Objective The aim of this review is to determine whether vitreomacular adhesion (VMA) or vitreomacular traction (VMT) has an influence on the outcomes of antivascular endothelium growth factor (anti-VEGF) treatment neovascular age-related macular degeneration (nAMD).
Methods A systematic literature search was performed in Pubmed.gov, Cochrane Library, Web of Science, China National Knowledge Infrastructure, Wanfang, SinoMed and ClinicalTrials.gov up to 30 June 2016 to identify eligible studies.
Results Nine studies and 2212 participants were finally identified. At month 6, the mean improvement in best-corrected visual acuity (BCVA) and mean decline in central retinal thickness (CRT) of the VMA/VMT(+) group was less than that of the VMA/VMT(-) group (95% CI −3.05 to –0.96 letters, p=0.0002; 15.53 to 32.98 μm, p<0.00001; respectively); at month 12, there was a small or only marginally significant difference (−0.01 to 2.00 letters, p=0.05; 0.17 to 23.7 μm, p=0.05; respectively) between the groups. During the 12 months, however, the VMA/VMT(+) group required more injections ((0.25 to 0.95), p=0.0008).
Conclusions In using anti-VEGF drugs to treat nAMD, clinicians should take into account the fact that concurrent VMA or VMT might antagonise the efficacy of anti-VEGF drugs during the early stage of treatment.
Statistics from Altmetric.com
Age-related macular degeneration (AMD) is a leading cause of vision loss worldwide. One late stages at which most visual loss occurs is known as neovascular (‘wet’) AMD (nAMD), which is characterised by choroidal neovascularisation that breaks through to the neural retina, leaking fluid, lipids and blood and leading to fibrous scarring.1 Large randomised controlled trials have shown that antivascular endothelial growth factor (anti-VEGF) treatments can be effective for most nAMD; as a result, anti-VEGF drugs given via intravitreal injections have become firmly established as the standard of care for nAMD.1–5 Despite the relatively good efficacy of anti-VEGF drugs in many patients, there are still some who show a weak response or even no response to the treatment. Many factors have been shown to be predictive of the response to anti-VEGF treatment for nAMD.6 These include genetic factors (eg, single-nucleotide polymorphisms in the complement factor H and VEGF-A genes),6–8 anatomical factors (eg, lesion size, subretinal fluid)6 9 and other variables (eg, age, number of injections, visual acuity).10–12 Other possible predictive factors are now being researched, including the role of the vitreomacular interface (VMI), which has generated much interest.13
Initially the vitreous is completely attached to all contiguous internal limiting membrane of the retina by the macromolecular attachment complex—composed of fibronectin, laminin and other extracellular components that form like matrix at VMI1 14 15 and by chondroitin sulfate throughout the vitreoretinal interface.1 16–18 In the course of ageing process, degenerative processes occur (eg, gel liquefaction, weakening of vitreoretinal adhesion and gel contraction), initiating vitreoretinal separation from the peripheral fundus. This process gradually progresses to the macula and optic nerve, finally resulting in complete posterior vitreous detachment (PVD).1 19 Under anomalous conditions, gel contraction outpaces detachment of the vitreous cortex, and there is an abnormal adhesion of the vitreous cortex to the internal limiting membrane. Thus, as vitreoretinal separation progresses, anomalous macular conditions including vitreomacular adhesion (VMA) and vitreomacular traction (VMT) can ensue.1 20 21 VMA is defined as a stage wherein partial detachment of the vitreous in the perifoveal area has occurred without associated retinal abnormalities. VMA progresses to VMT when detectable retinal anatomic changes occur.1 Some investigators do not separate the two stages and include them both under the heading VMA.22–24
The association between incomplete PVD and nAMD was first suggested by Weber-Krause et al in 1966.25 After that, many reports verified a higher prevalence of VMA in eyes with nAMD compared with eyes with non-neovascular AMD or with the eyes of normal controls.22 26–33 Some observations showed that VMA occurs at the vitreoretinal interface overlying the area of choroidal neovascularisation.26–28 In addition, the association of VMI and nAMD was indirectly proved by studies showing that vitrectomy to detach the posterior vitreous cortex and relieve VMA and VMT could improve choroidal neovascular regression.34 35 These observations indicate that anomalous VMI may play a part in the initial pathogenesis or progression of nAMD. With the establishment of anti-VEGF drugs as a treatment for nAMD, many studies further investigated the influence of VMA or VMT on anti-VEGF treatment for nAMD. The results were controversial, however, with some suggesting that eyes with VMA or VMT showed less anatomic or functional improvement13 19 24 36 37 or required more frequent dosing9 19 24 37 38 than eyes without VMA or VMT. Yet others suggested that there was no such difference.23
To our knowledge, there has been no meta-analysis of this disputed issue of whether VMA or VMT has an impact on anti-VEGF treatment for nAMD. In the current study, we assessed the possible influence of VMA or VMT from the standpoints of functional outcome, anatomic outcome and management burden.
Two of the authors independently conducted a systemic search of Pubmed.gov, Cochrane Library, Web of Science, China National Knowledge Infrastructure, Wanfang, SinoMed and ClinicalTrials.gov up to 30 June 2016. The following search strategy was used:
(Vitreomacular Interface [Title/Abstract] or VMI [Title/Abstract] or Vitreomacular adhesion [Title/Abstract] or VMA [Title/Abstract] or vitreomacular traction [Title/Abstract] or VMT [Title/Abstract])
(Vascular Endothelial Growth Factors [Title/Abstract] or VEGFs [Title/Abstract] or Bevacizumab [Title/Abstract] or Avastin [Title/Abstract] or Ranibizumab [Title/Abstract] or RhuFab V2 [Title/Abstract] or Lucentis [Title/Abstract] or Aflibercept [Title/Abstract] or AVE 005 [Title/Abstract] or AVE-005 [Title/Abstract] or Zaltrap [Title/Abstract] or ZIV-aflibercept [Title/Abstract] or AVE 0005 [Title/Abstract] or AVE-0005 [Title/Abstract] or Eylea [Title/Abstract] or KH902 [Title/Abstract] or conbercept [Title/Abstract] or Lumitin [Title/Abstract] or pegaptanib [Title/Abstract] or Macugen [Title/Abstract])
(randomised controlled trial [Publication Type] or controlled clinical trial [Publication Type] or randomised [Title/Abstract] or controlled [Title/Abstract] or trial [Title/Abstract] or random [Title/Abstract] or placebo [Title/Abstract] or groups [Title/Abstract])
(1, 2 and 3)
Retrieved studies were imported into EndNote (V.X7 for Windows), where duplicate articles were deleted. Titles and abstracts of the remaining articles were independently scanned by two authors. Extracted studies were compared and inconsistencies were resolved by consensus. The full texts of the remaining studies were then read to determine whether they met our inclusion criteria. In addition, the references of all identified studies were reviewed.
Studies included in this meta-analysis had to meet the following criteria: (1) The study design was a prospective cohort or retrospective case–control study. (2) All patients were diagnosed of nAMD based on leakage on fluorescein angiography and optical coherence tomography (OCT). (3) All patients received anti-VEGF treatment for nAMD in the study eye. (4) Participants had concurrent VMA or VMT, whereas controls did not have VMA or VMT. (5) Patients were followed for a minimum of 6 months. (6) Direct or indirect records of at least one of the three following outcomes were available: mean best-corrected visual acuity (BCVA) change from baseline 6 and 12 months after the first anti-VEGF injection; mean central retinal thickness (CRT) change from baseline 6 and 12 months after the first anti-VEGF injection; mean number of injections during the 12-month follow-up period. BCVA was determined using the ETDRS chart at 4 m and logarithm of the minimal angle of resolution (logMAR) was converted to letters for analysis. CRT was defined as the mean value of retinal thickness of the central points, measured manually on OCT scans using the built-in software in the OCT device.
Exclusion criteria were as follows: individuals with coexisting pathologies or choroidal neovascularisation secondary to other pathologies were excluded, as were studies published in languages other than English or Chinese. When multiple publications from the same study population were available, we checked for duplicate analyses and included the most recent publication.
Quality assessments and data extraction
The retrieved articles were reviewed for risk bias using the Newcastle-Ottawa Scale.39 From each study, the following characteristics were extracted: last name of first author and publication year, enrolment period, study design, age, gender, sample size, treatment regimen and follow-up period. Primary outcome measures were as follows: functionally mean BCVA change 6 and 12 months after the first injection and anatomically mean CRT change 6 and 12 months after the first injection. The secondary outcome measure was the mean number of injections during the 12-month follow-up period.
The quantitative data were entered into Review Manager (RevMan, V.5.3, Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2014). Meta-analysis was performed on the two primary outcome measures and one secondary outcome measure. Summary estimates, including 95% CIs, were calculated. The means and SD were used to calculate the estimated mean difference (MD) between groups for all the continuous outcome data. A random or fixed-effects model was use for analysis according to the situation. Statistical heterogeneity was tested using the χ2 test and I2 statistic. A sensitivity analysis that investigates the contribution of each study to the heterogeneity was performed by sequentially omitting one study and reanalysing the pooled estimate for the remaining studies. A two-sided p value less than 0.05 was regarded as significant for all analyses.
The process of the systemic search is demonstrated in figure 1. The search yielded 452 articles; 109 duplications were removed. Review of the title and abstract removed 308 of the articles, and thorough full-text assessment removed 26. Nine studies and 2212 participants were finally identified and analysed for the meta-analysis.9 13 19 22–24 36 38 40 Characteristics and quality assessment results of the included studies are presented in table 1. Quantitative synthesis and analysis of BCVA and CRT change at month 6 was conducted on available data from three studies and at month 12 from seven studies. There was no difference between the baseline BCVA or CRT of the VMA/VMT(+) and VMA/VMT(-) groups in any one study. Quantitative synthesis and analysis of injection frequency during 12 months of follow-up was conducted on data from five studies.
The mean BCVA improvement of the VMA/VMT(+) group at month 6 was less than that of the VMA/VMT(-) group (mean difference (95% CI), −2.01 (−3.05 to –0.96) letters; p=0.0002); at month 12, the difference became marginally significant (1.00 (-0.01 to 2.00) letters; p=0.05) (figure 2).
The mean CRT decline of the VMA/VMT(+) group at month 6 was smaller than that of the VMA/VMT(-) group (24.26 (15.53 to 32.98) μm; p<0.00001); at month 12, there was no more significant difference (11.94 (0.17 to 23.71) μm; p=0.05) (figure 3).
During the 12-month follow-up period, the mean number of injections among patients in the VMA/VMT(+) group was greater than that among those in the VMA/VMT(-) group (0.60 (0.25 to 0.95); p=0.0008). (figure 4).
Funnel plots of each comparison are shown in figure 5. Data from the Krishnan40 study were not synthesised in the analyses because, according to the sensitivity analysis, it would have increased heterogeneity and decreased robustness. After ruling out the Krishnan study, heterogeneity declined greatly, and in the analysis of BCVA change at month 12, I2 declined from 72% to 37%, CRT changed from 82% to 80% and injection frequency from 72% to 59%. CRT decrease at month 12 changed from 95% CI −4.26 to 20.08 μm to 0.17 to 23.71 μm. Finally, sensitivity analysis demonstrated that the current meta-analysis were robust.
According to the results of our analysis, the concurrent VMA or VMT affected the efficacy of anti-VEGF drugs for patients with nAMD during the first 6 months of treatment. Patients with VMA or VMT achieved less BCVA improvement and less CRT decline. After 12 months of treatment, this difference gradually diminished. Patients with VMA or VMT required more injections during the −12 month follow-up period.
With regard to the role VMI plays in the pathophysiological process of nAMD, there are three possible mechanisms by which anomalous VMI may antagonise the effect of anti-VEGF drugs. First, anomalous adhesion causes a traction force at VMI. This force is directly applied to Müller cells and causes stretching of the retinal pigment epithelial cells. This effect could lead to the secretion of various proinflammatory cytokines, thus inducing local inflammation, including VEGF, which plays an important role in the pathogenesis and progression of nAMD.13 41–44 Second, the traction force on the retina causes tissue to be pulled apart,42 45 a structural change that induces macular oedema. As a result, oxygen diffusion and nutrient supply to retinal tissue are influenced, and the blood-retinal barrier breaks down. Consequently, the lesion of nAMD becomes more active.13 41 42 46 47 Lastly, the traction force also causes local ischaemia and disrupts the choroidal supply of the macula, so that cytokines, including VEGF, are confined within the retinal structure.48 49 The fixed amount of anti-VEGF drug is not sufficient to neutralise the increased level of VEGF, which compromises the treatment outcome. As a result, more injections of anti-VEGF drugs are required.13 50
There is other indirect evidences pointing to the influence of VMA or VMT on anti-VEGF treatment for nAMD and the possible mechanisms involved. In humans and animals, pharmacological vitreolysis and vitrectomy have been shown to increase oxygen diffusion.51–53 Pharmacological vitreolysis, vitrectomy and PVD result in a more liquefied vitreous with less viscosity; therefore the diffusion of cytokines, oxygen, growth factors and medications may increase. This is just the opposite of VMA or VMT, which acts functionally as a diffusion barrier.9 Pharmacological vitreolysis agents or surgical vitrectomy are seen as an adjuvant treatment option for VMT and as a way of dealing resistance to conventional anti-VEGF treatments in selected patients.27 36 54 55
During treatment, VMA advanced to VMT in some patients or even worsened and progressed to PVD.1 Actually anti-VEGF agents can exacerbate VMA and VMT by causing contraction of the vitreous and posterior hyaloid membrane. This will further worsen a patient’s prognosis.56 In this study, patients with such unstable progressive VMA or VMT were not removed from the VMA/VMT(+) group. Besides, in this study patients with VMA and VMT were also not divided into two groups. It might seem more rigorous to further divide VMA/VMT(+) group into three subgroups, namely persistent VMA, persistent VMT and dynamic VMA/VMT.19 38 However, we consider such a step might be unnecessary and impractical. The grouping method of this meta-analysis is more in accord with actual clinical situations, because the evaluation of prognosis and formulation of the therapeutic plan are largely conducted before treatment, when dynamic changes in the patients’ VMI condition are still unknown. In addition, in distinguishing VMT from VMA by OCT, the results of different readers are not always the same, since they are highly dependent on the reader’s personal understanding of the definitions and characteristics.
Our meta-analysis showed that the antagonistic effects of VMA or VMT on the anatomic and functional outcome of anti-VEGF treatment of patients with nAMD diminished as the treatment continued. The possible mechanisms behind the disappearance of the effects were, as already mentioned, VMA/VMT release, which led to PVD in some patients during the 12 months of follow-up,1 either owing to natural disease progression or exacerbated by anti-VEGF agents.56 The effect of VMA/VMT vanishes along with the disappearance of VMA/VMT, as expected. The difference in the treatment outcomes between the two groups gradually grew smaller. If patients were monitored beyond 12 months, the difference in the mean number of injections would either persist or vanish, as functional and anatomic differences did. According to the two studies that had a follow-up duration of 2 years, the differences persisted.9 More studies with longer follow-up times are needed to draw a reliable conclusion regarding this issue.
This study has clinical significance. First and foremost, clinicians can get more information about the probable response of a given patient to anti-VEGF treatment by whether the patient has concurrent VMA or VMT. The presence of VMA or VMT may serve as a potential indicator of slower response and the need for more frequent injections. When clinicians encounter patients with VMA or VMT, adequate counselling before commencing treatment is recommended, so that patients will be fully informed as to what to expect.40 Second, based on the findings of our study, clinicians are advised to be cautious in attempting to extend intervals between visits or treatments in patients with VMA or VMT.13 Moreover, our study provides some justification for recent clinical investigations recommending earlier vitrectomy or intravitreal injection of the proteolytic enzyme ocriplasmin, used alone or in combination with anti-VEGF agents, to release VMA and VMT.27 40 54 57
This study has some limitations. First, the VMA/VMT(-) group consisted of patients with complete posterior vitreous attachment (PVA) or PVD. Although the OCT-based anatomic classification of VMI diseases has been a gold standard,38 58 it is still difficult to differentiate when the posterior vitreous boundary could not be seen in the entire scan set.19 However, the difference induced by PVA may be negligible because PVA is rare in elderly patients.19
Another major limitation is the great heterogeneity in the analysis of CRT change at month 12. This heterogeneity emerged from many possible factors, such as the variety of demographic characteristics, different types of drugs, different treatment regimens and so on. Because there are no two identical clinical studies, heterogeneity is a common feature of meta-analysis. The topic of this meta-analysis is a disputed issue, with previous studies holding different conclusions. It is for this reason that the heterogeneity of this meta-analysis is naturally high, and that it is especially meaningful to conduct this evidence-based research. We adopted the following methods to ensure the reliability and robustness of our analyses: (1) We use the random-effect model instead of the fixed-effect model.59 60 (2) Sensitivity analysis was conducted by excluding data from one study at a time and recalculating the results of the meta-analysis, whereby the current study was proved to be robust.61 (3) We also performed subgroup analyses based on different conditions,62 including patients’ race, age, drug type and treatment regimen. The results are available in the online supplementary files.
In conclusion, during the early stage of treatment, concurrent VMA or VMT may antagonise the efficacy of anti-VEGF drugs in patients with nAMD. Patients with VMA or VMT achieved less improvement in BCVA and less CRT decline, but these differences gradually diminished over time. We also found that patients with VMA or VMT required more injections. These findings should be taken into consideration when clinicians evaluate prognosis or make decisions regarding individualised therapeutic regimens for patients with nAMD.
Contributors PX and XZ contributed equally and are coauthors of this paper, who composed the manuscript. QL is the corresponding author, having designed the study and revised the manuscript. YY and XY participated in data extraction. ZH and DY carried out the statistical analysis. All authors read and approved the final manuscript.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement Any interested reader may make an inquiry of additional unpublished data to corresponding authors via email.
Correction notice This paper has been amended since it was published Online First. Owing to a scripting error, some of the publisher names in the references were replaced with 'BMJ Publishing Group'. This only affected the full text version, not the PDF. We have since corrected these errors and the correct publishers have been inserted into the references.
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.