Prediction of Retinal Ischemia in Branch Retinal Vein Occlusion: Spectral-Domain Optical Coherence Tomography Study

Abstract

Purpose: To investigate the relationship between spectral-domain optical coherence tomography (SD-OCT) measurements and retinal nonperfusion in patients with branch retinal vein occlusion (BRVO).

Methods: Forty-one patients with BRVO who had recovered from macular edema and had been followed for ≥2 years were included via retrospective, medical record review. Patients were divided into two groups that included 20 nonischemic eyes and 21 ischemic eyes, and 41 fellow control eyes were also included. Using SD-OCT, we measured the thickness of the macular layer, ganglion cell–inner plexiform layer (GC-IPL), and retinal nerve fiber layer (RNFL) in both the BRVO-affected and fellow eyes. The eyes were subdivided into affected and nonaffected areas of BRVO. Each area of normal fellow eyes and BRVO eyes was compared between the two groups. Receiver operating characteristic (ROC) curves were used to evaluate the diagnostic ability of the OCT measurements for ischemic BRVO.

Results: The macula, the GC-IPL, and RNFL in the BRVO-affected area were significantly thinner compared to those of the fellow eyes in the two groups. The thickness of macula, GC-IPL, and RNFL in the ischemic BRVO group was also significantly less than in the nonischemic BRVO group. In the ROC curve analysis of a total of 82 eyes, the area under the ROC curve of the RNFL on the ischemic BRVO was the highest at 0.906, followed by the GC-IPL (0.824) and the macula (0.768).

Conclusions: The thickness of macula, GC-IPL, and RNFL in the ischemic BRVO group was significantly reduced compared to the nonischemic BRVO group, especially in the RNFL.

Retinal vein occlusion (RVO) is the second most common cause of visual loss following diabetic retinopathy.^1–3 There are three types of RVO, which differ in the location of the occluded vessel: central retinal vein occlusion (CRVO); branch retinal vein occlusion (BRVO); and hemiretinal vein occlusion (hemi-RVO), an anatomical variation of CRVO.⁴ 

Branch retinal vein occlusion usually occurs at the sites of arteriovenous crossing.^5–7 Several risk factors for BRVO have been discussed in previous studies, such as arterial compression of the vein at the arteriovenous crossing sites,^8,9 turbulent blood flow,^10,11 and venous thrombus.^9,10 Two major complications of BRVO include macular edema and retinal nonperfusion.^12–14 These complications are closely associated with increased production of vascular endothelial growth factor (VEGF). Anti-VEGF therapy has become the mainstay of treatment for BRVO worldwide as it is effective for management of BRVO-related complications.^15–17 

Retinal ischemia can occur due to various diseases such as diabetic retinopathy, retinal artery occlusion, retinal vein occlusion, and hypertensive retinopathy. Retinal angiography is useful for identifying retinal ischemia, but intravenous administration of fluorescein can result in adverse reactions, including deterioration of renal function and hypersensitivity,¹⁸ and requires the additional equipment necessary for angiography. 

Spectral-domain optical coherence tomography (SD-OCT), which is the latest generation of a noninvasive and high-speed imaging technique, has become popular and can reveal the 10 layers of the retina. Spectral-domain OCT imaging is an effective tool to determine the thickness of the macula, the ganglion cell–inner plexiform layer (GC-IPL), and the retinal nerve fiber layer (RNFL). Based on the results of SD-OCT imaging, some studies have reported thinning of the macular layers, the GC-IPL, and the RNFL in patients suffering from ischemic retina and optical nerve.^19–26 

In the present study, we sought to evaluate differences in SD-OCT values between BRVO patients with and without retinal ischemia who recovered from macular edema after the onset of BRVO, and to investigate whether these values may be used as indicators for discrimination of ischemic from nonischemic BRVO. 

Methods

Subjects

The study sample was 41 eyes of 41 patients. The patients, who were diagnosed with a monocular temporal BRVO between January 2011 and January 2015 in Chungnam National University Hospital, were included in this study via retrospective medical record review. The study protocol was approved by the Institutional Review Board of Chungnam National University Hospital (Daejeon, Republic of Korea). The study adhered to the tenets of the Declaration of Helsinki. At initial diagnosis, all patients had retinal edema and were treated with intravitreal bevacizumab (Avastin; Genentech, Inc., San Francisco, CA, USA) injection (IVB). The injection was performed monthly and as needed until the macular edema resolved. After recovering from macular edema, patients visited the clinic at intervals of 1 month. After 3 months without macular edema, patients were followed up every 3 months. 

For this study, a group of 41 patients who recovered from retinal edema were followed from the time of onset for a period of at least 2 years. Fluorescein angiography (FA) was performed using SPECTRALIS Heidelberg Retina Angiograph (Heidelberg Engineering, Heidelberg, Germany) on all BRVO patients at the time of onset and at 3 months. Based on FA findings of BRVO at 3 months, the patients were divided into two groups: the ischemic BRVO group (retinal ischemia showing more than five disc diameters) and the nonischemic BRVO group (retinal ischemia showing less than five disc diameters), as shown in Figures 1 and 2. 

Figure 1

 

Fundus photography and fluorescein retinal angiography. Fundus photography initially (A) and 32 months later (B), and early (C) and late (D) fluorescein retinal angiography of a nonischemic BRVO patient.

Figure 2

 

Fundus photography and fluorescein retinal angiography. Fundus photography initially (A) and 32 months later (B), and early (C) and late (D) fluorescein retinal angiography of an ischemic BRVO patient.

We analyzed the overall findings of the patients, including a review of their medical history, refraction test, best-corrected visual acuity (BCVA), intraocular pressure (IOP), slit-lamp biomicroscopy, fundus examination/photography, OCT, and FA. We investigated the frequency of intravitreal injections and the incidence of vitreous hemorrhage and retinal neovascularization over the 2-year follow-up period. We excluded known corneal diseases, glaucoma, neuro-ophthalmologic disorders, previous intraocular surgery (i.e., cataract surgery, vitrectomy) or laser photocoagulation, uveitis, and other retinal diseases. 

When complications such as retinal neovascularization or vitreous hemorrhage were observed during the follow-up, patients received laser treatment, intravitreal anti-VEGF injection, or vitrectomy. Data were collected prior to the commencement of the procedure or surgery for analysis. We used the contralateral eye in the BRVO patients as a control for the BRVO-affected eyes. Contralateral eyes had a normal anterior segment and a retina with a BCVA of 20/20 vision (Snellen) or better, a normal range of IOP, and a spherical equivalent of ±1 diopter. 

Optical Coherence Tomography

Spectral-domain OCT imaging was performed using a Cirrus HD-OCT (Carl Zeiss Meditec, Inc., Dublin, CA, USA): macular cube 512 × 128 combination scan mode and an optic disc cube 200 × 200 scan mode. 

The macular thickness measurements were obtained by the macular cube 512 × 128 combination scan mode. The 6 × 6-mm circle corresponds to the Early Treatment Diabetic Retinopathy Study (ETDRS) subfield, which is segmented by a concentric circle into central, inner, and outer circles (1, 3, and 6 mm). The inner and outer circles are subdivided into four-quadrant sectors (superior, inferior, nasal, and temporal), consisting of nine areas in total. The thickness of each area is measured. 

In the macular cube scan, we measured the GC-IPL thickness using the ganglion cell analysis algorithm. The algorithm can provide the average of the six sectors (supero-temporal, superior, superonasal, inferonasal, inferior, inferotemporal) of the elliptical annulus (dimensions: vertical inner and outer radius of 0.5 and 2.0 mm, horizontal inner and outer radius of 0.6 and 2.4 mm) centered on the macular area. 

The optic disc cube 200 × 200 scan mode was used to image the optic disc and the RNFL over a 6 × 6-mm optic nerve head using 200 × 200 axial measurements. The RNFL thickness of the four-quadrant sectors (superior, inferior, nasal, and temporal) was then measured. 

The two scans were performed on both eyes of all patients by an experienced examiner. We selected patients with good-quality scans showing signal strength (SS) ≥ 7 while excluding patients with a SS < 7. The latest measurement data of the patients who were followed up for at least 24 months from the initial diagnosis were used for analysis. Based on the BRVO area, we further divided the OCT measurements maps into areas affected and nonaffected by BRVO. We defined the affected area in the superotemporal BRVO as inner and outer superior areas of the ETDRS subfield in macular analysis, as the superior segment of GC-IPL map in GC-IPL analysis, and as the superotemporal area (from 12–3 o'clock) of the 12-hour map in the RNFL analysis. The side opposite the affected area was defined as the nonaffected area. We also defined the affected area in the inferotemporal BRVO as the inner and outer inferior area of ETDRS subfield in the macular analysis, as the inferior segment of the GC-IPL map in the GC-IPL analysis, and as the inferotemporal area (from 3–6 o'clock) of the 12-hour map in the RNFL analysis. Mean measurement values defined as affected and nonaffected areas were used for analysis (Fig. 3). 

Figure 3

 

Optical coherence tomography measurements of macular (A), GC-IPL (B), and RNFL (C) thicknesses of a BRVO (superotemporal area) patient (left eye). In the macular analysis (A), the affected area (blue square) was defined as the inner and outer superior area of the ETDRS subfield, and the nonaffected area as the opposite area in the ETDRS subfield (red square). In GC-IPL analysis (B), the affected area (blue square) was defined as the superior segment in the GC-IPL measurement map, and the nonaffected area as the opposite area in the GC-IPL map (red square). In the RNFL analysis (C), the affected area was defined as the superotemporal area (from 12–3 o'clock, blue square) of the 12-hour thickness map, and the nonaffected area as the opposite area in the 12-hour thickness map (from 6–9 o'clock, red square).

Statistical Analysis

For the comparison of diverse clinical factors between the nonischemic and ischemic groups, the Mann-Whitney U and χ² tests were used. Optical coherence tomography measurement values of both normal and BRVO eyes were analyzed using the Mann-Whitney U test, and OCT measurement values were compared between the two groups. All statistical analysis was performed using the SPSS analytical software version 18.0 (SPSS, Inc., Chicago, IL, USA). In all analyses, differences were considered significant at a level of P < 0.05. 

Receiver operating characteristic (ROC) curves and the area under receiver operating characteristic curves (AUC) on the OCT-based mean measurement values of the macula, GC-IPL, and RNFL thickness in the lesion were calculated to evaluate the diagnostic ability of OCT measurement values to discriminate ischemic from nonischemic BRVO in the total group (total group contained the nonischemic group, the ischemic group, and the contralateral normal eyes). The cutoff value was determined as the highest Youden index (sensitivity + specificity-1). Statistical analyses for the ROC curve and the AUCs were performed using Medcalc version 12.7 (MedCalc Software bvba, Ostend, Belgium). 

Results

Patient Characteristics

The study included a total of 82 eyes, 41 BRVO eyes and 41 fellow control eyes of 41 patients. The patients were divided into two groups: 20 nonischemic and 21 ischemic eyes. The mean age of the ischemic group was 65.91 ± 7.34 years, which was higher than that of the nonischemic group (63.92 ± 10.05 years), but there was no significant difference in age between the groups (P = 0.109). Hypertension was more prevalent in the ischemic group (16 patients, 76.19%) compared to the nonischemic group (10 patients, 50%) (P = 0.087). Final OCT examination time was 28.51 ± 4.34 months in the nonischemic group and 27.41 ± 2.23 months in the ischemic group. During the follow-up period, the mean frequency of IVB injections in the nonischemic and ischemic groups was 2.71 ± 0.85 and 3.38 ± 1.09 times, respectively. The onset of BRVO in the nonischemic group occurred in the superotemporal (17 eyes) and inferotemporal (3 eyes) areas. In the ischemic group, BRVO occurred in the superotemporal (18 eyes) and inferotemporal (3 eyes) areas. The nonperfusion area at the 3-month FA findings had a disc diameter of 2.18 ± 0.97 in the nonischemic group and 10.78 ± 3.38 in the ischemic group. During the follow-up period, there was no occurrence of retinal neovascularization or vitreous hemorrhage in the nonischemic group, but these complications were observed in the ischemic group: retinal neovascularization in 10 eyes and vitreous hemorrhage in 5 eyes (23.81%) (Table 1). 

Table 1

 

Baseline Characteristics of BRVO Patients

 

OCT Measurements in the Ischemic and Nonischemic Groups

The mean thickness values of the macula, the GC-IPL, and the RNFL layers in the BRVO-affected area of the nonischemic group were 296.81 ± 14.94, 79.03 ± 7.84, and 88.83 ± 8.96 μm, respectively. These layers were thinner than those of the contralateral normal eyes (308.25 ± 10.13, 86.26 ± 5.95, and 95.53 ± 8.87 μm). The differences for each retinal layer between the normal and BRVO eyes were statistically significant (macula: P = 0.016, GC-IPL: P = 0.005, and RNFL: P = 0.037). There were no significant differences in the mean measurement values in the nonaffected areas between the normal and BRVO eyes of the nonischemic group (Table 2). 

Table 2

 

Mean Macular, GC-IPL, and RNFL Thickness According to BRVO-Affected Area in the Nonischemic Group

 

The mean thickness values of the macula, the GC-IPL, and the RNFL layers in the BRVO-affected area of the ischemic group were 267.47 ± 35.16, 62.63 ± 16.64, and 68.81 ± 14.17 μm, respectively. These layers were thinner than those of the contralateral normal eyes (294.72 ± 16.91, 80.26 ± 6.42, and 87.98 ± 11.97 μm, respectively). The difference for each of the retinal layers between the normal and the BRVO eyes was statistically significant (macula: P = 0.030, GC-IPL: P < 0.001, and RNFL: P < 0.001). There were no significant differences in the mean measurement values of the nonaffected area between the normal and BRVO eyes of the ischemic group (Table 3). 

Table 3

 

Mean Macular, GC-IPL, and RNFL Thickness According to BRVO-Affected Area in the Ischemic Group

 

Comparison of the OCT measurement values in each lesion of the retina showed significant differences between the nonischemic and ischemic patients, that is, macula thickness (296.81 ± 14.94 vs. 267.47 ± 35.16), GC-IPL thickness (79.03 ± 7.84 vs. 62.63 ± 16.64), and RNFL thickness (88.83 ± 8.96 vs. 68.81 ± 14.17) (macula: P = 0.002, GC-IPL: P < 0.001, RNFL: P < 0.001). In contrast, there were no significant differences between these thickness parameters in the nonlesion areas (Table 4). 

Table 4

 

Comparison of Mean Macular, GC-IPL, and RNFL Thickness Between Nonischemic BRVO and Ischemic BRVO According to BRVO-Affected Area

 

Diagnostic Performance for Ischemic BRVO

When the diagnostic ability of the OCT measurement values for ischemic BRVO in the total group (including the nonischemic, ischemic, and contralateral normal eyes, n = 82 eyes) was assessed using ROC curves and AUCs, the AUC of RNFL thickness was 0.906, which was higher than that for macular thickness (0.768) and GC-IPL thickness (0.824). If the reference level of RNFL thickness in the lesion was defined as 78.00 μm, the sensitivity and specificity for ischemic BRVO detection were 85.71% and 92.59%, respectively (Fig. 4). 

Figure 4

 

Prediction of ischemic BRVO based on the receiver operating characteristic (ROC) curve of affected macular, GC-IPL, and RNFL area.

Discussion

A number of ischemia-related retinal diseases, such as diabetic retinopathy,^19,27–29 RVO,^25,26,30 retinal artery occlusion,^31–33 and severe hypertensive retinopathy,^34,35 are responsible for thinning of the retinal layers. Several studies have indicated that retinal ischemia associated with diabetic retinopathy and RVO induces thinning of the inner retina.^26,30,36,37 Ebneter et al.³⁸ reported that the thickness of the layers of the outer retina in the ischemic retinal area is well preserved, whereas that of the layers of the inner retina is reduced, in animal models of RVO. Two-thirds of the inner retina receives its blood supply from the retinal blood vessels and one-third of the outer retina from the retina and choroid. The inner retinal layers may be particularly at risk of hypoxic damage because they are supplied with oxygen by the retinal vasculature, which is relatively sparse compared with the choroidal circulation that supplies most of the outer retina. This was demonstrated by Yu et al.³⁹ in animal experiments. 

Based on the results of SD-OCT imaging in the patients, we identified thinning of the BRVO-affected area in the macula, GC-IPL, and RNFL, and the thickness of the GC-IPL and the RNFL in the inner retina was significantly reduced compared with the full thickness of the retina. The results of this study are consistent with the findings of numerous previous studies. The OCT-based thickness profiles in the two groups revealed that the thinning of the RNFL layer in the ischemic BRVO group was greater than that in the nonischemic BRVO group. 

Regarding the diagnostic capabilities of the OCT measurement values to discriminate ischemic from nonischemic BRVO using ROC curves, we found that the AUC of RNFL thickness was 0.906; moreover, an improved diagnostic performance for the detection of ischemic BRVO was achieved in terms of sensitivity (85.71%) and specificity (92.59%) when the mean RNFL thickness value in the lesion was less than 78.00 μm. Retinal edema and hemorrhage in the lesion and macula occur following the onset of BRVO, and the RNFL thickness will decrease considerably over time due to the presence of retinal ischemia. The reason for the improved diagnostic ability of the RNFL thickness in this study could be that the peripapillary RNFL reflected the ischemia-related thinning of the entire retina compared with the macula or the GC-IPL thickness, which reflected only that in the macular area. 

In this study, all patients underwent IVB treatment to resolve macular edema, but its effect on RNFL thickness remains controversial.^40–43 The mean frequencies of IVB injections in the nonischemic and ischemic groups were 2.71 ± 0.85 and 3.38 ± 1.09 times, respectively; this difference was not significant. Furthermore, since the frequency of IVB injections was small in the present study, the RNFL defect likely occurred due to ischemia rather than the intravitreal injection. 

Frucht et al.⁴⁴ and Hayreh et al.⁴⁵ reported that the IOP may increase in RVO patients. However, in their study, all patients presented an IOP in the normal range with no known glaucoma, and there were no significant differences between the two groups in terms of IOP and the average cup-to disc (C/D) ratio. Therefore, the results of this study suggest that elevation of IOP is not correlated with a RNFL defect. 

Since the inception of retinal angiography, FA has been widely used as one of the most crucial and accurate tests for retinal nonperfusion.⁴⁶ Nevertheless, it has been reported that intravenous administration of sodium fluorescein can result in various side effects, ranging from mild allergic reactions such as nausea and skin rash to life-threatening anaphylactic shock.¹⁸ In place of FA, new noninvasive optical imaging modalities such as phase-variance OCT^47,48 and swept-source OCT^30,49 have been investigated. In addition to newly developed phase-variance OCT or swept-source OCT, RNFL imaging with the currently popular SD-OCT can provide important diagnostic and prognostic information in the management of BRVO. 

Our study had several limitations. First, each test method should cover different retinal areas. The RNFL thickness was measured over the rim of the optic nerve, whereas measurement of the macula and the GC-IPL thickness was performed in the macular area. Thus, we could not directly compare the thickness of RNFL with that of GC-IPL in the ischemia-susceptible inner retina. To overcome these shortcomings, we compared the three thickness measurement values for the macula, the GC-IPL, and the RNFL layers using ROC curves, and found that the mean thickness values of the RNFL were significantly reduced compared to those of the macula and GC-IPL centered on the macular area only. Second, this study included a small sample size of 41 eyes, so we failed to include diverse BRVO eyes in the study. Last, this was a retrospective study. Thus, our insufficient data analyses should be clarified by further large-scale, prospective studies. 

In summary, the thickness of the macula, the GC-IPL, and the RNFL layers was reduced in the affected area in patients with nonischemic or ischemic BRVO during the 2-year follow-up period. In particular, the thickness of the RNFL in ischemic BRVO was significantly reduced when compared to nonischemic BRVO due to retinal thinning. These findings suggest that if the overall retinal layers in the BRVO lesion are thinned and RNFL thickness is less than 78.00 μm during the follow-up of BRVO patients, the presence of ischemia BRVO should be suspected and appropriate action taken to prevent the onset of retinal neovascularization and its possible complications. 

Acknowledgments

Disclosure: H.-B. Lim, None; M.-S. Kim, None; Y.-J. Jo, None; J.-Y. Kim, None