Objectives The aim of this study is to systematically review the evidence for anti-vascular endothelial growth factor (VEGF) therapy in choroidal neovascularisation secondary to conditions other than age-related macular degeneration.
Data sources MEDLINE, MEDLINE in-process, EMBASE and CENTRAL databases and conference abstracts were searched (from inception to Jan 2014).
Study eligibility criteria, participants and interventions Randomised and non-randomised comparative studies with follow-up of at least 6 months were included and were used to assess clinical effectiveness.
Study appraisal and synthesis method Risk of bias was assessed using the Cochrane risk of bias tool and modified Newcastle-Ottawa Scale. Meta-analysis was not possible due to methodological heterogeneity.
Results 16 studies met the inclusion criteria (1091 eyes; 963 pathological myopia, 74 other conditions). There was large variation in risk of bias across studies. An improvement in best-corrected visual acuity in anti-VEGF arms over comparators was reported in all studies. The proportion of patients improving by at least 15 letters in anti-VEGF arms ranged from 27.3% to 70%. There were no significant differences between bevacizumab and ranibizumab.
Limitations Owing to the rarity of choroidal neovascularisation secondary to conditions other than age-related macular degeneration or pathological myopia, there are unlikely to ever be sufficiently powered trials in these populations.
Conclusions Bevacizumab and ranibizumab appear to be effective in improving visual acuity for patients with choroidal neovascularisation secondary to conditions other than age-related macular degeneration. The evidence base is strongest for choroidal neovascularisation secondary to pathological myopia, however, based on current evidence and likely pharmacological pathways, clinicians should consider treatment with either bevacizumab or ranibizumab for rarer causes.
- PUBLIC HEALTH
This is an Open Access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/
Statistics from Altmetric.com
Strengths and limitations of this study
A broad search has been undertaken, and data interpreted to maximise usefulness to clinicians.
There is a lack of evidence for choroidal neovascularisation secondary to conditions other than age-related macular degeneration or pathological myopia, and there is unlikely to ever be sufficiently powered trials in these populations.
The evidence base is strongest for pathological myopia, but based on current evidence and likely pharmacological pathways, clinicians should consider treatment with either bevacizumab or ranibizumab for rarer causes of choroidal neovascularisation.
Choroidal neovascularisation (CNV) is a common and severe complication of a number of different diseases affecting the posterior segment of the eye, and has the potential to cause blindness. It has a significant impact on functioning and quality of life.1 It is characterised by neovascularisation originating from the choroid which grows through Bruch's membrane and under the retinal pigment epithelium (RPE) or retina.2 Loss of vision usually results from haemorrhage and leakage, and ultimately fibrosis.3 Vascular endothelial growth factor (VEGF) is recognised as a key signalling molecule in this process. The most common disease associated with CNV is neovascular (wet) age-related macular degeneration (ARMD).
Pathological myopia (PM) is the commonest non-ARMD condition associated with CNV. It is estimated to affect up to 3% of the population, of which 5–11% may develop myopic CNV.3–6 Other conditions associated with CNV include angioid streaks, multifocal choroiditis, punctate inner choroidopathy, pseudoxanthoma elasticum and presumed ocular histoplasmosis. CNV may be associated with trauma and can be idiopathic. These conditions tend to affect younger patients leading to lifelong impairment.3 These conditions are relatively uncommon individually, but are more frequently seen as a combination. There is only limited evidence available about their treatment.7
The use of anti-VEGF agents has emerged as an effective therapy for a number of ophthalmological conditions. They have been shown to be superior to photodynamic therapy (PDT) in ARMD in large randomised controlled trials (RCTs)8–10 and in the treatment of macular oedema following retinal vein occlusion and diabetic macular oedema.11 ,12 There are a number of trials that show the effectiveness of anti-VEGF antibodies in the treatment of CNV associated with PM.13 ,14 Case reports and case series in the literature report improvements in vision and regression of CNV secondary to conditions other than ARMD with anti-VEGF therapy,15–18 but there are few interventional studies.
The aim of this study is to systematically review the evidence for anti-VEGF therapy in CNV secondary to conditions other than ARMD.
A systematic review was undertaken. The following electronic databases were searched from inception to January 2014: MEDLINE, MEDLINE In-process, EMBASE and CENTRAL. Conference abstracts from the annual meetings of the Association for Research in Vision and Ophthalmology, The Royal College of Ophthalmologists, and the American Academy of Ophthalmology for years 2011–2013 were searched using choroidal neovascularisation terms.
The search strategy for MEDLINE is shown in the online supplementary material. This was adapted for EMBASE and CENTRAL. Terms for ARMD were included in the search strategy to prevent excluding studies in which non-ARMD subgroups were included, or comparison with ARMD was used.
Only trials with a comparative design were included. This included RCTs, controlled trials (CTs), non-randomised trials, and comparative studies. Studies including adults over the age of 18 with a diagnosis of CNV that was secondary to non-ARMD conditions were eligible for inclusion. However, studies including patients with and without ARMD with reporting of subgroups were eligible.
Included interventions were intravitreal bevacizumab, ranibizumab, pegaptanib and aflibercept. Eligible comparators were placebo/sham treatments, other pharmacological interventions, usual care and observation. There were no language restrictions. Studies with length of follow-up of less than 6 months were excluded.
Outcome measures were: (A) best-corrected visual acuity (BCVA): mean change in, proportion of patients improving, and proportion of patients worsening; (B) mean change in central macular thickness (CMT) as determined by optical coherence tomography (OCT) and (C) adverse events. All BCVA data were converted to number of letters for consistency.
Screening and data extraction
Screening of titles and abstracts were undertaken independently by two authors (AS and SD). Differences were resolved through discussion with a third author (JAF). Data was extracted in a prespecified data extraction form. Non-English articles were translated.19–21 Data extracted included baseline characteristics, mean change in BCVA, proportion of patients improving, proportion of patients worsening, mean change in CMT, and adverse events. Risk of bias for the RCTs was assessed using the Cochrane risk of bias tool.22 A modified Newcastle-Ottawa Scale was used to assess the risk of bias for non-RCT studies. It was not possible to assess publication bias using a funnel plot because of heterogeneity and a limited number of studies.
Data were assessed for suitability for meta-analysis, but this was not possible due to methodological heterogeneity.
Sixteen studies met the inclusion criteria after screening 1251 titles and abstracts (figure 1).13 ,19–21 ,23–34 The main reasons for exclusion at full text stage was the absence of a separate analysis of trial arms, ARMD as cause of CNV, absence of comparator, invalid comparator and condition not CNV.
Table 1 shows that 5 studies were RCTs and 11 were non-randomised comparative studies. Studies were from a range of different countries. Only one trial was multicentre and industry funded. Follow-up ranged from 6 to 24 months.
Across included studies, the total number of eyes was 1091 (426 in RCTs), of which 684 received an anti-VEGF. Study size ranged from 27 to 277 eyes. Mean age ranged between 35.2 and 67 years, and between 60% and 100% were female. Mean baseline BCVA was between 81 and 99 letters.
Thirteen of the studies (4 of the 5 RCTs, 1017 eyes) included participants with CNV secondary to PM. The remaining studies examined CNV associated with multifocal choroiditis, punctate inner choroidopathy, or that was idiopathic.
The treatment and comparator therapies used in the included studies were intravitreal bevacizumab (IVB), intravitreal ranibizumab (IVR), photodynamic therapy (PDT), and in one study a traditional Chinese medicine (fufang xueshuantong (FXT)). The dose used in all studies was IVR 0.5 mg or IVB 1.25 mg. All studies using PDT as comparator reported standard PDT protocol as per the verteporfin in photodynamic therapy study. The mean number of IVB/IVR injections varied from 1.5 to 4.72, and the number of PDT treatments from 1.3 to 2.5. No studies assessing pegaptanib or aflibercept were found.
One study used a herbal agent, FXT.19 FXT is a Chinese herbal formula used in ophthalmological conditions, and consists of Panax notoginseng, Salvia miltiorrhizae, Astragalus membranaceus and Scrophularia ningpoensis. It is purported to have a vasodilatory effect, and has been studied in the treatment of diabetic retinopathy.35
Risk of bias
Generally, the RCTs were of low or unclear risk of bias, except for blinding of participants that was high or unclear in four studies (table 2). This reflects the difficulty of blinding participants in these trials. The majority of studies used assessors who were blinded to the received interventions when evaluating visual acuity after treatment, but this was not discussed in one study.19 Sequence generation was not reported in two studies. Two studies used sequentially numbered envelopes,23 ,33 but it was unclear if these were opaque envelopes.
The comparative studies had low risk of bias for selecting participants from the same cohort, comparability of participants, incomplete data and selective reporting, but a high risk of bias for outcome assessment (table 3). No studies blinded assessors to the interventions received.
All studies using PDT reported using a standard regime as per the verteporfin in Photodynamic Therapy Study. After baseline treatment, all studies based re-treatment on fluoroscein angiography (FA) findings at three monthly assessments. The mean number of treatments over the duration of follow-up ranged from 1.331 to 3.0.21
All studies using anti-VEGFs reported standard doses of 0.5 mg of ranibizumab, and 1.25 mg bevacizumab intravitreally. Dosing regimens varied by study. Three studies19 ,29 ,33 used a three monthly loading regime followed by further treatment based on clinical assessment (see table 4). All other studies based re-treatment on the findings of FA and OCT at 1–3 monthly follow-up visits. Mean number of injections over the follow-up periods ranged from 1.6 25 ,31 to 4.72 injections.23
Anti-VEGF versus PDT
Randomised controlled trials
Pathological myopia In the RADIANCE trial (eyes=277), an RCT of ranibizumab for CNV secondary to pathological myopia, the results for the three separate treatment arms are presented from the 3-month end point, as the control group received ranibizumab thereafter.13 Treatment arms consisted of two IVR groups re-treated based on different criteria (on the basis of assessed disease activity (DA), and on the basis of assessed disease stabilisation (STAB) and a PDT group. Mean change in BCVA was the same in both IVR groups, at a gain of 10.6 letters. The gain in letters in the PDT group was 2.2. The proportion improving (gain of ≥15 letters) was 43.1% and 38.1% in the respective IVR arms (DA and STAB), and 14.5% in the PDT group. The proportion of patients worsening was not reported. The mean decrease in CMT was 77.5, 60.9 and 12 µm between IVR DA, IVR STAB and PDT arms, respectively (statistical significance not reported). At 12 months, all three arms reported improvements in BCVA.
Other CNV causes
Parodi et al33 compared the effectiveness of PDT and IVB in patients with subfoveal CNV secondary to multifocal choroiditis (eyes=27). They reported a mean gain of 9 letters in the IVB group compared with 1 letter in the PDT group at 12 months. The difference was statistically significant. The proportion of patients with a gain of >15 letters was 36% in the IVB group compared with 0% in the PDT group; 8% of patients in the PDT group had a loss of >15 letters compared with none in the IVB group (statistical significance not reported). The mean CMT change was 44 and 55 µm in the PDT and IVB groups, respectively (statistical significance not reported).
Pathological myopia Seven of the eight comparative studies20 ,21 ,25–27 ,30 were in PM (eyes=541). The mean change in BCVA improved for all anti-VEGF arms compared with PDT. In studies in which the gain in BCVA in anti-VEGF arms over PDT was reported as statistically significant, the gain in letters ranged from 630 to 12.5 letters.31
The proportions of patients improving by >15 letters in the anti-VEGF groups ranged from 27.3%21 to 70%;20 however, neither of these groups reported statistical testing. In those in which a statistically significant difference was found (p≤0.05),25 ,26 the gain was 41.9% and 39.7% compared with 20.4% and 17.7% in the PDT groups, respectively.
Six of the seven comparative studies20 ,21 ,25 ,27 ,30 ,31 reported the proportion of patients with worsening vision. In all studies, there was a greater proportion that deteriorated ≥15 letters in the PDT groups versus the anti-VEGF groups.
Other CNV causes
One study32 was in idiopathic CNV (eyes=29). The gain in the anti-VEGF group was 17.5 vs 14 letters in the PDT group. In total 53.5% of patients in the anti-VEGF group compared with 42.9% of patients in the PDT group had a gain of >15 letters. No patients had a loss of >15 letters in the anti-VEGF group, compared with 21.3% in the PDT group. All differences were reported as statistically significant.
Ranibizumab versus bevacizumab
Randomised controlled trials
Pathological myopia Iacono et al23 (eyes=48) reported no statistically significant difference in either mean letter gain, or proportion improving by at least 15 letters between IVR and IVB groups. Of those worsening, slightly more deteriorated in the IVB group at 24% versus 17% in the IVR; statistical significance was not reported.
Similarly, Gharbiya et al24 (eyes=32) reported no statistically significant difference in the number of letters gained, or proportion of participants gaining more than 15 letters.
Pathological myopia Yoon et al28 (eyes=40; IVB=26, IVR=14) reported no statistically significant difference between intervention groups, with a mean gain of 13.5 and 14 letters in IVR and IVB groups, respectively.
Lai et al29 (eyes=37, IVB=22, IVR=15) also did not report a statistically significant difference, with a mean gain of 14 and 25.5 letters between IVB and IVR groups, respectively.
Other CNV cause
Cornish et al34 studied treatment of punctate inner choroidopathy (eyes=18; IVB=6, IVR=12). Mean gain in BCVA was 23 letters in the IVR group and 8.5 letters in the IVB group. Sixty-seven per cent of patients in the IVR group had a gain of at least 15 letters versus 83% in the IVR group. Statistical testing was not reported.
Liu et al19 (eyes=42) compared IVB with no IVB in patients with PM taking oral FXT. In the IVB + FXT group, there was a mean improvement of 21 letters, and in the FXT group there was a statistically significant mean improvement of 10 letters.
Twelve studies reported no adverse events occurring, and one study did not present adverse event data. Generally speaking, anti-VEGF therapy, compared with PDT, had fewer significant adverse events (eg, endophthalmitis, retinal detachment, systemic events). Adverse events in the RADIANCE trial were similar between IVR and PDT.13 El Matri et al27 reported two cases of endophthalmitis (6.6%) and one vitreous haemorrhage (3.3%) in the IVB group. Only one study that compared IVR with IVB reported on adverse events (worsening of cataract, increase in myopic foveoschisis, retinal detachment, macular hole, systemic events); there were similar adverse events in both groups (table 5).29
Statement of principal findings
Evidence from RCTs and non-randomised comparative studies shows that anti-VEGF therapies show consistent benefit in non-ARMD CNV conditions. When compared with the previous ‘gold-standard’ (PDT), anti-VEGFs result in greater improvements in BCVA. There was no robust evidence to suggest superiority of ranibizumab or bevacizumab.
Strengths and limitations
The search strategy was robust and broad with no language restrictions, and included grey literature. Two reviewers screened titles and abstracts. Risk of bias in studies was assessed using the Cochrane Risk of Bias Tool for RCTs and a modified Newcastle-Ottawa Scale for comparative studies. Only studies with at least 6 months of follow-up were included to increase meaningfulness of outcomes.
A major limitation of this review was that the majority of evidence pertains to CNV caused by PM, however, this reflects the available evidence base in the literature. The included non-PM CNV conditions such as PIC, and POHS are of such rarity that it is unlikely there will ever be large RCTs of their treatment. Many of the non-randomised comparative studies were small and of low quality. There was one large industry-funded trial assessing ranibizumab, but none assessing bevacizumab.
Methodological heterogeneity between studies was too high to allow meta-analysis. Baseline BCVA varied considerably between studies, as did treatment regimes. No study reported on vision-related quality of life as an outcome measure, arguably the most important. Studies were powered for clinical efficacy, not to detect adverse events.
The largest RCT included in our review (RADIANCE)13 was limited by the fact that although the entire follow-up period was 12 months, after 3 months, patients were eligible to cross over into other arms of the study. We therefore have presented only 3-month data, as the relevance of the data after this point is questionable.
Context of these results
This is the first systematic review to include all causes of CNV except ARMD. Wang et al36 undertook a systematic review of anti-VEGFs in CNV secondary to only PM. It did not include the RADIANCE study13 or undertake as broad a search. The authors concluded that the evidence supported anti-VEGF agents as first-line treatment, which supports our findings.
Ranibizumab remains the only drug licensed for the treatment of CNV secondary to PM, and its short-term (up to 24 months) safety has been demonstrated in numerous studies.8–10 ,37 National Institute for Health and Care Excellence (NICE) has recently approved ranibizumab as an option for treatment for CNV secondary to PM, where it is provided at a discount through a patient-access scheme.38 The appraisal committee noted that while there was little long-term evidence, it had shown greater clinical effectiveness than the current standard treatment of PDT.
Bevacizumab has a similar mechanism of action, and is considerably cheaper. However, due to commercial reasons, it is unlikely ever to be licensed for intravitreal use. The CATT study demonstrated that bevacizumab and ranibizumab have equivalent effects on visual acuity in neovascular ARMD.39 A total of 1185 patients were randomised to receive either bevacizumab or ranibizumab, and at 24-month follow-up the authors found similar effects on visual acuity and no difference in rates of death or systemic arteriothrombotic events. In 2012, NICE evaluated 89 studies and concluded that there was no significant difference in adverse events between bevacizumab and ranibizumab.40 A recent systematic review of the treatments for macular oedema secondary to central retinal vein occlusion examined anti-VEGF agents, including bevacizumab and ranibizumab, and concluded that they were similar in improving visual acuity, and there was no evidence of difference in adverse events.11
Anti-VEGF agents are used off-label for the treatment of CNV secondary to conditions other than ARMD or PM. There are multiple case series that support their effectiveness. All case series are subject to several methodological weaknesses, most importantly, publication bias and lack of comparator groups. Troutbeck et al17 reported on the use of IVR in 41 patients with a range of conditions complicated by CNV, including multifocal choroiditis, peripapillary CNV, angioid streaks, central serous chorioretinopathy, macular telangiectasia and idiopathic CNV. They reported that 25–43% of patients experienced 15 letter or greater improvement in vision. Chang et al16 used bevacizumab in 39 eyes in the treatment of CNV associated with either multifocal choroiditis, angioid streaks, myopic and also idiopathic CNV. Median BCVA improved from 76 letters at baseline, to 85 letters at mean follow-up of 58.8 weeks, and there were no adverse events.
What do these results mean for clinical practice?
The evidence for the use of anti-VEGF in the treatment of CNV associated with ARMD and, recently, PM is well established. The evidence for the use of these agents in the treatment of CNV complicating other diseases is mixed. This represents a heterogeneous group of conditions, often found in younger people and frequently with devastating visual outcomes. Despite a limited evidence base, the use of anti-VEGF therapy is likely to provide the best outcomes for patients. Patients expect and demand treatment in advance of best evidence being available, and healthcare planners and commissioners need to make decisions about the use of anti-VEGF molecules in these circumstances with limited evidence base for the relatively rare cases. Marginal cost–benefit analysis is often used in these circumstances, and this is likely to be favourable if it takes account of the overall costs to society and the individual patient in the event of a devastating loss of vision. Given that anti-VEGFs are superior to PDT and its use is off-label in treatment of CNV secondary to conditions other that ARMD and PM, considering the cheapest drug (sourced and administered) would prove to be the most cost effective and affordable option for clinical commissioners.
While the use of anti-VEGFs in ARMD and, recently, PM has been investigated in a number of large robustly conducted RCTs, there is a corresponding lack of high-quality, long-term evidence for the use of these drugs in CNV of other causes.
Large RCTs with head-to-head comparison of anti-VEGFs and other standard treatments are unlikely to be conducted in CNV secondary to conditions other than ARMD or PM, because of the heterogeneous and rare nature of these conditions. It may also be unethical to randomise participants to PDT considering the evidence that currently exists, and that the scientific equipoise is more in favour of anti-VEGFs. High-quality multicentre comparative studies which compare different anti-VEGFs are needed, especially considering the cost difference. This will become more important with the advent of aflibercept which has recently been licensed for choroidal neovascularisation secondary to pathological myopia.41 Further, small case series are unlikely to change clinical practice. Further studies are needed to establish the place of each anti-VEGF in the treatment pathway, and the frequency of injection.
Bevacizumab and ranibizumab appear to be more effective in improving visual acuity in patients with CNV secondary to pathological myopia. Based on the current knowledge of the condition, small RCTs, non-randomised comparative studies and robust RCT data from other conditions, clinicians should consider bevacizumab or ranibizumab as an option for treating patients with CNV secondary to other rarer causes. There is no evidence of difference in outcomes between bevacizumab and ranibizumab.
The authors would like to thank Rosetta Yuen for her help with the translation.
Review history and Supplementary material
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online supplement
Contributors AP conceived the idea. All author contributed to the design of the study. AS and SD screened titles and abstracts and extracted data. JF supervised day-to-day activities. CJ provided clinical expertise throughout. All authors interpreted the results. AS drafted the initial manuscript and all authors were involved in revising and agreeing the final manuscript. AS is the guarantor.
Funding This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement No additional data are available.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.