Introduction

Branch retinal vein occlusion (BRVO) is a common retinal disease [1,2,3] that causes sudden impairment of visual function. In several clinical studies, anti-vascular endothelial growth factor (VEGF) therapy has been used to rapidly reduce macular edema [4,5,6,7,8,9]. This condition can be naturally resolved over a period of 2 years in approximately 80% of patients [10], although the visual outcomes may not be as good as those in treated eyes considering the neurodegeneration that occurs due to prolonged edema. Macular edema decreases as early as 1 week after the initial treatment, and best-corrected visual acuity (BCVA) improves in response to this anatomical recovery [4,5,6,7,8]. However, recurrent macular edema may occur, which should be treated as soon as possible to protect the retinal neural elements and obtain better visual outcome.

Clinicians have used a pro re nata (PRN) regimen for anti-VEGF treatment of macular edema due to BRVO [11,12,13]. Thus, recurrences of macular edema must be detected early to ensure that appropriate additional treatments are delivered. Monthly follow-up visits may be helpful in this regard; however, it is not practical to implement this schedule in all patients with BRVO. Instead, it would be useful to determine which patients should receive frequent follow-up or be treated with a treat and extend regimen to avoid recurrent macular edema. Researchers must therefore develop methods to predict recurrence of macular edema due to BRVO.

Previous studies have shown that decreased macular vessel density, as determined with optical coherence tomography angiography (OCTA) [11] and fluorescein angiography (FA) [14], is related to recurrence after anti-VEGF treatment. However, before treatment, it may be difficult to perform OCTA and FA because massive retinal hemorrhage and/or edema in patients with fresh BRVO can cause segmentation error and fluorescence blockade, respectively.

In contrast, sectional images of the retina can easily be obtained using OCT. In this regard, previous reports have focused on findings from the external limiting membrane (ELM), ellipsoid zone (EZ), and interdigitation zone (IZ) as predictors of visual outcome after macular edema due to BRVO [15,16,17,18]. However, the predictive factors for recurrences have remained unclear.

Because BRVO lesions originate in the inner retinal layers, it may be useful to measure the impact of inner retinal condition with sectional OCT images. A recent report stated that disorganization of the retinal inner layer (DRIL) may influence improvement of visual acuity in patients with central retinal vein occlusion (CRVO) or BRVO after 6 intravitreal injections of bevacizumab delivered monthly [19]. Conversely, 2 additional reports have shown that DRIL is not related to visual outcome in BRVO [20, 21]. However, these prior studies did not assess recurrence. DRIL was originally reported as a biomarker for use in predicting the prognosis of diabetic macular edema in terms of visual acuity [22, 23] and was later reported to be related to abnormal blood flow [24, 25]. Recurrence of macular edema can theoretically occur in relation to the vulnerability of the blood vessels. Thus, we hypothesized that factors related to abnormal blood flow may predict the risk of recurrent macular edema.

In the present study, we first analyzed the influence of recurrent macular edema on visual acuity to determine the impact of recurrence predictors, then, we analyzed factors that may be predictive of recurrence by using retinal sections obtained with OCT at baseline (prior to treatment). Information regarding these predictive factors may be useful for clinicians in managing macular edema due to BRVO.

Materials and methods

This retrospective study was based on a detailed medical chart review; it followed the tenets of the Declaration of Helsinki and was approved by the Ethics Committee of Keio University School of Medicine. All patients provided informed consent for the use of their data for research purposes.

Study participants

In total, 117 consecutive eyes of 117 patients were diagnosed with BRVO. From these, we included in the present study 65 eyes of 65 patients with BRVO-induced macular edema who had received intravitreal ranibizumab (IVR) monotherapy at the Medical Retina Division Clinic of the Department of Ophthalmology, Keio University Hospital (Tokyo, Japan) between October 2013 and August 2017. All patients had attended our clinic for at least 12 months, during which no treatment other than IVR was administered. Patients were excluded if they had a history of any anti-VEGF treatment (10 patients) or photocoagulation (12 patients), regardless of both with or without corticosteroid use within 6 months prior to the initial IVR injection. Those who received corticosteroid and/or photocoagulation therapy (12 patients), who had been sent back to the local clinicians (2 patients), as well as those who gave up or denied IVR upon recurrence for patient-specific reasons (7 patients), and who defaulted visits were lost to follow-up (9 patients) within 12 months after the first IVR treatment were also excluded.

Ophthalmologic examinations

At each follow-up visit, all patients underwent BCVA measurement using refraction tests, slit-lamp examinations, and binocular indirect ophthalmoscopy—ophthalmoscopy was performed after pupil dilation with 0.5% tropicamide. BCVA improvement or worsening was defined as a change of > 0.2 in LogMAR.

FA

FA was performed and fundus photographs were obtained using a Topcon TRC 50DX retinal camera (Topcon Corporation, Tokyo, Japan) to diagnose BRVO and macular edema and to evaluate foveal hemorrhage and distance from disc to occlusive lesion.

OCT

OCT images were recorded at every follow-up visit using a Heidelberg Spectralis OCT system (Heidelberg Engineering GmbH, Dossenheim, Germany). The images were used to evaluate the central retinal thickness (CRT), central choroidal thickness (CCT), DRIL, subretinal fluid build-up, macular edema, and disruptions of ELM and EZ. CRT was defined as the distance between the internal limiting membrane and the putative retinal pigment epithelium at the fovea. CCT was defined as the distance from the hyper-reflective line corresponding to Bruch’s membrane beneath the retinal pigment epithelium and the inner surface of the sclera at the foveal center. DRIL was identified when the boundaries between the ganglion cell layer and inner plexiform layer complex, inner nuclear layer, and outer plexiform layer (OPL) could not be distinguished, as described in a previous report [25]. DRIL within 500 μm of the fovea in nasal, temporal, superior, and inferior directions was defined in respective horizontal and vertical OCT images (Fig. 1a–c). Measurements were performed using the caliper function of the OCT device, with manual adjustment. Image analyses were performed by 2 retina specialists (MS and YO) in a blinded manner; if the results differed between the 2 specialists, a 3rd specialist (NN) was asked to perform analysis and results were determined by consensus.

Fig. 1
figure 1

Disorganization of the retinal inner layer (DRIL) in optical coherence tomography (OCT) images in a representative eye. DRIL was analyzed within 500 μm of the fovea in both the vertical and horizontal directions (a). OCT images (b, horizontal section; c, vertical section) of a left eye with macular edema due to branch retinal vein occlusion at baseline. DRIL was observed temporal to the fovea within 500 μm of the fovea (bi-arrows). However, it was not observed on other sides of the fovea

Full size image

Intravitreal ranibizumab monotherapy and follow-up

During the induction phase of the treatment, ranibizumab (0.5 mg, 0.05 mL) was injected intravitreally via the pars plana under sterile conditions. This injection was then repeated as needed (PRN) if follow-up OCT showed macular edema. The need for retreatment was determined by retinal specialists (MS, NN, HS, and YO). Follow-ups were conducted every 1 month after therapy initiation until the macular edema disappeared with the PRN regimen of anti-VEGF therapy. When no macular edema was detected for more than 2 months, the follow-up interval was extended to 2 months.

Statistical analyses

Data are expressed as mean ± standard deviation. Commercially available software (IBM SPSS Statistics, version 24.0; IBM Japan, Tokyo, Japan) was used for all statistical analyses. The Wilcoxon signed-rank test, Mann–Whitney U test, chi-squared test, and logistic regression analyses were used. All p values < 0.05 were considered to be statistically significant.

Results

Sixty-five eyes of 65 patients (male, 33 eyes [50.8%]; mean age, 66.5 years; Table 1) were included. The mean number of injections was 1.9 ± 1.1 (range 1–5) for the initial resolution of macular edema and 3.5 ± 1.8 (range 1–7) over the 12-month follow-up period. Eleven eyes (16.9%) required no additional injections during the 12-month follow-up period after the initial injection. Recurrence of macular edema after remission was observed in 40 eyes (61.5%) during the 12-month follow-up period (Table 2).

Table 1 Baseline characteristics
Full size table
Table 2 Differences between eyes with or without recurrent macular edema
Full size table

Overall, the mean BCVA (Fig. 2a) and CRT (Fig. 2b) were improved at 3, 6, and 12 months, compared with the baseline values (p < 0.01 for all); this improvement was observed regardless of the presence of recurrent macular edema after initial resolution of macular edema (Fig. 2a, b, p < 0.01 for all). BCVA improved or was maintained in 62 eyes (95.4%) after 12 months (data not shown).

Fig. 2
figure 2

Mean best-corrected visual acuity (BCVA) and central retinal thickness (CRT) at each time point. Mean ± standard deviation of the BCVA (a) and CRT (b) at baseline and at 3, 6, and 12 months after initial intravitreal ranibizumab therapy with PRN regimen. Overall, the mean BCVA (a) and CRT (b) were improved at 3, 6, and 12 months, compared with baseline values, in the recurrent and non-recurrent groups by Wilcoxon signed-rank test (**p < 0.01 for all). Mean BCVA did not differ at baseline between the 2 groups; however, the values at 3, 6, and 12 months after therapy were worse in the recurrent group than in non-recurrent group, as determined by the Mann–Whitney U test (a, ††p < 0.01 for all). Mean CRT was greater in the recurrent group at baseline and at 3 months by Mann–Whitney U test (b, ††p < 0.01 for both); however, at 6 and 12 months, it was comparable between the groups

Full size image

Interestingly, when the mean values between the groups with and without recurrence were compared, mean BCVA at each time point after treatment, including at the 12-month follow-up, was worse in patients who had recurrent macular edema (Fig. 2a, p < 0.01 for all), whereas the initial mean BCVA was nearly identical in both groups (Fig. 2a, Table 2, p = 0.396). Mean CRT was different between the groups at baseline (Table 2, p < 0.001) and at the 3-month follow-up (Fig. 2b, p < 0.01); however, the values became comparable at the 6-month follow-up, with additional injections as needed (Fig. 2b).

Differences in the initial conditions of the eyes with and without recurrence are shown in Table 2. In eyes with recurrence, the mean age of the patients was older (68.8 ± 10.0 years vs. 62.7 ± 9.6 years; p = 0.01); the mean CRT was greater (563.3 ± 194.3 μm vs. 380.2 ± 117.2 μm; p < 0.001); and the incidences of DRIL nasal (73% vs. 28%; p < 0.001), inferior (73% vs. 32%; p = 0.001), and temporal to (80% vs. 24%; p < 0.001) the fovea were higher, relative to those in eyes without recurrence. The incidences of disruptions of ELM and EZ did not differ between eyes with and without recurrence. The total number of injections required during the 12-month follow-up period was significantly greater in eyes with recurrence (Table 2, 4.4 ± 1.6 vs. 2.0 ± 1.1 injections, respectively; p < 0.001). The mean duration to the first recurrence after the first remission was 4.4 ± 5.8 months (data not shown). The incidence of DRIL temporal to the fovea was also higher at the time of first macular edema remission and at the 12-month follow-up in eyes with recurrence (Supplementary Table 1).

Next, the risk of recurrence was analyzed using logistic regression analysis (Table 3). Crude analysis showed that increased age (p = 0.01); increased CRT (p = 0.001); and eyes with DRIL nasal (p = 0.002), inferior (p = 0.004), and temporal (p < 0.001) to the fovea initially exhibited increased risk for recurrence of macular edema. Moreover, after adjustment for age, gender, and initial CRT at baseline, eyes with DRIL temporal to the fovea at baseline exhibited a significantly greater risk of recurrence (odds ratio = 7.74, 95% confidence interval 1.62–37.08, p = 0.01), indicating that DRIL temporal to the fovea was an independent risk factor for recurrent macular edema.

Table 3 Risk factors for recurrent macula edema within 12 months
Full size table

Discussion

In the present study, we showed that, in eyes with macular edema due to BRVO, both BCVA and CRT were improved by IVR monotherapy after 12 months, using a regimen of 1 initial injection followed by a PRN approach. However, recurrent macular edema occurred in 61.5% of the eyes; recurrence affected BCVA at the 12-month follow-up. The initial CRT was greater in the group with recurrent macular edema; however, CRT decreased and became comparable to that in the non-recurrent group with treatment. Presence of DRIL temporal to the fovea was an independent risk factor for recurrent macular edema due to BRVO, as shown by logistic regression analysis after adjustment for age, gender, and initial CRT.

Both mean BCVA and CRT were improved rapidly 3 months after the initial IVR injection, corroborating previous reports of prospective clinical trials using either ranibizumab or aflibercept injection [6, 26]. In these trials, 6 injections were delivered at monthly intervals, followed by PRN treatment thereafter. The current study observed similar outcomes with only 1 injection followed by a PRN regimen. In addition, none of the patients in the current study required 6 consecutive injections to initiate therapy until the first remission of macular edema, with monthly follow-up until the first remission (data not shown).

The mean number of injections delivered in the current study was 3.5, corroborating the findings of a previous study by Miwa et al. [27], which showed that 3.8 injections were required, using the same protocol for PRN treatment. Macular edema recurred in 61.5% of patients in the current study, which was consistent with the results of a previous study by Hasegawa et al. [11] (59.3%); they reported respective injection numbers of 5.1 and 2.1 in eyes with and without recurrence, which were also consistent with the findings of the present study.

Notably, the visual outcome was worse in eyes with recurrent macular edema, although CRT was successfully reduced to levels similar to those in the non-recurrent group. However, retinal neurons may have been more severely affected by changes in the microenvironment, such as cytokine and ionic changes, related to repeated macular edema over a prolonged period. Although VEGF plays a neuroprotective role [28, 29], the extracellular fluid that causes edema may contain various neurotoxic cytokines that can damage the retinal neurons; previous reports have shown that levels of various inflammatory cytokines (e.g., interleukin-6 and monocyte chemotactic protein-1) were increased in the vitreous fluid of BRVO patients [30].

Alternative or additional pathways may contribute to visual outcomes. Previous studies regarding visual outcomes have focused on the outer layers of the retina, such as the ELM, EZ, and IZ [15,16,17,18, 20]. In contrast, there were no significant differences between the recurrence of macular edema and the incidences of ELM or EZ disruption in the current study. Nevertheless, eyes with recurrent macular edema had worse visual outcomes, likely because recurrence was associated with DRIL, which is reportedly related to visual function, as demonstrated in diabetic retinopathy patients [22]. DRIL represents a disorganized microenvironment involving synapses between photoreceptors and secondary neurons (bipolar cells) that construct the OPL. In addition, a persistent synaptic abnormality may cause retrograde damage to photoreceptors. Indeed, this idea is supported by the observation that visual prognosis is related to the flow in the deep capillary plexus of the inner retinal layer [31, 32]. Another previous report stated that non-perfused areas in the inner layer, which can be detected by FA, correspond to EZ disruption that reflects photoreceptor damage [33]. Thus, abnormal inner retinal blood flow and recurrent macular edema may cause damage to photoreceptors. Analysis of a hyper-reflective band involving the inner retinal layer in OCT images, known as paracentral acute middle maculopathy (PAMM), showed that ischemic lesions in vascular occlusive diseases of the retina develop at the venular pole of the deep capillary plexus [34] and may then progress laterally and anteriorly into the inner retina at the level of the superficial capillary plexus. This ischemic mechanism also supports our abovementioned hypothesis that inner retinal findings involve photoreceptor disorder.

Importantly, recurrence was associated with the presence of DRIL temporal to the fovea, but not with the disruption of either ELM or EZ. Given that DRIL is related to flow abnormality in the deep plexus [24], the areas affected by DRIL may have experienced persistent hypoxia, leading to excessive VEGF expression and thus recurrent macular edema. Alternatively, hypoxia may have interrupted the recovery of vessels that were influenced at the time of occlusion due to insufficient nutrient and oxygen supply to the affected vascular cells. This interpretation of our results is consistent with a previous report using FA, which found that 35% of patients showed recurrence 6 months after a single injection of an anti-VEGF drug; the risk was increased if the patients had a non-perfused area in the deep capillary plexus within 1 mm of the fovea [14]. Notably, the previous study using FA had a limitation in the recording of clear images if the eyes had severe hemorrhage causing blockade of the signals [14]. In contrast, the current study did not have this limitation; the inner retinal layers in the sectional OCT images could be evaluated regardless of the extent of hemorrhage and edema before treatment. DRIL, as identified in sectional OCT images, would be of value for predicting recurrent macular edema in daily clinical practice.

DRIL temporal to the fovea was responsible for the recurrence of macular edema. Notably, retinal vessels temporal to the fovea are at a greater distance from the disc, and the presence of microvascular changes on the temporal side may result in a greater area of vascular abnormality. This would increase the risk of extracellular fluid accumulation. Alternatively, because the capillary densities in the neurofiber layer, ganglion cell layer, inner plexiform layer, and inner nuclear layer are lower on the side temporal to the fovea [35], we speculate that further reductions in blood flow caused by BRVO, as suggested by DRIL, could lower the ability of the retina to absorb extracellular fluid, thereby resulting in edema.

The limitations of the current study are its relatively small sample size and retrospective design; moreover, the need for retreatment was determined by attending doctors, rather than by the reading center. However, recurrent macular edema was recorded in OCT images that were reviewed during the current study, and we confirmed that retreatment was performed properly, based on the criterion, if follow-up OCT showed macular edema.

In summary, visual prognosis was worse in eyes with recurrent macular edema, despite reduction in CRT that indicated sufficient treatment. DRIL temporal to the fovea was a significant risk factor for recurrence. Use of OCT sectional images is valuable for diagnosis of macular edema, as well as for evaluation of DRIL temporal to the fovea at baseline; this can help to predict prognosis and determine follow-up schedules for early retreatment of recurrent edema and/or the treatment protocol in the daily clinic.