Abstract
BACKGROUND Age-related macular degeneration (AMD) is a highly prevalent condition in an ever-increasing elderly population. Although insidious in the early stages, advanced AMD (neovascular and atrophic forms) can cause significant visual disability and economic burden on health systems worldwide. The most common form, geographic atrophy, has no effective treatment to date, whereas neovascular AMD can be treated with intravitreal anti-vascular endothelial growth factor (anti-VEGF) injections. Geographic atrophy has a slow disease progression and patients tend to have preserved central vision until the final stages. This tendency, coupled with the use of modern imaging modalities, provides a large window of opportunity to intervene with validated methods to assess treatment efficacy. As geographic atrophy is an increasingly common condition with no effective intervention, many treatments are under investigation, one of which is visual cycle modulators. These medications have been shown to reduce lipofuscin accumulation in pre-clinical studies that have led to several clinical trials, reviewed herein. OBJECTIVES To assess the efficacy and safety of visual cycle modulators for the prevention and treatment of geographic atrophy secondary to AMD. SEARCH METHODS We searched the Cochrane Central Register of Controlled Trials (CENTRAL) (which contains the Cochrane Eyes and Vision Trials Register) (2020, Issue 1); MEDLINE Ovid; Embase Ovid; Web of Science Core Collection; Scopus; Association for Research in Vision and Ophthalmology (ARVO) website; ClinicalTrials.gov and the WHO ICTRP to 11 January 2020 with no language restrictions. We also searched using the reference lists of reviews and existing studies and the Cited Reference Search function in Web of Science to identify further relevant studies. SELECTION CRITERIA We included randomised controlled trials (RCTs) and quasi-randomised clinical studies (if available) that compared visual cycle modulators to placebo or no treatment (observation) in people diagnosed with AMD (early, intermediate or geographic atrophy). DATA COLLECTION AND ANALYSIS Two authors independently assessed risk of bias in the included studies and extracted data. Both authors entered data into RevMan 5. We resolved discrepancies through discussion. We graded the certainty of the evidence using the GRADE approach. MAIN RESULTS We included three RCTs from the USA; one of these had clinical sites in Germany. Two studies compared emixustat to placebo while the other compared fenretinide to placebo. All assigned one study eye per participant and, combined, have a total of 821 participants with a majority white ethnicity (97.6%). All participants were diagnosed with geographic atrophy due to AMD based on validated imaging modalities. All three studies have high risk of attrition bias mainly due to ocular adverse effects of emixustat and fenretinide. We considered only one study to be adequately conducted and reported with high risk of bias in only one domain (attrition bias). We considered the other two studies to be poorly reported and to have high risk of attrition bias and reporting bias. People with geographic atrophy treated with emixustat may not experience a clinically important change in best-corrected visual acuity (BCVA) between baseline and 24 months compared to people treated with placebo (mean difference (MD) 1.9 Early Treatment Diabetic Retinopathy Study (ETDRS) letters, 95% confidence interval (CI) -2.34 to 6.14, low-certainty evidence). Emixustat may also result in little or no difference in loss of 15 ETDRS letters or more of BCVA compared with placebo at 24 months (16.4% versus 18%) (risk ratio (RR) 0.91, 95% CI 0.59 to 1.4, low-certainty evidence). In terms of disease progression, emixustat may result in little or no difference in the annual growth rate of geographic atrophy compared with placebo (mean difference MD 0.09 mm2/year (95% CI -0.26 to 0.44, low-certainty evidence). All three studies reported adverse events of both drugs (emixustat: moderate-certainty evidence; fenretinide: low-certainty evidence). The main adverse events were ocular in nature and associated with the mechanism of action of the drugs. Delayed dark adaptation (emixustat: 54.5%; fenretinide: 39.3%) and chromatopsia (emixustat: 22.6%; fenretinide: 25.2%) were the most common adverse events reported, and were the most prevalent reasons for study dropout in emixustat trials. These effects were dose-dependent and resolved after drug cessation. No specific systemic adverse events were considered related to emixustat; only pruritus and rash were considered to be due to fenretinide. One emixustat study reported six deaths, none deemed related to the drug. None of the included RCTs reported the other pre-specified outcomes, including proportion of participants losing 10 letters or more, and mean change in macular sensitivity. We planned to investigate progression to advanced AMD (geographic atrophy or neovascular AMD) in prevention studies, including participants with early or intermediate AMD, but we identified no such studies. Two of the included studies reported an additional outcome - incidence of choroidal neovascularisation (CNV) - that was not in our published protocol. CNV onset may be reduced in those treated with emixustat but the evidence was uncertain (risk ratio (RR) 0.67, 95% CI 0.27 to 1.65, low-certainty evidence), or fenretinide (RR 0.5, 95% CI 0.26 to 0.98, low-certainty evidence) compared to placebo. A dose-dependent relationship was observed with emixustat. AUTHORS' CONCLUSIONS There is limited evidence to support the use of visual cycle modulators (emixustat and fenretinide) for the treatment of established geographic atrophy due to AMD. The possible reduction in the incidence of CNV observed with fenretinide, and to a lesser extent, emixustat, requires formal assessment in focused studies.
