Impact of serous macular detachment on visual recovery in retinal vein occlusion treatment

Submitted for publication: December 8, 2020 Accepted for publication: April 6, 2021 Funding: This study received no specific financial support. Disclosure of potential conflicts of interest: None of the authors have any potential conflicts of interest to disclose. Corresponding author: Ilknur Tuncer Firat. E-mail: ilknurtuncer89@gmail.com Approved by the following research ethics committee: Inonu University (#2018/9-11). ABSTRACT | Purpose: The aim of this study was to evaluate the effect of serous macular detachment observed during retinal vein occlusion on treatment results. Methods: A total of 117 eyes from 115 patients who had been treated with intravitreal injections for macular edema secondary to retinal vein occlusion were retrospectively reviewed. Visual acuity, optical coherence tomography, and fundus fluorescein angiography findings were evaluated according to the status of serous macular detachment. Results: In the branch retinal vein occlusion group, a statistically significant increase was detected in the mean visual acuity compared to the baseline value at each visit in the absence of serous macular detachment, whereas the increase in the mean visual acuity was significant only at the 3and 6-month visits in the presence of serous macular detachment. In the central retinal vein occlusion group, there was an increase in the mean visual acuity compared to the baseline value at every visit in the absence of serous macular detachment, whereas the mean visual acuity decreased compared to the baseline value at every visit except at the 3-month visit in the presence of serous macular detachment. The ellipsoid zone defect was more prominent in the presence of serous macular detachment in eyes with branch retinal vein occlusion, whereas there was no significant difference in the ellipsoid zone in the absence or presence of serous macular detachment in eyes with central retinal vein occlusion. Conclusions: In the group with macular edema due to retinal vein occlusion, the initial mean visual acuity increase observed in the first year was maintained in cases without serous macular detachment but not in those with serous macular detachment. Serous macular detachment could be a negative factor in eyes with retinal vein occlusion.


INTRODUCTION
Retinal vein obstruction (RVO) is the most common retinal vascular disorder that develops after diabetic retinopathy (1) . One of the basic reasons for vision loss in RVO is macular edema (2) . Some eyes with macular edema in RVO have demonstrated a less pronounced response to the same treatments than other eyes (3)(4)(5)(6) . This finding has led to the investigation of various prognostic factors. Previously detected prognostic factors for branch retinal vein obstruction (BRVO) include initial visual acuity (VA), age, BRVO duration, macular ischemia, cystic areas of >600 µm in the fovea, and defects in the ellipsoid zone and the external limiting membrane (ELM) (4)(5)(6)(7) . Some studies have reported that serous macular detachment (SMD) secondary to RVO has a positive or negative effect on the results, whereas other studies have found no effect (4,6,8) . The aim of this study was to determine the effect of SMD on the treatment and prognosis of macular edema due to RVO in a large patient group.

METHODS
We retrospectively reviewed the charts of patients who had been diagnosed with macular edema due to RVO at the İnönü University Faculty of Medicine Turgut Özal Medical Center's Ophthalmology Clinic between January 2011 and July 2017 and subsequently treated with an intravitreal (IV) sustained-release dexamethasone implant (Ozurdex; Allergan, Inc., Irvine, CA, US), bevacizumab (Avastin; Genentech Inc., South San Francisco, CA, US), ranibizumab (Lucentis; Genentech Inc., South San Francisco, CA, US), or aflibercept (Eylea, Regeneron Inc, NY, US). This study was conducted according to the principles of the Declaration of Helsinki and patient treatment. Consent from the Malatya İnönü University Ethics Committee was obtained before conducting the study (İnönü University Scientific Research and Publication Ethics Committee -Health Sciences Noninterventional Clinical Studies Ethics Committee -Decision No: 2018/9-11).
Patients who participated in the first visit and at least one of the 3-, 6-, 9-and 12-month visits with an optical coherence tomography (OCT) image suitable for evaluation were included in the study. The last visit the patient could attend was accepted as the final visit.
We excluded patients treated at an external center. Those with the following conditions were also exclu-ded: presence of choroidal neovascularization, diabetic retinopathy, or significant media opacity; vitreomacular disorders at baseline; a history of other disorders that could affect vision such as retinal artery occlusion and ocular inflammation; past vitreoretinal surgery; grid laser treatment to the macula; and poor OCT image quality.
Before the injection and laser treatment, written and verbal informed consent was obtained from all patients. Patients recruited early in the study could not initially receive anti-vascular endothelial growth factor (VEGF) treatment for RVO-related macular edema because of limitations in the Turkish National Health System and were instead treated with IV sustained-release dexamethasone twice a year at most. After the approval of IV anti-VEGF treatment for RVO, patients who had not responded or had demonstrated a poor response to previous IV sustained-release dexamethasone treatment were started on one of the IV anti-VEGF treatments as an additional treatment. After the approval of IV anti-VEGF treatments, patients with RVO receiving treatment for the first time were primarily assigned to one of the anti-VEGF options, but an IV sustained-release dexamethasone implant was placed later in case the treatment response was inadequate. After IV anti-VEGF treatment had been administered monthly for 3 months, or after the first IV sustained-release dexamethasone treatment was administered, the patient was called in for a 3-month follow-up if there was a good response or for monthly treatment if the response was poor.
The criteria for treatment initiation were the presence of a central foveal thickness (CFT) of >250 µm and/ or a significant intraretinal cyst or subretinal fluid due to RVO. The criteria for retreatment and treatment continuation were an increase of >50 µm in CFT since the last visit and/or a marked increase in intraretinal cysts, an inadequate response to previous treatment, and a loss of two or more lines in VA on the ETDRS scale. IV treatment was stopped if there was no response to three consecutive IV injections, and patients were called for a follow-up visit 3-6 months later according to their risk for other complications.
Data regarding age, sex, the presence of diabetes mellitus or hypertension, eye with RVO, RVO type, ischemic status, VA (logMAR), CFT measurement (µm) by OCT (spectral-domain OCT), the presence of SMD, the presence of foveal ellipsoid zone and ELM damage at the final visit, the presence of initial macular ischemia, the type of injection administered, and the number of doses were collected. The results of fundus fluorescein angiography (FFA) were considered to be ischemic if the area was >5 disk areas for BRVO and 10 disk areas for central retinal vein occlusion (CRVO) (9) . CFT was defined as the mean thickness within the central, 1-mm-diameter foveal area and was obtained from the OCT thickness map. SMD was defined as the presence of any hyporeflective area between the neurosensorial retina and RPE, together with macular elevation. Any disruption of the ellipsoid zone or ELM at the fovea was accepted as the presence of ellipsoid zone or ELM damage. Patients were evaluated in two groups as BRVO and CRVO according to the type of RVO. They were further divided into two subgroups as SMD absent (SMD (−)) and SMD present (SMD (+)).

Statistical analysis
The SPSS Windows version 22.0 (SPSS Inc., Chicago, IL, US) software was used for statistical analysis. Results are expressed as mean ± standard deviation (SD). The Kolmogorov-Smirnov test was used to determine whether the quantitative data conformed to a normal distribution. In the dependent groups, a paired t-test was used for data with a normal distribution, and the Wilcoxon signed-rank test was used for data without a normal distribution, and the respective tests used in the independent groups were an unpaired t-test and the Mann-Whitney U test. Qualitative variables were evaluated using Pearson chi-square or Fisher's exact tests. The level of statistical significance was set at p<0.05.

RESULTS
A total of 117 eyes from 115 patients were included in this study, of which 75 eyes (64.1%) had BRVO, and 42 eyes (35.9%) had CRVO. There were 62 (53.0%) SMD (−) eyes and 55 (47.0%) SMD (+) eyes in the entire group. No significant difference was found in the mean age, sex, presence of hypertension or diabetes mellitus, laterality, RVO type, presence of ischemia, initial mean VA, or mean CFT value between the groups when divided according to the SMD status (Table 1). SMD was present in 31 (41.34%) eyes in the BRVO group and in 24 (57.1%) eyes in the CRVO group. There was no difference in sex, presence of hypertension and diabetes mellitus, laterality, presence of ischemia, or the total number of injections according to the SMD status in the BRVO and CRVO groups (p>0.05). Among patients with BRVO, the mean age was 60.63 ± 11.78 years in the SMD (−) group and 59.22 ± 12.12 years in the SMD (+) group (p=0.830). in the CRVO group, the mean ages were 64.78 ± 6.25 and 55.0 ± 15.43 years, respectively (p=0.008).
The mean number of anti-VEGF injections was 2.37 ± 1.46 in the SMD (−) group and 2.43 ± 1.43 in the SMD (+) group (p=0.828). The mean number of Ozurdex injections was 1.71 ± 0.91 in the SMD (−) group and 1.56 ± 0.97 in the SMD (+) group (p=0.281). The distribution of IV treatments was similar in SMD (−) and SMD (+) cases in both the BRVO and CRVO groups ( Table 2).
In the absence of macular ischemia, in the SMD (−) group (n=50), the mean initial and final VA values were 0.96 ± 0.53 and 0.66 ± 0.51 logMAR (p<0.001), respectively, and in the SMD (+) group (n=43), the respective values were 0.79 ± 0.44 and 0.64 ± 0.48 logMAR (p=0.056). In the presence of macular ischemia, in the SMD (−) group (n=12), the mean initial and final VA values were 1.43 ± 0.66 and 1.38 ± 0.37 logMAR   Arq Bras Oftalmol. 2022 -Ahead of Print (p=0.498), respectively, and in the SMD (+) group (n=12), the respective values were 1.45 ± 0.72 and 1.51 ± 0.71 logMAR (p=0.678). Ellipsoid zone damage was significantly more prominent in SMD (+) eyes than in SMD (−) eyes in the BRVO group (p=0.032). It was also more prominent in SMD (+) eyes than in SMD (−) eyes in the CRVO group, but there was no statistically significant difference (p=0.276). There was also no statistically significant difference between ELM damage rates in SMD (−) and SMD (+) cases in either the BRVO or CRVO group. Only in the absence of macular ischemia did the ellipsoid zone damage increase significantly in the SMD (+) group (p=0.003). In the presence and absence of macular ischemia, the ELM damage was not affected by the presence of SMD (p=0.105) (  (p<0.001), respectively, in SMD (−) cases and 645.29 ± 240.63 and 264.00 ± 120.94 µm (p<0.001), respectively, in SMD (+) cases. No statistically significant difference was found in the mean CFT values between SMD (−) and SMD (+) eyes at any time point in the BRVO and CRVO groups. The mean CFT showed a decrease compared to the baseline value at all time points in both SMD (−) and SMD (+) eyes with BRVO or CRVO (Table 5).

DISCUSSION
The mechanism underlying SMD development is still not completely clear. The possible causes of subretinal fluid collection are vascular leakage from the retinal and/or choroidal circulation and disturbed fluid removal function of the retinal pigment epithelium secondary to increased inflammation (10,11) . The fact that both macular edema and SMD respond to anti-inflammatory treatment confirms the effect of inflammation in RVO on the pathogenesis of both conditions. Some studies have reported that increased VEGF levels in the vitreous and aqueous humor support the role of increased inflammation in SMD development (12,13) . Inflammation plays a significant role in SMD development; however, the fact that SMD is not observed in every RVO case indicates that inflammation is not the only factor in the mechanism underlying SMD development. Therefore, the effect of inflammation on SMD development can be explained by changes in the natural barriers and the microstructure (7) .
The possible involvement of the outer blood-retina barrier during SMD development indicates a pathological spread of the process of RVO from inner retinal layers to outer retinal layers. The inflammatory mediators that pass toward outer retinal layers could play a role in this spread. The ELM consists of cells in the interface between the intercellular connections of the inner segment of photoreceptors and Müller cells (14) , and its pores are not permeable to large serum proteins such as albumin and globulins (15,16) . It is possible that inflammatory molecules of a size that can pass through the ELM pores can also pass through the intercellular space (17) , and the outer blood-retina barrier could be broken down due to the effect of these molecules (14,18,19) . Therefore, proteins can collect in the subretinal area (20,21) and draw water by crea ting a hyperosmotic section compared to that in inner retinal layers. Moreover, the decreased water removal capacity of the RPE due to inflammation can increase the collection of fluid beyond capacity and result in SMD development by opening up the potential space within the subretinal region. However, ELM defects due to inflammation can develop at later periods and prevent the function of the ELM osmotic barrier. Proteins can easily pass through the ELM defects when the ELM loses its protein barrier function, thereby preventing the development of an osmotic gradient between the two sides of the ELM (14) . This thesis is supported by the facts that there were only a few SMD relapses despite severe macular edema relapses and no case with SMD development without RVO initiation in this study. Early ELM damage and inflammation that does not cause adequate disruption of the other blood-retina barrier could explain the other cases in which no SMD development was observed.
SMD is a complication reported at a rate of 28.3%-71.4% in BRVO cases and 50.0%-100.0% in CRVO cases; SMD has been evaluated as a factor that could influence treatment outcomes (5,8,(22)(23)(24) . Similarly, we found the rate of SMD to be 41.3% in eyes with BRVO and 57.1% in those with CRVO. Considering the higher inflammation and vascular leakage in eyes with CRVO than in eyes with BRVO (25) , the higher rate of SMD development in eyes with CRVO could provide some proof that increased inflammation and vascular leakage are associated with SMD development.
Although some studies associate the presence of SMD in RVO with a poor visual prognosis, the effect of SMD on visual prognosis still remains unclear (4,6,18,22,26,27) . The evaluation of our patients after dividing them into BRVO and CRVO groups revealed that the early gain in mean VA in the SMD (+) group was either lost or became insignificant over time in both patients with CRVO and those with BRVO. Furthermore, there was a meaningful increase in the mean VA value at the final visit in SMD (−) BRVO eyes but not in SMD (+) eyes. The mean CFT decreased at all time points in both SMD (+) and SMD (−) eyes with BRVO or CRVO. This result indicates that the final visual prognosis was worse in the SMD (+) group despite the similar anatomic improvement in the two groups. The increase in VA in the early period could have been obtained as a result of the recovery of contact between the neurosensorial retina and RPE following successful SMD treatment. However, this increase in VA could be lost over time because of the larger quantity of inflammatory factors (12,13) , the mechanical photoreceptor damage that occurs during SMD development, and the gradual increase in metabolic photoreceptor damage as a result of the loss of contact between photoreceptors and RPE due to SMD (7,22,28) . The absence of a significant difference between the SMD groups with regard to the final CFT result has similarly been reported in several studies and supports our findings (4,13,18,22) .
Our results demonstrate that the therapeutic potential of IV treatments used for macular edema of RVO origin also applies to SMD treatment. Although there were patients with SMD recurrence, all recurrences had been resolved by the final visit. Nevertheless, the low number of patients with SMD recurrence prevents us from reaching a conclusion regarding the prognosis. The present medical treatments cannot improve or recover the photoreceptor damage that develops before treatment but can decrease vascular damage and leakage and therefore treat SMD and macular edema with every administration. This could explain the differentiation of CFT and VA results in the SMD (+) group over time.
The negative effect of SMD on visual prognosis is also supported by the OCT findings. Tsujikawa et al. reported a significantly more prominent IS/OS defect in the SMD (+) group, but they did not find a significant difference in ELM defects between the SMD (−) and SMD (+) groups (7) . Similarly, we found more prominent ellipsoid zone damage in SMD (+) BRVO eyes. In contrast, ellipsoid zone damage was similarly distributed in SMD (−) and SMD (+) cases in the CRVO group. We believe that this could be the result of the much larger ellipsoid zone damage potential from CRVO itself rather than from SMD. Consequently, the ellipsoid zone damage of SMD may be masked by CRVO. Similarly, due to the masking of macular ischemia, the negative effect of SMD on the ellipsoid zone becomes evident only in the absence of macular ischemia. Our microstructural findings are supported by similar findings in the course of VA.
We found no significant difference in ELM damage based on the presence of SMD irrespective of the RVO type and macular ischemia status. This result may indicate that ELM damage is more a part of the RVO process than a harmful effect of SMD. When eyes with and without ELM damage were examined in two separate groups, the mean VA of those with ELM damage was lower at baseline and at the final visit. However, no significant difference was detected in the mean VA based on the presence of SMD in each group. These results indicate that ELM damage is associated with poor visual prognosis from the beginning and that ELM status is more a determinant of VA than SMD status. The limitations of our study include its retrospective nature, the wide range of options for the treatment regimen, the possibility that the result was influenced by the chosen treatment, and the small sample size for the evaluation of some subgroups.
In conclusion, we detected a negative effect of the presence of SMD on VA in eyes with BRVO or CRVO. We believe that the reason for this finding is the additional photoreceptor damage due to inflammatory, mechanical, or metabolic causes in SMD (+) eyes. Moreover, it is possible that ELM integrity plays a significant role in the development and recurrence of SMD. Nonetheless, further randomized controlled studies on larger series are required to clarify these results.