisting of 15 retinal squares per eye in 43 subject eyes. I investigated the correlations between perfusion status and the corresponding light sensitivity.

Statistical analysis

The statistical analyses were conducted using StatFlex (Version 6.0 Artec Inc. Osaka, Japan). A Kruskal–Wallis test was used to compare light sensitivity among the three groups of retinal squares from the different perfusion states. Multiple comparisons of similar data were evaluated using a Dunn test. P-values of less than 0.05 were considered statistically significant.

Results

The characteristics of the subjects are described in Table 1. Approximately half of the eyes were diagnosed as non-perfused BRVO based on the criterion of retinal capillary non-perfusion of 5 disk diameters or more within the retinal area in the available FA images.

Table 1 Patient demographics
Full size table

Complete agreement was obtained among the three evaluators for the perfusion status in 459 of the 645 retinal areas (71.2%). Light sensitivity corresponding to these 459 areas is plotted in Fig. 3. The light sensitivity was significantly different among the three perfusion groups (P < .000001, Kruskal–Wallis test). The median (first and the third quartiles) values were − 4.0 dB (− 7.0, − 2.0), − 8.0 dB (− 13.0, − 5.0) and − 16.0 dB (− 26.0, − 9.0) for perfused, partially perfused and non-perfused retinal areas, respectively, which were significantly different from each other (P < .01 each, Dunn test) (Fig. 3).

Fig. 3
figure 3

Distribution and box plot of light sensitivity for the three perfusion statuses of retinal squares. The horizontal bars indicate the median and the boxes show the lower and upper quartiles

Full size image

Discussion

A previous report [2] demonstrating inner retinal thinning in eyes with non-perfused BRVO similar to glaucoma suggests that static perimetry could reveal regional reductions in light sensitivity corresponding to retinal capillary non-perfusion in eyes with BRVO, demonstrating relative or absolute scotoma similar to that observed in eyes with glaucoma.

Therefore, I studied the correlation between light sensitivity and perfusion status in 15 retinal squares in eyes with BRVO using the total deviation display of a Humphrey perimeter. The results revealed reduction in light sensitivity depending on the perfusion status in the order of non-perfused, partially perfused, and perfused retinal squares.

Previous work using a fundus-related microperimeter (MP-1, Nidek), shows light sensitivity reduced in areas of retinal capillary non-perfusion in eyes with diabetic retinopathy [3]. Another investigation from the same institute reports reduced light sensitivity in eyes with BRVO, related to retinal thickening due to retinal edema [4]. The implication of retinal capillary non-perfusion upon the reduction of light sensitivity in eyes with BRVO was not investigated.

I have also previously reported that results of investigations using a Humphrey perimeter show that in eyes with BRVO without apparent retinal capillary non-perfusion, light sensitivity is decreased by retinal thickening due to retinal edema [5]. However, the extent of the reduction in light sensitivity was relatively mild, within 10 dB. This was far smaller than the reduction due to impaired perfusion in eyes with BRVO, which ranged up to 30 dB, as shown in Fig. 3. Moreover, the reduction in light sensitivity due to retinal edema may be reversible [4].

When observed in eyes with BRVO, hemorrhage, edema, and capillary non-perfusion can all reduce regional light sensitivity by blocking light, by impaired neural transduction, or by neuronal loss in the inner retina. All eyes in the current study had recovered from macular edema and hemorrhage either through intravitreal injection of anti-VEGF agents or spontaneously. Therefore, the present results were derived from eyes with BRVO with minimal retinal hemorrhage and edema.

The results indicate that impaired retinal perfusion correlates closely with the regional reduction of light sensitivity in eyes with BRVO recovering from retinal hemorrhages and edema with a small number of exceptional data (Fig. 3). There are only a few points with lowest light sensitivity around − 30 dB in the perfused group in Fig. 3, which might be due to fixation instability during Humphrey testing. Compared with these perfused retinal areas showing lower light sensitivity, a larger number of points in the non-perfused group in Fig. 3 shows normal or near normal light sensitivity around 0 dB. The probable mechanism of reduced light sensitivity in the retinal area of non-perfusion is loss of the inner retinal neurons leading to impaired transduction of light signals from photoreceptors to the optic nerve. Several points in the non-perfused retina exhibited relatively higher light sensitivity which might be explained by the gradual loss of neurons in the inner retina in the specified area.

Although this investigation did not examine light sensitivity longitudinally, the reduction of light sensitivity may be irreversible because thinning of the retina due to reduced perfusion may never recover, as demonstrated in eyes with branch retinal artery occlusion [6]. Although recent literature reports increased mean light sensitivity after vitrectomy for macular edema in eyes with ischemic BRVO using a fundus-related microperimeter [7], this increase in light sensitivity may have resulted from reduced macular edema.

Many clinical studies on BRVO employed visual acuity measurements as the main outcome measure of visual function [1, 8, 9]. However, light sensitivity has rarely been assessed using static perimetry in these studies. Visual acuity and light sensitivity assessed by perimetry have different implications in terms of quality of vision. Static perimetry can reveal scotoma, i.e., a regional reduction of light sensitivity in any area in the posterior retina, while visual acuity loss is a spatial threshold defect of visual discrimination only at the fovea. The visual dysfunction in eyes with retinal diseases with extrafoveal pathology cannot be fully assessed with visual acuity alone. For example, a patient with non-perfused BRVO in the superotemporal quadrant of the retina may complain of a severe scotoma in the lower visual field despite normal visual acuity. His or her complaints may consequently be underappreciated unless perimetry is performed.

For this reason, I routinely employ Humphrey perimetry in eyes with BRVO to better understand the quality of vision in these patients. Figure 4 presents representative cases of perfused and non-perfused BRVO, where visual acuity was 1.0. While light sensitivity within 30° in eyes with perfused BRVO (Fig. 4, top) is almost normal, light sensitivity in the inferior visual field in eyes with superotemporal non-perfused BRVO (Fig. 4, bottom) is greatly reduced, demonstrating deep or absolute scotoma, which could induce marked visual disability despite good visual acuity.

Fig. 4
figure 4

Fluorescein angiogram (FA) and grey scale of the Humphrey central 30-2 of eyes with perfused and non-perfused BRVO. (Top left) FA OD of 73-year-old woman with perfused BRVO shows no retinal capillary non-perfusion in the affected retinal area along the dilated superotemporal retinal vein. (Top right) Grey scale of the Humphrey central 30-2 of the same eye demonstrates no apparent scotoma. Her visual acuity was 1.0. (Bottom left) FA OD of 50-year-old man with non-perfused BRVO demonstrates a large retinal capillary non-perfusion area along the superotemporal branch of the retinal vein. (Bottom right) Grey scale of the Humphrey 30-2 of the same eye demonstrates an absolute scotoma in the inferior field, which corresponds to the retinal capillary non-perfusion area in the FA. His visual acuity was 1.2. The yellow dotted circular line in each FA image indicates the retinal area corresponding to the Humphrey results

Full size image

Although the classification of perfused and non-perfused BRVO is prevalent in clinical practice, their differentiation is not always easy. In eyes with acute BRVO, within a few months of symptom onset, FA images show many areas where fluorescence is blocked by retinal hemorrhages, which may mimic hypofluorescence due to non-perfusion of the retina. To avoid misreading the blocked hypofluorescence due to retinal hemorrhages as non-perfusion hypofluorescence, I studied FA images taken during the chronic stage of the disease when most retinal hemorrhages had disappeared, results of these studies being very important to treat microaneurysms and capillary non-perfusion using laser photocoagulation in eyes with BRVO [10].

So-called non-perfused BRVO does not always refer to eyes with a complete dropout of the capillary network. Even with FA images of good quality without retinal hemorrhages, determination of non-perfused BRVO may still be ambiguous in some cases. The affected retinal area in such eyes includes both areas with the complete absence of capillaries and of reduced capillary density to various degrees. Therefore, in the current study, I divided the affected retinal area into 15 squares corresponding to the Humphrey test points. It was difficult to differentiate between non-perfused and perfused even for the small retinal squares shown in Fig. 3 because of the widely variable size of the capillary-free area within each retinal square. Thus, I defined partially perfused squares as having a capillary-free area between 1/3 and 2/3 of the total area of each square and others as either perfused or non-perfused. Owing to this subjective definition, I only adopted data that were agreed upon completely by the three evaluators. Consequently, the number of partially perfused squares available for the analysis was small. This is one potential limitation of the study.

Another limitation was the method of superimposing Humphrey results onto FA images, which may be inferior to using a fundus-related microperimeter, i.e., MP-1, in terms of accuracy of the correlation between the affected retinal area and light sensitivity. MP-1 can be advantageous for accurately correlating retinal location and function. However, one disadvantage is that it is less common in clinical practice compared with the Humphrey perimeter. Moreover, the dynamic range of its 20 dB measurement is far less than the 50 dB of the Humphrey perimetry. Thus, zero dB sensitivity measured with MP-1 does not always mean absolute scotoma, which is a major drawback of MP-1 in assessing light sensitivity in eyes with non-perfused BRVO.

In conclusion, light sensitivity is decreased in retinal areas showing impaired perfusion in eyes with BRVO present without retinal edema or hemorrhage.