Management of macular oedema due to retinal vein occlusion: An evidence‐based systematic review and meta‐analysis

Central retinal vein occlusion and branch retinal vein occlusion are common causes of visual loss due to associated macular oedema. The aim of this review was to assess the effectiveness of interventions improving vision and treating macular oedema in central retinal vein occlusion and branch retinal vein occlusion.


| INTRODUCTION
Retinal vein occlusion (RVO) is the second most common vision-threatening vascular disorder of the retina after diabetic retinopathy. It can be categorised based on the location of the luminal obstruction of the venous outflow system within the retinal vasculature. In central retinal vein occlusion (CRVO), blockage of the central retinal vein occurs within the optic nerve at the level of, or posterior to, the lamina cribrosa. All retinal veins and venules appear dilated and tortuous often with retinal haemorrhages in all four quadrants of the retina. In hemiretinal vein occlusion, signs are restricted to the superior or inferior half of the retina. 1 In branch retinal vein occlusion (BRVO), occlusion tends to occur at an arteriovenous crossing with haemorrhages localised to the area drained by the branch retinal vein. 2 CRVO is less common than BRVO with a reported prevalence of 0.1%-0.4%. In eyes with CRVO, visual impairment is most commonly due to the development macular oedema (MO) but can also be caused by macular ischemia, and/or neovascular glaucoma. 3 The pathophysiology of CRVO is believed to be due to the atherosclerotic changes occurring in the central retinal artery affecting the central retinal vein following the principles of Virchow's triad for thrombogenesis. The vein and artery share a common adventitial sheath and atherosclerotic changes of the artery can cause compression of the vein at or proximal to the lamina cribrosa of the optic nerve. Patients with CRVO have similar systemic vascular risk factors to patients with cardiovascular disease, including hypertension, age over 55 years, hyperlipidaemia, diabetes, and smoking. 4 Younger patients are more likely to have other coexistent systemic disease including hypercoagulability.
The estimated prevalence of BRVO is 0.7%. 5 Like CRVO, BRVO can have multiple underlying causes, including age, hypertension, diabetic retinopathy, or hypercoagulability. The pathophysiology of BRVO involves increased hydrostatic pressure within thinwalled veins proximal to a luminal obstruction. 3 The superotemporal quadrant in the retina is the most commonly affected in 63%-66% of eyes. 6 Similar to eyes with CRVO, the most common cause of vision loss in BRVO is due to the development of MO. 3,7 Hypoxia and vascular endothelial growth factor upregulation occur due to the resistance to outflow in eyes with RVO. The resultant hypermeability leads to exudation and MO. The upregulation of vascular endothelial derived growth factor can also lead to further progression of retinal ischaemia. As MO is the primary cause of visual loss in patients with RVO, various treatments have been trialled aimed at treating MO due to RVO. 8 Prior to the advent of intravitreal therapy, there was little evidence of any effective treatment to improve vision in eyes with MO due to CRVO. Whereas macular laser offered modest improvement in eyes with BRVO. 9 The purpose of this review was to examine the effectiveness of macular laser, current intravitreal anti-VEGF agents and intravitreal corticosteroids that have been studied in large cohorts in randomised controlled trials (RCTs) and real-world outcome studies (RWOS). We aim to provide a narrative review of evidence of individual studies and meta-analyses to overview the strength of the studies included. This contrasts with previously published meta-analyses.

| METHODS
This systematic review is reported in accordance with the preferred reporting items for systematic reviews (http:// www.prisma-statement.org; accessed 20 May 2022).

| Types of interventions
Randomised controlled studies for the treatment of MO due to CRVO and BRVO comparing any known therapies (including laser treatment, intravitreal therapies such as anti-VEGF agents or corticosteroid and no/sham treatment) were included in this review. We included all treatment algorithms (fixed, as needed [PRN] and treat and extend [T&E]), as the goal was to assess efficacy of treatments irrespective of their treatment algorithm. We also included RWOS with visual outcomes whether anatomical outcomes were reported or not.

| Outcome measures
Outcome measures included the mean change in best corrected visual acuity (BCVA) from baseline to end of study, and the mean change in MO usually measured as a change in central macular thickness (CMT) on optical coherence tomography (OCT). Where available, ocular adverse events were also reported.

| Study selection process and data extraction and management
Articles were retrieved using a keyword search of medical internet search engines-Cochrane, PubMed, MEDLINE Ovid; the ISRCTN registry; ClinicalTrials.gov; and the WHO ICTRP. The date of the last search was August 2022. Two investigators (E. C., S. Z.) independently identified all RCTs which evaluated the management of MO in CRVO or BRVO with at least 90 patients and RWOS of CRVO or BRVO with more than 100 patients, and with a follow up period of at least 6 months. RCTs that investigated only the dose and duration of treatment without a comparator were excluded. Discrepancies between reviewers were resolved by discussion and referral with a third party (S. F. B.).
Keywords search used was-BRVO/CRVO and macular (o)edema

| Meta-analysis
Two reviewers (S. F. B., K. S.), working independently and in replica, evaluated the risk of bias using the Cochrane Risk of Bias Tool 9 for RCTS, and the Risk of Bias in Nonrandomised Studies-of interventions (ROBINS-I) tool. 10 The risk of bias for RCT studies was considered according to the subsequent fields; selection bias, reporting bias, and other bias. Then the articles were classified into 'low risk', 'high risk', or 'unclear risk' of bias for each domain. For cohort studies, the risk of bias was evaluated through its selection, comparability, and result.
Study findings were summarised with regards to trends in BCVA improvement over time. Meta-analyses of visual outcomes were performed using ETDRS letters. Studies that reported visual acuity in logMAR or decimal values had those values approximated to ETDRS letter scores 11 with an approximate SD. 12 Studies that presented outcome data at various time intervals were included in the meta-analysis using pairwise associations of treatments evaluated at each of the different time points. The summary measure of primary outcome was the difference in mean change of VA from baseline. Subgroup analyses were performed at 6, 12, 24 and 36 months, as results at these time points were reported most commonly in studies. A post-hoc analysis was performed in order to compare the outcomes of various therapies used to treat MO due to RVO. Treatment effect estimates were assessed for all patients as well as subcategories based on VA.
Comprehensive Meta-Analysis version 4.0 was employed to generate the meta-analysis and to evaluate the data from included studies. The standardised mean difference (SMD) between the two groups was calculated using an inverse variance statistical approach with 95% confidence interval (CI) as the effect measure. To characterise any heterogeneity among studies, random effects analysis was used. The chi-square test and the 12 value were used to evaluate the statistical heterogeneity between studies. p Values below 0.10 in chi-square analyses were regarded as indicators of statistically significant heterogeneity. This was quantified using I 2 statistics. I 2 > 30% was deemed to indicate moderate heterogeneity, I 2 > 50% substantial heterogeneity and I 2 > 75%. considerable heterogeneity. Studies with substantial or considerable heterogeneity were examined for possible sources of heterogeneity.

| Study selection
Following a search of databases and search engines, 372 RCTs and 106 RWOS examining the management of MO in CRVO or BRVO were identified ( Figure 1). Studies with less than 6 months follow up and/or less than 90 patients for RCTs and less than 100 patients for RWOS were excluded. A total of 21 RCT and 15 RWOS met the eligibility criteria to be included in this review (Tables 1-5).

| Summary of evidence for CRVO
There were 185 RCTs identified which examined treatment of MO due to CRVO after duplicates were removed. After exclusion of studies with less than 90 eyes and/or less than 6 months follow up, or failure to meet inclusion criteria (studies required visual and/or anatomical outcomes), there were 11 studies which met the inclusion criteria ( Figure 1). Of the 11 included RCTs (Table 1), one compared observation to grid pattern laser (CVOS; Central retinal Vein Occlusion Study), 11 one compared two doses of preservative free triamcinolone acetonide to sham (SCORE Standard Care vs. cOrticosteroids for Retina vein occlusions), 25 two compared intravitreal dexamethasone implant (Dex-implant) to sham 13,17 and one compared dexamethasone implant (Dex-implant) to ranibizumab (COMARADE C). 16 One study compared ranibizumab to sham (CRUISE), 7 two studies compared aflibercept to sham (COPERNICUS, GALILEO), 15 and three compared two or three of the commonly used anti-VEGF agents; bevacizumab ranibizumab and aflibercept (SCORE 2, 18

| Intravitreal corticosteroids
The first intravitreal agent used in the management of MO due to CRVO was triamcinolone acetonide. The SCORE study compared triamcinolone acetonide at doses 1 and 4 mg to sham/no treatment which was the standard of care at the time. Macular grid photocoagulation had been studied in the Central Vein Occlusion Study, but no significant difference in visual outcome between eyes treated with macular laser and those observed was found at any follow up time point, except there was a trend towards significance in younger patients. 14 In the SCORE study, one intravitreal injection of triamcinolone acetonide improved vision compared to observation at 12 months. 12 The ocular adverse events of raised intraocular pressure (IOP) and cataract progression were less in the triamcinolone acetonide 1 mg group than the 4 mg group leading to the recommendation that triamcinolone acetonide 1 mg dose be used instead of 4 mg. Later the GENEVA (Global Evaluation of implaNtable dExamethasone in retinal Vein occlusion with macular edema) study group compared the use of Dex-implant 0.7 mg and Deximplant 0.35 mg to sham. At both doses there was a reduction in the risk of vision loss and improved speed and incidence of improvement in vision. 13 The GENEVA studies found that earlier treatment of MO was associated with better visual outcomes which peaked around 60 days after treatment. A Chinese RCT comparing the Dex-implant with sham also reported better anatomical and visual outcomes with Dex-implant compared to sham treatment. Anatomical and visual improvement peaked at 3-4 months after injection. Ocular adverse events included an increase in IOP which normalised by 4 months after each injection of DEX-implant, even with repeat treatments. 17

| Intravitreal VEGF-inhibitors
The first large RCT using anti-VEGF was the CRUISE study (Central Retinal vein occlUSIon Study). 14 The study compared two doses of ranibizumab (0.3 and 0.5 mg) to sham in eyes with MO secondary to CRVO. There were significantly better visual and anatomical outcomes in eyes treated with ranibizumab at 6 months compared eyes treated with sham. 14 No differences in AEs between the two groups p = 0.51 6 months Bev given 4 weekly was noninferior to Ran 4 weekly. Anatomic and safety outcomes did not differ between treatment groups.
Low ranibizumab for 7 months followed by PRN treatment (with monthly visits) to fixed monthly injections of ranibizumab for 15 months (SHORE, The Study evaluating dosing regimens for treatment with intravitreal ranibizumab injections in subjects with MO following retinal vein occlusions). Visual and anatomical outcomes did not differ significantly between the two treatment regimens.
The GALILEO and COPERNICUS studies compared aflibercept to sham in eyes with MO due to CRVO 7,15 (VEGF TRAP-EYE: investigation of efficacy and safety in CRVO). Both studies reported better visual and anatomical outcomes in eyes treated with aflibercept compared to sham, particularly when treatment was started early. However, when the patients were switched to PRN treatment and less frequent reviews (three monthly), the effect of treatment diminished. SCORE 2 study compared the effect of bevacizumab to aflibercept given every 4 weeks for 6 months to eyes with MO due to CRVO and found that bevacizumab was a noninferior treatment to aflibercept when given monthly. 18 The LEAVO (Lucentis, Eylea, Avastin in Vein Occlusion) study compared three anti-VEGF agents: ranibizumab, aflibercept and bevacizumab. Three monthly doses were given followed by PRN dosing for 100 weeks. 19 Bevacizumab was non-inferior to ranibizumab but was not noninferior to aflibercept with regards to the mean change in vision. However, bevacizumab was less effective in resolving MO than ranibizumab or aflibercept.
A single Dex-implant given at baseline was compared to ranibizumab given as three 4-weekly loading doses followed by PRN treatment in the COMRADE C trial. 16 At 6 months, vision and anatomical outcomes were better for eyes randomised to ranibizumab than eyes randomised to Dex-implant. There were more ocular adverse events in Dex-implant treated eyes, including cataract, increased IOP and the development retinal neovascularisation.

| Results of meta-analysis of visual outcomes of RCTs for management of MO due to CRVO
A total of nine RCTs reported change in BCVA at 6 months encompassing 3512 eyes. Seven studies included an anti-VEGF arm with a mean improvement in BCVA at 6 months of 15 Figure 2). All nine studies favoured intravitreal treatment compared to sham or no treatment (2.1 in Figure 2). The pooled SMD of the nine studies was 0.97 letters (95% CI 0. 40 In the three studies with 12 months of follow-up (2.2 in Figure 2) the BCVA gains in the anti-VEGF arms, were maintained, with a mean improvement of 15.07 letters (95% CI 12.11-18.04, p < 0.01) from baseline. The IVTA and sham arms trended lower with a mean change of À1.2 letters (95% CI À1.57 to À8.23, p < 0.01) from baseline for IVTA, and À1.54 letters (95%CI À14.12 to 11.03, p = 0.81), for sham. This trend continued at 24 months (2.3 in Figure 2) with two anti-VEGF studies reporting a mean gain of 12.68 letters (95%CI 11.24-14.11, p < 0.01), whereas IVTA treated eyes lost À3.4 letters (95%CI À3.90 to À2.89, p < 0.01), perhaps a result of cataract progression in the IVTA group. However, any intravitreal treatment with anti-VEGF or corticosteroid resulted in significantly better vision than sham or no treatment at 12 months with a SMD of 2.22 letters (95% CI 0.44-4.01, p < 0.01) (2.2 in Figure 2). Table 2) A UK retrospective cohort study of 231 patients with MO secondary to CRVO receiving ranibizumab or aflibercept reported on treatment outcomes. Both drugs were effective in improving vision. 23 However, because only 14% of the cohort received ranibizumab there was inadequate power to compare the relative efficacy between the two agents. The FRB! registry reported outcomes of 296 eyes with MO due to CRVO. Again, it was reported that both ranibizumab and aflibercept were effective treatments. 29 However, there were greater anatomical and visual outcomes seen in eyes treated with aflibercept at 12 months compared to ranibizumab. It is also important to note that despite receiving anti-VEGF, neovascular complications still occurred in the anterior segment (n = 16 eyes) and posterior segment (n = 17 eyes) and were associated with poorer visual outcomes. Laser pan-retinal photocoagulation was performed in 83 eyes. Rubeotic glaucoma more frequently developed in ranibizumab-treated eyes than in aflibercept-treated eyes (12 vs. 4 aflibercept p = 0.01). However the number of injections, irrespective of the agent were also strongly associated with rubeotic glaucoma ( p < 0.001). 29 Ciulla et al. 24  T A B L E 5 Summary of clinical recommendations for retinal vein occlusion.

Intervention
Effects CRVO Anti-VEGF All anti-VEGF agents studied better than sham. 7,14,15 Real world results not as good as clinical trials. 24,28 Better than Dex-implant in real-world setting. 21,22 Most studies found no difference in outcomes when comparing different anti-VEGF agents 23,24 except the registry study FRB! which reported better visual and anatomical outcomes compared with ranibizumab 29 Intravitreal bevacizumab Non-inferior to ranibizumab when given monthly. 1,2 Non-inferior to aflibercept when each given monthly 1,18 but when given monthly for three injections followed by PRN it was not non-inferior 19 Intravitreal ranibizumab Is an effective treatment for MO from CRVO with both 0.3 and 0.5 mg 14 and is safer and more effective than 6 monthly Dex-implant 16 Intravitreal aflibercept Efficacious when given monthly less effective when switched to PRN with three monthly assessments 14 however was effective when given PRN with monthly reviews. 19 Non-inferior to ranibizumab 7,15 more effective than ranibizumab in real-world setting 40 Dex-implant Better than sham. 13 Treatment effect approximately 3-4 months. 16,22 More ocular adverse events than anti-VEGF 16,22 but with fewer injections required 22 IVTA Better than sham. 1 mg is as effective as 4 mg but is associated with lower risk of ocular adverse events, thus 1 mg dose recommended 12 Macular laser Does not improve visual outcomes compared to observation 11 Treatment timing Early treatment (shorter duration of MO) associated with better visual outcomes 7,12,15 When switching to PRN dosing, monthly monitoring prevents decline in visual and anatomical gains 15,41 BRVO Anti-VEGF No difference between anti-VEGF agents when given 4-weekly. 23,24 In a RWOS, aflibercept had greater effect on anatomy and time to resolution of MO but not vision 39 Real world outcomes not as good as clinical trial outcomes 24,28,42 Intravitreal bevacizumab As effective as ranibizumab 38 Intravitreal ranibizumab Better with or without laser, than laser alone 34 Intravitreal aflibercept Better than laser 33 Dex-implant Improvement in visual and anatomical outcome best at 6 weeks to 3 months post injection 20 Less effective than anti-VEGF and with higher incidence of ocular adverse events 43 IVTA No better than macular laser but with higher rates of adverse events 12

Macular laser
Better than observation 43 Treatment timing Early treatment (shorted duration of MO) associated with better visual outcomes 20,33 Abbreviations performed a multicentre retrospective case series of 102 eyes treated with bevacizumab for MO due to CRVO and reported a mean gain of 0.22 LogMAR letters (approximately nine letters) at 5 years. Eyes with better baseline BCVA and shorter duration of symptoms were more likely to achieve better BCVA. 26 The LUMINOUS F I G U R E 2 Meta-analysis 1-Visual outcomes for central retinal vein occlusion (CRVO) randomised controlled trials (RCT). study 25 reported visual improvement in CRVO eyes treated with ranibizumab at 1 year in routine clinical practice (+10.8 letters; SD19.7). Anti-VEGF real world data in eyes with MO due to CRVO have not demonstrated unexpected non-ocular safety outcomes. The UK EMR database users' group 22 retrospectively reviewed 4619 files of patients with MO secondary to CRVO and found visual improvements were greater in eyes treated with anti-VEGF than those treated with Deximplant, although with a larger treatment burden. The endophthalmitis rate was also higher in the Dex-implant treated eyes (0.01% anti-VEGF vs. 0.2% Dex-implant). The rate of endophthalmitis following Dex-implant in another prospective observational study was similar at 0.25% (2/790 injections), 27 whereas a German Dex-implant study reported no cases of endophthalmitis among 573 patients who were followed for 6 months. 20 Both these prospective, observational studies treating eyes with Dex-implants for MO found better visual outcomes in treatment naive eyes with a shorter duration of MO.
3.2.6 | Results of meta-analysis of visual outcomes of real-world studies of eyes with MO due to CRVO A total of 3 RWOS reported change in BCVA at 6 months encompassing 479 eyes. There was a mean improvement of 12.14 letters (95%CI 7.34-16.94, p < 0.01) (4.1 in Figure 4) in the anti-VEGF arms, and 14.70 letters (95% CI 12.43-16.97, p < 0.01) in the Dex-implant arms. The forest plot resulted in considerable heterogeneity between the included studies (I 2 = 86.3%, p < 0.01), a reflection of the different study designs and intravitreal therapy used ( Figure 3).
Six RWOS reported 12 months of follow-up in eyes with CRVO (4.2 in Figure 4). There was a mean improvement in BCVA of 11.35 letters (95% CI 8.86-13.84, p < 0.01) from baseline in the anti-VEGF treatment arms. Two studies reported visual outcomes at 24 months (4.3 in Figure 4). Eyes with CRVO treated with ant-VEGF reported a mean gain of vision of 9.29 letters (95%CI À1.00 to 19.58, p = 0.08). However, visual gains declined at 36 months with 2 studies reporting a mean of 3.79 letters (95% CI À2.34 to 9.92, p = 0.02). Eyes receiving anti-VEGF therapy did better on average than those receiving Dex-implant with a SMD 0.37 (95% CI 0.16-0.59, p < 0.01) at 36 months (4.4 in Figure 4).

| Summary of evidence for BRVO
There were 187 RCTs identified which examined treatment of MO due to BRVO after duplicates were removed.

| Macular laser
Macular laser photocoagulation was the first proven treatment to improve vision in eyes with MO due to BRVO in the BVOS. 9 The results of the BVOS 9 led to macular laser becoming the standard of care for eyes with visual impairment due to MO secondary to BRVO. The BVOS 9 was conducted in the mid-1980s, and patients waited 3 months before enrolment to allow for spontaneous resolution of MO before receiving laser. Eyes without macular ischaemia and vision between 20/40 and 20/200, had a 65% chance of gaining two lines of visual acuity when treated with macular laser versus 37% of control eyes. 9

| Intravitreal corticosteroids
After the BVOS, the SCORE (Standard Care vs. cOrticosteroids for REtina vein occlusion) study compared macular laser to intravitreal triamcinolone acetonide. 30 Compared to macular laser, intravitreal triamcinolone acetonide did not demonstrate a greater visual benefit but was linked to increased incidence of ocular adverse events. 30 Later, the effect of another corticosteroid formulated for extended-delivery, Dex-implant, was assessed in the GENEVA study, a randomised, sham-controlled trial which included eyes with BRVO and CRVO. BRVO eyes randomised to Dex-implant 0.7 mg were more likely to improve vision than those randomised to sham, however, Dex-implant treated eyes were more likely to develop ocular adverse events such as cataract and elevated IOP. The treatment effect appeared to be close to 3-4 months, however, eyes were only eligible to receive a single Dex-implant at baseline with the primary outcome being at 6 months. The China Ozurdex study also compared a single injection of Dex-implant 0.7 mg to sham at 6 months with similar outcomes. 17 The COMRADE B 37 study compared ranibizumab 0.5 mg monthly for 3 months followed by ranibizumab PRN to a single Dex-implant 0.7 mg in eyes with MO due to BRVO. At 6 months, ranibizumab treated eyes had greater mean BCVA gains than Dex-implant treated eyes. This difference was more evident at 1 year in the F I G U R E 3 Meta-analysis 2-Comparison of visual outcomes within studies. COMRADE extension study 40 (even with a second Deximplant allowed in the second 6 months). Ranibizumab also had a better safety profile. Bandello et al. 37 compared the same agents although retreatment with Dex-implant was at 5 months. Although the study was underpowered, the authors concluded that Dex-implant was not 'non-inferior' to ranibizumab. Dex-implant treated eyes also had an increased risk of IOP elevation and cataract progression compared to ranibizumab treated eyes.

| Intravitreal VEGF-inhibitors
The first large RCT using anti-VEGF in eyes with MO due to BRVO was the BRAVO 40 study (Branch RetinAl Vein Occlusion Study). The study compared two doses of ranibizumab (0.3 and 0.5 mg) to sham, in eyes with MO secondary to BRVO. Eyes treated with ranibizumab had significantly better visual and anatomical outcomes at 6 months than eyes randomised to receive sham. 14 The BRIGHTER study found that ranibizumab with or without macular laser was superior to laser alone. 33,34 The VIBRANT study compared intravitreal aflibercept to macular laser in eyes with MO due to BRVO at 24 and 52 weeks. Monthly aflibercept led to significant greater visual and anatomic outcomes than macula laser. In the study, intravitreal aflibercept was given 4 weekly for 6 months, then 8-weekly. Visual and anatomic improvements seen at 6 months were maintained to 1 year. 14,31,32 The BRAVO, 40 HORIZON 45 and BLOSSOM 35 studies compared injections of aflibercept to sham in eyes with MO due to BRVO. As with CRVO eyes, better visual and anatomic outcomes were seen in eyes treated with aflibercept compared to sham and the treatment effect was greater when treatment was started early. 33,34 Although there are few data on long term outcomes of eyes with RVO receiving anti-VEGF injections, 46 VA appears to be better maintained compared to eyes receiving anti-VEGF for neovascular age related macular degeneration.

| Results of meta-analysis of visual outcomes of RCTs in eyes with MO due to BRVO
A total of eight RCTs of BRVO reported change in BCVA at 6 months encompassing 2623 eyes. There was a mean improvement in vision of 14.54 letters (95%CI À13.2 to 15.88, p < 0.01) (5.1 in Figure 5) in the anti-VEGF arms, 6.29 letters (95%CI 5.51-7.08, p < 0.01) in the Dex-implant arms, and 4.65 letters (95%CI 3.99-5.31, p < 0.01) in the IVTA arms. All eight studies favoured intravitreal treatment compared to laser or sham (3.3 in Figure 3). The pooled SMD of six studies comparing treatments was 0.5 letters (95% CI 0.21-0.78). The forest plot resulted in considerable heterogeneity between the included studies (I 2 = 87.5%, p < 0.01).
Intravitreal therapy was significantly better than laser or sham at 12 months with a SMD of 0.51 letters (95% CI 0.23-0.79, p < 0.01) (3.4 in Figure 3). This effect was lost at month 24 with a SMD of À0.23 (95% CI À1.35 to 0.90, p = 0.69) (3.5 in Figure 3). (Table 4) The visual acuity gains seen in RCTs have not been replicated in RWOS. 24,42 However, they have still confirmed the primary role of anti-VEGF for the treatment of MO associated with BRVO. From the six included studies, two examined outcomes using Dex-implant, 20,47 one compared anti-VEGF to Dex-implant, 21 one compared ranibizumab to aflibercept, 23 one ranibizumab to bevacizumab, 38 and one examined outcomes using any anti VEGF agent. 24 As seen in RCTs, early treatment was associated with better visual gains. 20 The two non-comparative RWOS 20,48 which assessed outcomes of eyes treated with Dex-implant reported that Dex-implant was an effective treatment of MO due to BRVO. Best visual and anatomical outcomes were seen 6 weeks post treatment. 20 However, treatment was associated with an increased risk of cataract and elevated IOP.

| Real-world evidence
All anti-VEGF agents (bevacizumab, ranibizumab and aflibercept) were found to be effective treatments of MO associated with BRVO in RWOS. 24,38,39 Anti-VEGF was found to be more effective than Dex-implant in improving vision and reducing MO. 21 Ang et al. 42 performed a meta-analysis of real-world evidence on outcomes of treatment of MO due to BRVO. The review included 2530 eyes followed for 12 months. Although visual and anatomical gains were achieved in eyes treated with anti-VEGF in RWOS, they were not as impressive as seen in the seminal RCTs. This is likely due to multiple factors including poorer treatment compliance, fewer injections in RWOS and differences in patient baseline characteristics.
3.3.6 | Results of a meta-analysis of visual outcomes of real-world studies of BRVO A total of three RWOS reported change in VA at 6 months encompassing 692 eyes. There was a mean improvement F I G U R E 5 Meta-analysis 4-Visual outcomes for branch retinal vein occlusion (BRVO) randomised controlled trials (RCT).
in BCVA of 13.00 letters (95%CI 11.31-14.69, p < 0.01) in the one study reporting outcomes for eyes treated with anti-VEGF and 5.82 letters (95% CI 5.03-6.61, p < 0.01) in the two studies reporting outcomes for eyes treated with Dex-implant (6.1 in Figure 6).
Three RWOS including eyes with BRVO reported VA change at 12 months, encompassing 9322 eyes. There was a mean improvement of 10.57 letters (95% CI 9.06-12.07, p < 0.01) in BCVA from baseline for eyes receiving anti-VEGF therapy (6.2 in Figure 6).

| Clinical recommendations for CRVO
Intravitreal treatment with either anti-VEGF agents or corticosteroids in eyes with MO due to CRVO is more effective than the natural history. Our results support previous meta-analyses despite having different inclusion criteria. 8,49 Anti-VEGF agents are considered first line treatment due to fewer ocular adverse events than intravitreal corticosteroid and better visual outcomes when corticosteroids are given 6-monthly. Prompt treatment is associated with better visual outcomes. 7,12,13,15 The choice of VEGF inhibitor does not appear to be important with similar visual and anatomic outcomes, as long as it is given as often as required, and patients kept under close observation when treatment is given PRN. 7,15 Ocular adverse events appear comparable between the different anti-VEGF agents.
Intravitreal corticosteroids have been used for treatment of MO 50 for decades. The risks of cataract and raised IOP 12,13,16,17,36,51 mean that they should be cautiously used, especially in the setting of younger phakic patients. For this reason, they are considered second line treatments for MO due to CRVO. There is a need for more long-term studies using corticosteroid with more frequent treatment than 6-monthly to properly assess their maximal effectiveness when ocular adverse events are managed carefully. Macular laser is not recommended as a treatment for eyes with MO due to CRVO. 11 Data from RWOS remind clinicians of the risk of eyes developing neovascular glaucoma, and retinal neovascularisation, which is more common in eyes that receive fewer injections. 29 Although one RWOS suggested that aflibercept was associated with better visual and anatomical outcomes at 12 months than ranibizumab or F I G U R E 6 Meta-analysis 5-Visual outcomes for branch retinal vein occlusion (BRVO) real-world studies (RWOS). bevacizumab, a large USA RWOS did not find any difference between any of the anti-VEGF agents. 24 Overall, the visual outcomes in RWOS are not as good as those of RCTs. This is likely to be due to undertreatment but may be due to a difference in inclusion and exclusion criteria. RWOS tend to include all patients receiving treatment for the condition, whereas RCTs exclude many patients due to concurrent eye or systemic disease, (which may limit visual potential or be associated with poorer compliance).

| Clinical recommendations for BRVO
As in eyes with CRVO, anti-VEGF agents are considered first line treatment for eyes with MO due to BRVO, supporting previous meta-analyses despite different inclusion criteria. 8,46,49 Again, prompt treatment is paramount in order to gain maximum visual outcomes. 20,32,35 The anti-VEGF agent used does not appear to influence outcomes. Ideally intravitreal injections should be given monthly until stability is reached, followed by close monitoring if a PRN regime is used.
Intravitreal triamcinolone acetonide was no more effective than laser in eyes with MO due to BRVO but was associated with a greater risk of ocular adverse events 9 and is therefore not recommended. Although intravitreal dexamethasone implant was more effective than observation, it is generally reserved as a second line treatment due to higher rates of ocular adverse events that anti-VEGF therapy (progression of cataract and raised IOP) [20][21][22]27,37,44,47 and inferior vision acuity outcomes when given 6 monthly.
Laser treatment for MO secondary to BRVO has been proven to be more effective than observation, however it is less effective than anti-VEGF and is therefore also reserved as a second-line or adjunctive treatment. Although not a focus of this review, laser pan retinal photocoagulation remains the standard of care for the treatment of neovascular complications associated with RVO (CVOS 11 BVOS 9 ). Of note, a significant number of eyes receiving anti-VEGF still developed neovascular complications. 21,23,29

| What is next?
A current phase III, randomised, clinical trial is investigating faricimab, a bispecific anti-VEGF, ang-2 inhibitor for treatment of MO due to CRVO (COMINO, ClinicalTrials.gov Identifier NCT04740931, estimated to be completed in late 2023). A similar study is also investigating faricimab for MO secondary to BRVO (BALATON; ClinicalTrials.gov Identifier NCT04740905). The RAP-TOR (ClinicalTrials.gov Identifier NCT03802630) and RAVEN (ClinicalTrials.gov Identifier NCT03810313) studies investigated brolucizumab for the treatment of MO due to RVO, but these trials were ceased prematurely due to the concern of increased risk of retinal vasculitis. RIPPLE-1 is a phase II, single-masked dose-ranging study evaluating IBE-815 (a slow-release low dose dexamethasone implant; ClinicalTrials.gov Identifier NCT04576689) for MO due to diabetes or RVO.

| Summary
Anti-VEGF agents should be considered as first line therapy for eyes with MO due to CRVO or BRVO. Visual loss from MO due to CRVO or BRVO is more likely to recover when intravitreal treatment is started early, regardless of which anti-VEGF agent is used. Monthly injections of anti-VEGF agents are required until stability of VA and MO is achieved, after which treatment can be given PRN with assessments 1-2 monthly. Although an alternative is to use a treat and extend approach, there are few data on its outcomes. Second line treatment with intravitreal Dex-implant in eyes with CRVO and BRVO still offers potential for visual improvement with fewer treatment visits, although non treatment visits for evaluation of cataract and IOP are still required. Triamcinolone acetonide 1 mg can be considered for eyes with MO due to CRVO, but not BRVO, as in BRVO eyes, it was no more effective than laser but associated with a greater risk of ocular adverse events. Macular laser is more effective than observation in eyes with MO due to BRVO, but it is less effective than anti-VEGF and thus is used as a second line or adjunctive therapy. Macular laser is no more likely to improve vision in eyes with CRVO than its natural history. A significant number of eyes with CRVO still require laser panretinal photocoagulation to manage neovascular complications.
Clinical trial outcomes of treatment of CRVO and BRVO are superior compared to RWOS. This is likely to represent undertreatment but also several other factors, including a difference in eyes and patients included in RWOS compared to RCTs, and more frequent treatment and better patient compliance seen in patients enrolled in RCTs.

ACKNOWLEDGEMENT
Open access publishing facilitated by The University of Sydney, as part of the Wiley -The University of Sydney agreement via the Council of Australian University Librarians.

FUNDING INFORMATION
None.