Treatments for dry age-related macular degeneration and Stargardt disease: a systematic review

The sclera

The outer layer of the eyeball is the sclera, which forms part of the supporting wall of the eye. It is the ‘white of the eye’ and it surrounds most of the eye. However at the front of the eye, it is replaced by the cornea, which is transparent and allows light through.

The choroid

Inside the sclera, the next layer at the back of the eye is the choroid, which is the vascular layer of the eye, composed of blood vessels and connective tissue. There are sublayers within the choroid, including the choriocapillaris and Bruch’s membrane. The choriocapillaris consists of the capillaries that provide oxygen and nutrients to the retina.

Bruch’s membrane

Bruch’s membrane is the innermost part of the choroid, in contact with the retina. The innermost part of Bruch’s membrane is formed by the basement membrane of the RPE, which transmits waste products of metabolism from the photoreceptors (PRs) in the retina into the blood vessels in the choroid. The RPE and the choroid provide nourishment to the retinal PR cells.

Bruch’s membrane gets thicker with age, and this slows the transport of metabolites. With ageing, lipids accumulate in Bruch’s membrane. The conduction of fluids (hydraulic conductivity – the ability to let fluids pass) through the membrane is reduced. 23,24 It is thought that oxidative change in the lipids may trigger an inflammatory process, including activation of complement. Reduced transport of nutrients into the retina and reduced transport of waste products of metabolism out of it may trigger a release of VEGF in an attempt to provide more blood supply, and this may lead to the development of abnormal new blood vessels in wet AMD. The RPE has a symbiotic relationship with the choriocapillaris; if the RPE is lost, then the choriocapillaris closes down. This is believed to be the result of reduced production of VEGF by the RPE.

There is a large variation in thickening of Bruch’s membrane with age. Lommatzsch et al. 23 suggest that half of the thickening is due to natural ageing and half is due to other factors, such as genetic susceptibility and environmental factors. In early AMD, there is thickening of Bruch’s membrane due to lipid and protein deposits (drusen). 25

The retinal pigment epithelium

The RPE lies between Bruch’s membrane and the PRs. Boulton and Dayhaw-Barker26 provide a good review of its functions, which include transport of ions, fluid and metabolites; support for the visual cycle; clearance of debris; protection against light and free radicals; and production of growth factors. The RPE changes with age. In AMD, accumulation of lipofuscin in the RPE can damage it.

The retina

The retina contains the PR cells, rods and cones, and it is only 0.5 mm thick. Retinal pigmentation is partly due to the presence of melanin in the RPE. There is more melanin in the macula so it appears darker. 26 Melanin is protective but the amount of melanin falls with age, and one effect is to reduce antioxidant potential. 26

The macula

The macula is an oval area near the centre of the retina, only 5.5 mm across. It is the most sensitive part of the retina. At the centre of the macula is the fovea. The macula is responsible for high acuity and colour vision. The macula is yellowish in colour due to the macular pigments.

Macular pigments

These consist of lutein, zeaxanthin and meso-zeaxanthin, which are found in high concentration in the macula and are known as macular pigments. The first two are obtained from the diet, and meso-zeaxanthin is formed in the macula from lutein. The levels are measured as macular pigment optical density (MPOD). Their distributions in the retina vary, with meso-zeaxanthin dominating in the centre of the macula and lutein at the periphery. Carpentier et al. 27 provide an overview, with points including:

• Adipose tissue may compete with the retina for uptake of lutein and zeaxanthin so obesity may lower MPOD.

• Macular pigments protect the retina from the effects of blue light.

• In the USA, the combined intake of lutein and zeaxanthin is about 2 mg per day.

• Higher intakes appear to reduce the risk of AMD.

• Taking supplements increases MPOD when that is low and some studies report improvements in visual function.

Lutein and zeaxanthin are members of the xanthophyll family. Most lutein and zeaxanthin comes from vegetables with highest concentrations, dark leafy vegetables such as spinach and kale, and egg yolk and maize. These carotenoids have antioxidant effects, protecting the RPE from oxidative stress. Increasing dietary intake leads to an increase in macular levels. A Dutch trial28 showed increases in VA and improvement in dark adaptation.

However, Stevens et al. 29 from Aston University reported that among 158 patients recruited via the Macular Society helpline, those with AMD consumed a daily average 3.3 mg of lutein and zeaxanthin (text – table says under 2 mg/day), which is well below the 10 mg recommended after the AREDS 2 study. 30 Many patients were not eating vegetables such as spinach and kale, but a control group of people without AMD had better intake.

Reticular pseudodrusen

Reticular pseudodrusen (RPD) are a specific phenotype of early AMD first described by Mimoun et al. 33 as a yellowish interlacing network in the outer macula of AMD patients, best visualised under blue light. Other terms that have been used for RPD are subretinal drusenoid deposits, reticular macular disease and reticular drusen.

Arnold et al. 34 reported RPDs typical predominant location between the upper edge of the fovea and the supero-temporal arcade. The fundus autofluorescence (FAF) findings of RPD and its common association with RAP were first noted by McBain et al. 35 and Lois et al. 36 and subsequently supported by others. 37

The prevalence of RPD was initially reported to be 0.7% in the Beaver Dam Eye study38 and 1.95% in the Blue Mountains Eye study. 39 The 15-year incidences were 3% and 4%, respectively. Later studies utilising multimodal imaging have reported a higher prevalence of the condition [4.9% in the Rotterdam study40 and 13.4% in the Alienor (Antioxydants, Lipides Essentiels, Nutrition et Maladies OculaiRes) study41]. This could be attributed to the fact that the former studies only utilised colour fundus photography (CFP) to detect RPD but the latter used newer imaging technologies. Known risk factors for RPD include age and female gender. 38–41

Reticular pseudodrusen are associated with all stages of AMD as well as being more prevalent in late AMD. 42–46 The prevalence of RPD in early AMD was reported to be 8.4% in the AREDS study,47 36–54% in neovascular age-related macular degeneration (nAMD) and ranging from 29% to 92% in GA. 42 The Beaver Dam Eye study reported that eyes with RPD are at a sixfold higher risk of progressing to late AMD within 5 years than eyes with indistinct soft drusen but no RPD. 38 The Blue Mountains Eye study reported a fourfold increased risk. 39 Gil and colleagues48 studied the fellow eyes of patients with unilateral wet AMD and found that 58% had RPD, and that RPD increased the risk of progression compared with patients without RPD.

Fellow eyes of patients with unilateral wet AMD are known to have a higher risk of progression to late AMD. 47,49 Studies in this group have shown that presence of RPD is an independent and additional risk factor (when combined with drusen and pigmentary changes) for progression to late AMD in the fellow eye. 46,48,50,51 Eyes with RPD tend to progress to GA50,51 but some studies have also reported a higher risk of wet AMD. 34,44 In patients with established GA, Marsiglia et al. 52 reported that eyes with RPD have a higher rate of progression than eyes without RPD.

Reticular pseudodrusen have been reported to cause significant deterioration in rod function,42 although central VA is preserved, as reported by Hogg et al. 50 using the Smith-Kettlewell low luminance acuity test and by Steinberg et al. 53 using microperimetry. Compared with areas with no pathologic morphology, areas with RPD demonstrated a large and sharp decrease of scotopic sensitivity while there was only a mild decrease in photopic sensitivity. 53 Other studies have shown that RPD are associated with reduction in photopic sensitivity when compared with healthy controls or people with typical drusen. 54–56 Ooto et al. 57 suggested that in order to truly reflect a patient’s visual function, other parameters, such as contrast sensitivity and mesopic sensitivity, should be measured along with VA, as RPD are associated with deterioration in both.

Corvi et al. 58 compared MPOD in patients with RPD and people without AMD, and reported lower levels. They also reported reduced best corrected visual acuity (BCVA) and retinal sensitivity. After 3 months supplementation with lutein 10 mg/day and zeaxanthin 2 mg/day, the mean MPOD in the RPD group improved to the same levels as in the control group. However, no significant improvements were seen in BCVA or retinal sensitivity. This may be because changes in function take longer to accrue. In the CREST study, Nolan and colleagues59 found that MPOD increased by 3 months but that changes in contrast sensitivity took 12 months to reach statistical significance.

Pigmentation

Melanin in the choroid is protective against oxidative damage, and a reduction in pigment in the eye may increase the risk of developing AMD.

Retinal hypopigmentation results in paler areas and is usually associated with loss of the RPE cells. Conversely, hyperpigmentation can occur in early AMD. Neither change is specific to AMD. Another pigment, lipofuscin, appears harmful. It is composed of lipids and protein. A major component is a retinoid A2E (N-retinyl-N-retinylidene ethanolamine) which is a by-product of the visual cycle.

The ‘visual cycle’

Light reaching the photoreceptors in the retina triggers the conversion of the light-sensitive retinoid 11-cis-retinal into a different form, 11-cis-retinol, thereby generating an electrical signal to the brain. The trans form is then converted back to the cis form in the RPE and then returns to the photoreceptors, completing the visual cycle. If the two molecules of the trans form combines with one of the lipids (phosphatidilethanolomine) in the RPE, A2E is formed, and this can impair RPE function. Because the edges of patches of GA are thought to have A2E accumulation [as reflected in increased autofluorescence (AF)], reducing that accumulation may be one target of drug treatment.

Oxidative stress

Oxidative stress is defined by Betteridge60 as a disturbance in the balance between the production of reactive oxygen species (free radicals) and antioxidant defences. The retina has a very high metabolic rate, reflected in high oxygen consumption, and has a high concentration of polyunsaturated fatty acids and exposure to light, which, if coupled with inadequate levels of antioxidants, can make it very susceptible to oxidative stress. Yehoshua and Rosenfeld61 report that the evidence for cumulative oxidative damage being the cause of AMD has been growing, but that a mechanism for it is not yet known.

Barnett and Handa62 suggest that oxidative stress can affect the immune system, turning it from a protective to a pathological response, and can also lead to chronic inflammation.

The immune system

Ambati and colleagues63 have reviewed the immunology of AMD. In brief, they consider that overactivity in the alternative pathway of the complement system is involved in the development of AMD. This is associated with the genetic susceptibility via a variant of the CFH gene, known as CFH (402His), which causes a greater than normal complement response to retinal injury. Ambati and colleagues63 suggest that in individuals with ‘a complement hyperinflammatory phenotype’ there is an over-reaction to cellular damage in the retina.

Anderson, Hageman and colleagues64 first described the role of inflammation in AMD, and put forward the hypothesis that drusen were the result of local immune-mediated processes and the junction of the RPE and Bruch’s membrane.

The pathological role of the complement system has led to trials of drugs to inhibit that system.

Age-related macular degeneration

Classification (Macular Research Classification Committee 2013):65

1. Normal ageing – people with small drusen (< 63 µm), also termed drupelets, should be considered to have normal ageing changes with no clinically relevant increased risk of late AMD developing.

2. Early AMD – medium drusen (≥ 63 to < 125 µm), but without pigmentary abnormalities thought to be related to AMD.

3. Intermediate AMD – large drusen or with pigmentary abnormalities associated with at least medium drusen.

4. Late AMD – neovascular (wet) AMD or GA (advanced dry AMD).

In this report, we use the term dry AMD to cover all stages from early AMD to GA.

Early and intermediate AMD is characterised by drusen, and/or by changes in pigmentation. 25 However, most people with drusen will not progress to severe visual loss and drusen may cause only mild or no visual symptoms. Up to 80% of people > 60 years have some drusen. Hard drusen are well-defined yellowish deposits with little risk of progression.

Soft drusen are larger, not well demarcated and are associated with a high risk of progression to late AMD. They may become larger and merge over time and can lead to RPE detachments, called drusenoid RPE detachments. They may disappear, but this is usually associated with atrophy of the outer retina. Drusen are associated with thinning of the overlying RPE. Fleckenstein and colleagues66 consider that GA is the natural end stage of soft drusen. A key component of drusen is amyloid beta,67 which is a waste product.

The underlying processes include locally intensive metabolism, oxidative stress, chronic inflammation, a pathological immune response and lipofuscin accumulation. 9 Lipofuscin is considered toxic.

In atrophic AMD, there may be a single patch of atrophy or several. Over time, the patches may get bigger and may merge. The foveal centre (the area responsible for central vision) is lost last as atrophy occurs around the centre of the macula first before expanding into the fovea, which is the very centre of the macula. This potentially gives time for treatment before the central vision is lost.

Vision is lost from atrophic patches and the gaps in vision are called scotomas.

The atrophy is due to loss of the RPE, outer layers of the retina and the underlying choriocapillaris. 11,66 On optical coherence tomography (OCT), GA appears as a flat patch where the retinal has withered away. A total of 20% of people with legal blindness have lost central vision due to GA. It tends to be of similar extent in both eyes66 but patients can have GA in one eye and wet AMD in the other, and can also have both GA and wet AMD in the same eye, if late dry AMD turns to wet AMD.

Geographic atrophy is also seen in patients treated with anti-VEGF drugs for wet AMD. 35,68 In both the Comparison of Age-Related Macular Degeneration Treatments Trials (CATT)69 and Inhibition of VEGF in Age-related choroidal Neovascularisation (IVAN)70 trials (of ranibizumab vs. bevacizumab in wet AMD) it was observed that about one-fifth of patients developed GA after 2 years of anti-VEGF treatment. 71 This GA appeared to be clinically similar to the GA that is seen in dry AMD and may occur because VEGF is required for the maintenance of the choriocapillaris by the RPE.

Progression and natural history

Data on natural history studies are important because natural history may be the only comparator for some interventions reported in observational studies.

Wet AMD will develop in 10–15% of people with intermediate AMD. 72 In the AREDS trial, the average time to atrophy was 5–6 years in people with large drusen and hyperpigmentation, but 2.5 years in those with hypopigmentation.

Most people with AMD are at the early stage,2 as shown in Table 1.

The KORA (Cooperative Health Research in the Region of Augburg) study from South Germany reported features of AMD in people < 50 years. 73

The Geographic Atrophy Study by Sunness et al. 74,75 reported that GA enlarged at 2.6 mm² per year over a median follow-up of 4.3 years, in 212 eyes in 131 patients, mean age 78 years. However, there was a very wide range of progression rates from none to almost 14 mm² per year. They noted a high concordance in rates of enlargement between eyes.

The Geographic Atrophy Progression Study,76 in patients with a mean age of 77 years, found that the GA enlarged by an average of 1.85 mm² over 12 months, based on AF, and this was slightly higher based on CFP.

The Beaver Dam Study77 found a progression rate of 1.3 mm² per year in 53 eyes of 32 patients (mean age about 81 years) over 5 years.

The AREDS trial group reported progression of 1.7 mm² per year in 251 eyes of 181 patients (mean age 70 years) over a median follow-up of 6 years.

The FAM (Fundus Autofluorescence in age-related Macular Degeneration) study78 reported a similar progression rate of 1.75 mm² per year (mean) or 1.52 mm² per year (median) in 195 eyes of 129 patients (mean age 74 years), but over a median follow-up of only 1.8 years. They also reported a wide range of progression rates. They used FAF to determine areas of GA.

Outcomes

The most important outcomes are those that matter to patients: VA, contrast sensitivity, adverse effects of treatment, reading speed, ability to drive, health-related quality of life, progression of disease and patient preference.

However, VA loss is a late manifestation of AMD and not a good primary outcome in most trials, especially when the treatment is aiming at prevention of visual loss before it occurs. Early AMD may cause minimal or no symptoms. VA depends only on the centre of the fovea but this tends to go last in atrophic AMD and many patients have large areas of atrophy and experience considerable problems before the fovea goes. Reading and seeing faces of people can be extremely difficult and the ability to drive may be lost.

Progression of dry AMD is slow, and so it could be years before a trial could show a decline in vision. Therefore, predictors or biomarkers of future decline can be accepted if there is good evidence that they are strong predictors of subsequent visual outcomes. These will include changes detectable by investigation, but not necessarily by people with AMD, including:

• Rod function, which may not correlate with VA as central VA acuity (as measured using VA charts) depends on foveal function, and the fovea is cone rich. But rod function is one of the earliest abnormalities detected in people who will later develop GA in AMD.

• Macular pigment density, because it appears to be protective.

• Integrity of the RPE layer, as determined by FAF and OCT.

• Drusen volume and number. Disappearance of drusen may be a sign of developing GA.

• Macular sensitivity, which can be measured by microperimetry.

• Dark adaptation.

Both photopic and scotopic vision need to be considered. Scotopic vision refers to low levels of light such as in near darkness.

One issue is the clinical significance of changes in VA. In past evaluations, for example of the anti-VEGF drugs, a clinically significant difference in VA has usually been considered as a change of ≥ 10 letters. Changes of < 5 letters are not regarded as clinically relevant as may indicate normal variability. Changes of 5–9 letters are not regarded as clinically useful but might be regarded a valuable outcome to investigate if seen in a short-term study, suggesting that a larger or longer trial is justified.

In dry AMD, no change (which could be lack of deterioration) could be regarded as clinically meaningful if observed over a long enough period.

Microperimetry

Microperimetry can detect changes in macular sensitivity in patients with early AMD and normal VA. 79

Macular sensitivity measured using microperimetry focuses on the central macula instead of the entire visual field. 80–83

Testing is performed either with a modified Humphrey Field Analyser or with a microperimeter. 83

There is limited evidence of the reproducibility of microperimetry in patients with AMD, but current studies have suggested that it provides consistent and reproducible readings. 84–86

Early AMD patients have rod sensitivity loss87 and impaired rod-mediated parameters of dark adaptation,88 which worsen as AMD progresses. The association between early AMD changes and macular sensitivity was further established by the observation that a correlation existed between altered AF signal and reduced macular sensitivity. 79,89 In GA, macular sensitivity was reduced in areas of increased fundus AF signal at the junctional zone of areas of atrophy. 90 However, this observation has not yet been proven to be a predictor of GA enlargement over time. 90

Current evidence suggests that macular sensitivity is a valuable biomarker for early AMD and microperimetry has proven to be an easy and reliable test to measure it. It is not widely used in clinical practice, but has been used in clinical trials to evaluate the effects of treatments on macular sensitivity. 91–96

Search strategies

MEDLINE, EMBASE, Web of Science and The Cochrane Library were searched from 2005 to 13 July 2017 for reviews, journal articles and meeting abstracts. Searches were limited to English language.

Initial searches of all databases were undertaken in June 2016 and updated searches were run in June 2017 to check for any articles added in the previous year. The Association for Research in Vision and Ophthalmology (ARVO) website was also searched for meeting abstracts.

References of reviews were checked for relevant studies and clinical experts were also consulted for any other relevant literature.

Studies were selected for inclusion through a two-stage process using predefined and explicit criteria. Titles and abstracts of 7948 articles from the full literature search results were screened independently by two reviewers to identify all citations that appeared likely to have met the inclusion criteria, and checked by a third. The full texts of 398 articles were obtained for further screening and checking of references and 112 articles were included in the final report.

ClinicalTrials.gov, the WHO search portal and UK Clinical Trials gateway were searched for ongoing and recently completed clinical trials.

Full details of the search strategies are in Appendix 1.

Inclusion and exclusion criteria

Reviews

One recent review of treatments for STGD has been published by Lu et al. 112 It claims to be systematic but no details of methods are given, and not all therapies were covered. Three other non-systematic reviews were identified. 100,117,118 A very thorough report from the National Horizon Scanning Centre (NHSC)119 was found, covering all inherited retinal conditions, based on extensive searches and discussions with experts. All these reviews were used to check for references, including to ongoing research. The NHSC report119 was particularly useful.

Lu et al. 112 conclude that the most promising drug treatments for STGD are drugs that reduce lipofuscin accumulation, such as deuterated vitamin A (ALK-001; Alkeus Pharmaceuticals, Boston, MA, USA), fenretinide (Sirion Therapeutics, Tampa, FL, USA) and A1120, a non-retinoid RB4 antagonist.

A more recent review by Tanna and colleagues120 provides a detailed review of diagnostic methods but adds little on treatments. Lambertus et al. 104 have also reported ways of monitoring progression of STGD, focusing on late-onset STGD.

Studies

Four studies and two subgroup analyses of studies included people with STGD. The interventions in these studies were microcurrent stimulation (MCS) (two studies),97,121 light protection (one study),122 RPE transplantation (one study)123 and nutritional supplements (two studies). 124,125

Summary of microstimulation

The small RCT by Röck and colleagues,121 with an unclear risk of bias, did not find any statistically significant benefit from MCS. We found no good evidence that MCS worked.

Summary of light protection

One very small study reported reduced progression in the light-protected eye in four out of five participants. So although the evidence base is very weak, and we need a proper trial, we could recommend that people with STGD should wear sunglasses or dark contact lenses to protect their retinas from light exposure.

Adverse events

There were no AEs specifically from the cellular therapy (e.g. acute transplant rejection or abnormalities in retinal vascular or choroidal circulations) but one patient developed an endophthalmitis following surgery, which was a potentially devastating complication.

Summary of retinal pigment epithelium transplantation

One small before-and-after study98,123 found improvements in VA in most of the nine eyes after 12 months, whereas there was little change in the untreated fellow eyes. Improvements in quality of life were also noted.

Discussion

The NHSC report119 discussed the prospects for RPE cell transplants, quoting expert opinion doubts about the effectiveness of hESC-derived RPE cells. This was on the grounds that most people with STGD may need replacement of both RPE cells and PRs, if the treatment is undertaken at a late stage of disease, when PR cells may be already lost and permanent damage to the choriocapillaries may have taken place. Indeed, the early clinical trials referred to above selected patients with very advanced STGD. The treatment could potentially be more promising if the disease were to be treated at an earlier stage and this could be done provided that the risks of the treatment were small.

Research in progress

Research in collaboration with Ocata Therapeutics (formerly Advanced Cell Technology, now part of Astellas) has been under way at Moorfields since 2012.

NCT02445612 is the long-term (15 years, to 2029) follow-up of the NCT01345006 study by Schwartz et al. described above.

NCT02941991 is a 5-year follow-up study in Moorfields and Newcastle of people treated with hESC RPE cells, due to complete data collection in December 2019. The sponsor is the Astellas Institute for Regenerative Medicine.

NCT01469832 was called ‘Safety and Tolerability of Sub-retinal Transplantation of Human Embryonic Stem Cell Derived Retinal Pigmented Epithelial (hESC-RPE) Cells in Patients With STGD Macular Dystrophy (SMD)’. It was being carried out in Moorfields, Newcastle and Edinburgh, aiming to end by 2015. It was sponsored by Astellas Institute for Regenerative Medicine. NCT02941991 looks to be a continuing follow-up, which is due to end in 2019. The entry for NCT01469837 has disappeared from Clinical trials.Gov.

NCT01625559 is a small study in Korea aiming to recruit three patients. No details have been added recently.

NCT02749734 aims to recruit 15 patients in China (Southwest Hospital) for a 12-month study.

Two stem cell studies are reported to be under way, called Stem Cell Ophthalmology Treatment Study (SCOTS) I and II, which are sponsored by MD Stem Cells. These are being carried out in nine different eye diseases, and involve injecting bone marrow cells (from the hip) into the eye by different routes. There is no control group. SCOTS I (NCT01920867) is due to end in August 2017. The start date for SCOTS II (NCT03011541) was January 2016 and end date will be 2020. The sites are Florida (Retinal Associates of South Florida) and Dubai (Al Zahra Hospital). A couple of single case reports have been published, but not in STGD, except for a case report from another centre of a retinal detachment following stem cell transplantation in a SCOTS I patient (Leung et al. 128). Media reports are that patients are being charged $20,000 for the treatment. 129

NCT02903576 is under way in São Paulo, Brazil, recruiting 18 patients with STGD and AMD, and due to end in 2018.

Many of the studies registered on ClinicalTrials.gov are not trials, and patients should be aware that registration does not mean approval by scientific authorities. Several who saw an entry for NCT02024269 had adipose tissue cells injected into their eyes. Kuriyan and colleagues130 report on the resulting visual loss. The patients were charged for the treatment and the ‘trial’ registration has been withdrawn.

Summary of nutritional supplementation

One small short-term study of lutein supplementation failed to find a beneficial effect on VA and dark-adapted sensitivity. 114 A trial of DHA supplementation found only slight improvement in VA in 4 out of 20 people after 6 months of treatment. 125 It should be noted that treatments may be effective without improving vision if they prevent further deterioration, but this could only be shown by longer-term RCTs.

We suggest an exploratory EME study of lutein and zeaxanthin in STGD. There is theoretical support for them being of value in protecting the centre of the macula, through protection of visual cycle products from photo-oxidation, and hence reducing accumulation of the toxic A2E. Aleman and colleagues114 found that the eyes of those with STGD have lower MPOD than normal eyes and about half the serum levels of the carotenoids, compared with a control group, despite similar dietary intakes. That study was published in 2007. No trials of carotenoid supplements are registered on ClinicalTrials.gov.

We suggest an EME ‘proof of concept’ trial using sensitive measures of retinal function rather than changes in VA. MPOD and VA would be measured, but the aim would be to recruit people before vision was significantly impaired to determine whether functional loss could be delayed or prevented. So the study should use macular microperimetry or multifocal ERG to test macular function, reading vision and reading speed, visual-related quality of life and AF. The power of the study would be increased by recruiting faster progressors in STGD, which would probably allow some outcomes to be determined after shorter follow-up, perhaps 1 year. Because treatment would be systemic, both eyes will be treated (so no possibility of using one eye as treated and one eye as control) and, therefore, adequate controls would be required. It would be essential that all patients (treated and controls) are similar with regard to characteristics known to affect speed of progression of disease as well as genotype, which could be achieved by a large enough trial, and/or by stratified randomisation. Although genotyping may pose problems if mutations on the ABCA4 gene are not found in both alleles. A matched placebo would be required to reduce the risk of the control group self-treating with supplements.

Soraprazan

Soraprazan is a proton potassium-competitive acid-blocker developed for use in dyspepsia, but discontinued for that use. It was granted orphan drug status by European Medicines Agency (EMA) for use in STGD in Germany in November 2013. 137 An orphan designation is not a marketing authorisation, but only approval for investigation. The EMA noted that no trials had been started.

The EMA document provides an accessible explanation of how soraprazan might work:138

Soraprazan is expected to be able to enter the cells of the retina, where it attaches to the abnormal deposits that damage the retina cells. Soraprazan is thought to cause the deposits to break up and partly dissolve. The broken-down deposits can then be expelled by the cell’s own natural mechanisms, reducing their build-up and the damage to the cell.

Reproduced with permission from © European Medicines Agency, 2013138

Reproduced with permission from © European Medicines Agency, 2013138

Isotretinoin

This was reported in a review by Battaglia Parodi et al. 139 to delay visual loss in STGD but to have adverse effects including liver toxicity. The evidence comes from the mouse model (ABCA4^–/–), in which it reduced A2E and lipofuscin granules, but at a dose too high for human use.

Dobesilate

Dobesilate has been reported in a single case history95,140 to improve VA 4 weeks after a single injection, but with no changes in fundus photography, fluorescein angiography or foveal thickness. Dobesilate is an inhibitor of fibroblast growth factor.

4-methylpyrazole

A placebo controlled trial (NCT00346853) of 4-methylpyrazole was conducted in Utah in healthy adults to see if it would improve dark adaptation. Jurgensmeier et al. 141 reported that there were no significant differences after six intravenous (i.v.) injections and they concluded that further trials in humans were not justified.

ALK-001

This compound is C20 deuterated vitamin A. In deuterated compounds, hydrogen is replaced by deuterium (deuterium is an isotope of hydrogen which has a neutron in the nucleus, as in ‘heavy water’).

In the retina, vitamin A has a tendency to ‘dimerise’, which means that two vitamin A molecules join together. The combined compound is A2E, which is a major component of lipofuscin. Deuterated vitamin A is much less likely to dimerise, and so the deposition of A2E in lipofuscin is reduced by about 80% in a study in mice by Charbel Issa and colleagues. 142 For a recent review of rationale and of animal studies, see Saad and Washington 2016. 143

The rationale for use is that ALK-001 interferes with the vitamin A processes that lead to the formation of A2E and lipofuscin accumulation. A Phase I trial assessed safety in healthy volunteers (NCT02230228) and a Phase II trial is under way, called TEASE – Tolerability and Effects of AKL-001 on Stargardt disease (NCT02402660), which aims to recruit 50 patients followed for 24 months. The sponsor is Alkeus Pharmaceuticals, Boston, MA, USA.

Fenretinide

Fenretinide is a synthetic form of vitamin A. Administration leads to reduced levels of retinol binding protein (RBP) which, in turn, leads to decreased levels of vitamin A in the eye and reduced accumulation of the toxic A2E. Safety concerns (angiosarcomas) have been raised following research in mice (see Lu et al. 112 for review) but it has been used in a large study144 of 246 patients with dry AMD in two doses: 100 mg and 300 mg (see Chapter 5). Adverse reactions were common in a group with a median age of 79 years, and 17% of the lower-dose group and 20% of the high-dose group stopped the drug, compared with 6% on placebo. It may be worth a trial in STGD, in a much younger and fitter population. We have not found any such trial on Clinicaltrials.gov. It has been trialled in many cancers and other diseases so safety data are available in younger age groups.

LBS-500

This drug (www.linbioscience.com/Pipeline/LBS008)145 from Lin BioSciences (Taipei City, Taiwan) is expected to work in a similar way to fenretinide by reducing RBP4. It was granted orphan drug status by the Food and Drug Administration (FDA) for STGD, and a Phase I trial will be carried out in the USA in collaboration with Columbia University, supported by NIH. 146

A1120

This was originally developed for type 2 diabetes (to improve insulin sensitivity) but was ineffective. However, it competes with vitamin A for binding on to RBP and reduces accumulation of A2E and lipofuscin in a mouse model of STGD. 147 We have not found any human studies in STGD.

Emixustat

A Phase II pharmacodynamic study (NCT03033108) started in January 2017, which will look at safety and retinal responses to flashes of light (measured by electroretinogram), and was sponsored by Acucela (Seattle, WA, USA). There will be three different doses but no placebo group. Data collection is due to end in December 2017.

Fenofibrate

A very recent conference abstract by Moiseyev and colleagues148 has reported that fenofibrate may be a visual cycle inhibitor, and, if so, may have potential to decrease the accumulation of A2E in STGD.

Fenofibrate has been used for many years in hyperlipidaemia, but is currently of interest in diabetic retinopathy. An unexpected finding from the FIELD (Fenofibrate Intervention and Event Lowering in Diabetes) trial149 was a reduction in progression of diabetic retinopathy. The LENS (Lowering Events in Non-proliferative retinopathy in Scotland) trial150 of fenofibrate to reduce progression of diabetic retinopathy is under way, funded by the HTA programme.

Full publication of the study by Moiseyev and colleagues148 is awaited, and further pre-clinical research is probably required before clinical trials can start. However, fenofibrate is an old, cheap and safe drug so if it is proven to be a visual cycle inhibitor, then a trial in STGD (and dry AMD) would be justified.

Avacincaptad pegol

Avacincaptad (Zimura, Ophthotech, New York, NY, USA) is a C5 complement inhibitor or anti-C5 aptamer. A trial (NCT03364153) in STGD is due to start in December 2017, aiming to recruit 120 patients. An earlier smaller trial, NCT00950638, does not seem to have been published yet.

Another trial (NCT02686658) is under way in the USA and Hungary. The aim is to recruit 300 people to three arms: two doses of the drug and sham. It is due to end in December 2018.

There is another Zimura registration on the EU Clinical Trials Register, EUCTR2015-003991-56-HU, for a trial with a target of 900 patients, but without any sites mentioned, although the ‘HU’ implies Hungary.

Prevention of Stargardt

Sohrab and colleagues151 report a single case in which a man with STGD and his wife, who was a carrier but unaffected (i.e. heterozygous, carrying only one copy of the responsible gene), had IVF and pre-implantation diagnosis to identify an embryo with a normal maternal allele before implantation. The child was born healthy and will be an unaffected carrier.

Quantity and quality of research

Results

Consecutive patients with dry AMD (details of stage not given), as diagnosed by their ophthalmologist, were included. After 2 weeks of acupuncture, the median VA at 3-m distance and at 40-cm distance improved in 44.2% and 88.4% of participants, respectively. Vision at 3 m and at 40 cm worsened in 4.3% and 2.7% of participants, respectively. No statistical analyses were presented.

Summary

One large before-and-after study found improvement in vision after 2 weeks of acupuncture; however, no statistical analyses were presented and the study was assessed as low quality.

Registered studies

The study NCT02255981 aims to use acupuncture to treat 33 patients in Columbia with unspecified macular diseases. With no control group, it is unlikely to be helpful.

Mechanism of action

The outer layer of the retina has an abundant supply of oxygen from the choriocapillaris and oxidative damage, triggered by the effects of light, is a risk. Several studies have shown that people most exposed to sunlight, like fishermen, have a higher risk of AMD (see Cuthbertson review154). The risk may be higher in people who are both exposed to sunlight and have low antioxidant levels. 155

The structure of the human eye reduces penetration of harmful light. Protective mechanisms include the structure in front of the retina (especially the lens), various enzymes and the macular pigments lutein and zeaxanthin, which absorb blue light.

Cataract surgery is common in older people and the natural lens is replaced by an artificial one – an artificial intraocular lens (IOL). These can be protective only against ultraviolet radiation (UVR) or against both UVR and blue light. The argument against additional blue-light-filtering appears to be based on the effects on vision in poor light (scotopic vision) and circadian rhythms and sleep. Scotopic vision declines with age and is reduced by blue-light-filtering IOLs.

People with artificial lenses are described as ‘pseudophakic’. People who still have their own lens are called phakic.

There are several uncertainties:

• Are people more at risk of AMD after cataract removal?

• If so, do blue-light-filtering IOLs prevent AMD?

• In people with existing AMD, do blue-light-filtering IOLs reduce or prevent progression?

Quantity and quality of research

Summary

The evidence is mixed. The Davison156 and Henderson157 reviews make a good case for routine use of blue-light-filtering IOLs. The Pipis trial162 provides some support for a reduction in progression but scores poorly on quality. The Chong,165 Lavric164 and Łak166 studies, which have a stronger design with patients having standard IOLs in one eye versus blue-light-filtering ones in the other, report no differences, but two (Chong165 and Łak166) are available only as abstracts. The adverse effects on scotopic vision and sleep disturbance do not seem to be clinically significant.

However, a very large trial is under way, called the CLOCK-IOL colour study. Nishi and colleagues168 from Nara in Japan report that they plan to randomise 1000 patients after cataract surgery to a standard IOL or a blue-light-blocking IOL, with AMD one of the primary outcomes, assessed first after 1 year but with follow-up planned for up to 20 years. The protocol has been published. 168 The trial is registered on the University Hospital Medical Information Network as UMIN000014680. 169

Mechanism of action

Brunner and colleagues170 suggest that a key feature in AMD is impaired choroidal perfusion, and that reduction of plasma viscosity can improve this. However, reduced viscosity may be more important in the retinal circulation where vessels are smaller (3–10 µm compared with 20–40 µm in the choroidal capillaries). Perfusion pressure may be more important than viscosity as choroidal vessels are wider, fenestrated and have lower pressure than retinal vessels. 171,172

Quantity and quality of research

Combinations of rheophoresis and other interventions

Two studies182,183 used rheophoresis in combination with plasma exchange and selective adsorption.

Brunner et al. 182 reported in a case series in 1996 that a combination of plasma exchange, selective adsorption and membrane differential filtration improved VA by at least 1 line, 24 hours after treatment in 11 out of 17 patients (15 out of 31 eyes) with AMD (about half with wet AMD and results not given separately). A single ETDRS line is not usually considered clinically important.

In a longer-term case series with 20 participants with dry AMD (available only as an abstract), Widder and colleagues183 reported significant improvements in BCVA 1.9 lines at 2 years and 1.2 at 3 years but not at 4 years follow-up, after combined plasma exchange, selective adsorption and filtration.

A small trial by Kamami-Levy and colleagues189 is as yet available only as an abstract from ARVO 2014. Patients were randomised to rheopheresis (n = 10) or controls (n = 11). The control intervention, if any, is not specified. After 24 months there were no statistically significant differences in VA or atrophic area.

Adverse events

Adverse event rates were low in three studies (0%, Studnička 2013;180 2%, Koss 2009;175 5%, Blaha 2013;179) (for specific events see Report Supplementary Material 2). However, Pulido 2006174 reported that during the period of follow-up, 34% of haemopheresis participants and 28% of control participants experienced an AE, with serious AEs in 24% of treated participants. Few details are given of what these serious AEs were. No serious AEs recorded by Brunner 2000. 170 In the study by Klingel 2010,181 0.24% of people with dry AMD experienced retinal bleeding (Table 8). AEs were not reported in the Rencova177 or Swartz178 studies.

Summary

Seven intervention studies, with unclear risk of bias in four170,174,176–178 and poor quality in two,190,191 were included. Only one study (Koss192) was thought to have a low risk of selection bias. Results were mixed, with statistically, but not really clinically, significant effects of haemopheresis on VA reported in only three studies, all of which included very small numbers of patients. The studies that did not show differences between groups in BCVA included Pulido,174 with by far the largest numbers, and Studnicka 2013180 and Blaha 2013179 with longer periods of follow-up.

Overall, the results do not provide a strong case for the value of haemopheresis on VA. The two studies from Germany by Koss et al. 192 and Brunner et al. 170 reported statistically significant differences, and Lane179 reports that the largest German medical insurance company, Deutsche Kranken-Versicherung AG, agreed to fund rheopheresis. The largest study, the MIRA trial by Pulido et al.,174 was deemed to be negative and we could not find any FDA approval for the treatment on the FDA website. One interesting feature, reported by Brunner and colleagues,170 was that the effect of rheophoresis persisted for much longer than the reduction in viscosity, implying that a relatively short period of reduced viscosity could have longer-lasting effects.

The practicality and acceptability of having older people attend for 2- to 4-hour sessions, about 10 times a year, would have to be considered.

A related technology is HELP (heparin-induced extracorporeal lipoprotein precipitation). A study, NCT01840683, was registered in 2013 to be carried out at the Canadian Centre for Advanced Eye Therapeutics, a private company in Mississauga, Ontario, Canada. Some results were published by Ali and Armogan in Retina Today (Sept/October 2008, 72–75). After eight HELP sessions, VA was reported to have improved by at least 1 line in just over half of 33 eyes of 19 patients, and by ≥ 3 lines in 15% (about 5 patients). Drusenoid macular thickness was reduced in over half the eyes. No full publications were found, but there is a conference abstract. 193

Conclusion: given that most positive studies were small, that they mostly had uncertain risks of bias, that effect sizes were modest, that the largest one was negative and that treatment would be inconvenient to older people patients, we do not see rheopheresis as a research priority.

Quantity and quality of research

Results

Anastassiou and colleagues 2013196 included 22 people with dry AMD, no neovascular disease and VA between 25 and 45 letters. Details of the stage of AMD were not given. At 6 months, there was no statistically significant difference in the change in VA, contrast sensitivity or macular sensitivity (Table 9). However, they did report an improvement in VA after 1 week, an improvement of > 10 letters in three patients at 4 weeks, and an increase in contrast sensitivity at 4 weeks. Anastassiou and colleagues 2013196 concluded that transpalpebral electrostimulation led to a temporary increase in visual function in some patients, but that further research was needed.

Shinoda and colleagues 2008199 included people with wet or dry AMD (wet not reported here). The inclusion criteria for dry AMD imply that GA was present. The study was graded as fair on the NIH checklist but it was very small, containing only seven patients with dry AMD. They reported a statistically significant improvement in the number of ETDRS letters in the dry AMD group from 39.8 to 42.9 (p = 0.04) after 4 weeks (see Table 9). However, they conclude that the limitations of the study (small numbers, short follow-up and the absence of a control group) means that it cannot be used to justify the use of MCS. They recommend a larger RCT with longer follow-up.

The Kondrot 2002208 study was graded as very poor quality, achieving only one positive answer out of 12 items on the NIH checklist. Kondrot 2002208 treated people with dry AMD (no details are given) with a combination of MCS, vitamins and nutritional supplements (details of those not given), starting with a pilot of 10 patients. There was no loss of vision, change in Amsler grid or change in intraocular pressure (IOP) in the first group of 10 patients, so the study was then expanded to treat 56 eyes of 28 patients with no control group. Before treatment, the range of VA was 20/25 to 1/400. After treatment, up to 1-year follow-up, the range of VA was reported to be between 20/20 and 3/800. The proportion of participants with improvement of VA was 66% and the range of improvement was mean 0.48 lines, range 0–2.5 lines. No other outcomes were reported. Dr Kondrot suggested that a double-blind study was desirable to address the possible placebo effect.

Chaikin and colleagues 2015198 included six people with wet and 25 with dry AMD (details not provided). MCS was given by the palpebral route, for 35 minutes once a week, for an average of 4.8 sessions (range 2–10). We report only results in dry AMD. There was no control group. The study was graded poor, achieving only four out of 12 items on the checklist. At 3 months, treated eyes showed an improvement in logMAR VA (see Table 9). There was a sharp improvement in the first week, continuing in the first month, followed by a levelling off. The authors stated that the mean letter change from baseline was statistically significant (p = 0.012), although a figure in the publication suggests this was not significant (p = 0.059). There was improvement in VA (defined as a gain of ≥ 1 letter) in 52% of eyes and deterioration in 28%. The authors cautiously conclude that the short follow-up and small numbers, the training effect of repeated VA testing and the lack of masking of the VA testers mean that long-term efficacy is unproven. They encourage further studies including a control arm with sham treatment. They conclude that if further studies supported the positive results from their study, that the treatment would be an easy one to administer in offices or at home.

In a large retrospective before-and-after study, graded poor on quality assessment, of a package of alternative treatments (i.v. nutrition, oxidative therapy, MCS and ‘syntonic light therapy’) combined with a stress reduction programme and a detoxification programme, Kondrot 201597 reported a 5.5-letter improvement in VA at follow-up (likely immediately following 3 days of treatment) (see Table 9). Improvement of < 1 line was apparent in 35.7%, 1–2 lines in 37.9% and > 2 lines in 15.7%. There was also an improvement in contrast sensitivity of 3.8 letters (Report Supplementary Material 2). However, given the combination of therapies, no conclusions on MCS stimulation can be reached.

In one publication, two linked studies of MCS in people with dry AMD (no further inclusion criteria) were reported, one of which had two separate cohorts of 12 and 34 patients, all patients treated, with no controls. 204,205 In the first study there was a decrease in VA after 2–7 years’ follow-up (see Table 9). In the second study, few results were reported; there was a mean loss of 3 letters of VA in cohort 1, and a mean gain of 8.5 letters in cohort 2. Few other results were reported (see Report Supplementary Material 2). In addition to electrotherapy, a nutrition extract of taurine, rutin and bilberry was administered in the second study by Allen et al. 205 There was no control group.

Adverse events

Anastassiou et al. 196 reported that there were no adverse effects of treatment. Shinoda et al. 199 reported that there were no ocular or systemic complications except in one participant who developed contact dermatitis on eye lids, which was treated as a serious AE. It is not clear from the publication if this was in a dry AMD patient. Three studies did not report AEs. 97,198,208

A pilot study (not included above) with 17 patients from Mumbai by Natarajan et al. 209,210 reported improvements in VA after MCS treatment with the ScyFIX device (Chanhassen, MN, USA; http://www.scyfix.org/.) Natarajan et al. 210 recommend a RCT with long-term follow-up.

Summary

The American Academy of Opthamology Taskforce concluded there was no strong evidence on the effectiveness of MCS stimulation, but it preceded many of the studies listed above. The studies reported mixed results of MCS. Follow-up was generally short and sample sizes small. In the RCT there were no significant benefits of treatment. In the before-and-after studies and the case series, results were difficult to interpret owing to differences in measures, time points, small samples, study quality and a lack of statistical analysis. In addition, in the case series, participants were given long-term nutritional supplements. In the immediate to short term following treatment, there may be some improvement in VA, but most authors expressed caution and suggested RCTs. Few studies reported other outcomes.

Overall, the evidence base for MCS is weak. Despite this, we note that Nova Oculus Partners (Indian Wells, CA, USA), manufacturers of an ‘electrotherapeutic device’ to treat dry AMD (unspecified), are hoping to get International Organization for Standardization (ISO) certification worldwide. 211

A device is being marketed by ScyFIX for ‘microcurrent neuromodulation’. It has a Conformité Européene (CE) mark in Europe. The manufacturer claims that microcurrent neuromodulation has been shown to be effective. 212 In their submission to the FDA, ScyFIX cite a study by Wallace 1997206 who reported treating 43 patients with dry AMD with microcurrent (200 µA for 20 minutes for 36 sessions) after which 54% improved by 1–4 Snellen lines. 212

ScyFIX212 also cite a paper by Paul (undated) which claims a 72% success rate in AMD after microcurrent and nutritional treatment. They also cite a presentation by Halloran and Reader213 (Fourth Annual Symposium on Biologically Closed Electrical Circuits 1997) in which improvement in visual field function was reported to be 4.61 decibels, and in VA 0.98 lines. The treatment was part of a package of treatments, including some described as applied kinesiology, neurolymphatic deep stimulation, deep tissue acupressure, colour-shape identification therapy and nutritional supplements.

Current research

We have identified five trials in ClinicalTrials.gov (accessed 16 April 2017).

NCT01600300 aimed to do a feasibility trial of the Tesmac device versus sham Tesmac (Acuity Medical International, Minneapolis, MN, USA). The completion date was 2004 but no results have been posted. The Tesmac device is now owned by The Eye Machine (Indian Wells, CA, USA). 214

NCT00804102: the study details mention a range of eye diseases including dry AMD but the only publication posted was on retinitis pigmentosa. However, the research group includes Röck from Tubingen, so this may have been the Röck 2013121 study described in Chapter 2.

NCT0170958 was an observational study in AMD, STGD and retinitis pigmentosa, led by Papastergiou in the Retinal Institute in Hawaii. No results have been posted. Completion date was 2012.

NCT02699216 is listed as currently recruiting in the Du Bois Vision Clinic, PA, USA, but expected completion date was December 2016. It will assess with the Novo Oculus device (The Eye Machine).

NCT02540148 was to be another trial of the Novo Oculus device against sham treatment, but is listed as not yet recruiting. The details were provided by The Eye Machine Canada (Indian Wells, CA, USA) but have not been updated since 2015.

We are aware of proposals for a further study of MCS being planned by Oxford Bioelectronics. A leaflet in February 2017 said they would be embarking on a pilot study to establish safety and clinical acceptability of the device as a treatment for dry AMD, and that if successful, a clinical trial for regulatory approval of a bioelectronic treatment would follow.

It is not clear if the pilot will have any control patients. If it does not, it will contribute little new. However, the proposal for a full clinical trial, if that is a RCT, is welcome. No further details were on the company website (www.oxfordbioelectronics.com/amd/; accessed 31 July 2017).

Laser photocoagulation treatment of early age-related macular degeneration

Prophylactic laser photocoagulation has been proposed to prevent progression of early AMD.

Quantity and quality of research

Results

Key results from these studies can be seen in Table 10.

Figueroa and colleagues 1997239 included two cohorts of people: (1) bilateral confluent soft drusen and pigmentary changes (eyes were randomised to treatment or no treatment) and (2) high-risk drusen in one eye and choroidal neovascular membrane in fellow eye (where the eye with dry AMD was lasered). A green argon laser was used, and 0.1 seconds and 160-µm spot sizes. Light grey-white take was considered the end point for the treatments. Drusen disappeared in 45 out of 46 patients during an average of 3.5 months and untreated drusen (located far from laser scars) disappeared in 43 out of 46 patients during an average of 8.6 months (cohort not stated). Initial improvement in VA was seen in 33% of patients in cohort 1, but this was lost after 3 years’ follow-up in half of these. After an average of 3 years’ follow-up, 17% of eyes in cohort 1 and 31% of participants in cohort 2 had at least 1-line improvement in VA, compared with none of the untreated eyes. No statistical analyses were presented.

Guymer and colleagues 2014232 included 50 people with bilateral intermediate AMD, excluding those with evidence of GA or presence of CNV. They describe their study as a pilot. Only one eye was treated. They used an ultra-low energy nanosecond laser aiming at ‘retinal rejuvenation therapy’: 12 spots around the macula, 400-µm diameter spots, 3-nanosecond pulse length, 532-nm wavelength. This was estimated to give 1000 times less radiant exposure than conventional macula thermal lasers. At 12 months’ follow-up, a treatment effect was seen in both treated and untreated eyes; BCVA improved by at least 5 letters in 16% and 8% of eyes, and drusen area was reduced in 44% of treated and 22% of untreated eyes, but p-values were not reported.

Guymer’s proposed explanation for the improvement in untreated eyes is that laser to one eye exposes the drusen to the immune system. She suggests that the exposure starts the immune system cleaning up of drusen in both eyes. 240 Another possible explanation is that laser treatment breaks up the drusen and triggers an inflammatory response.

Ivandic and colleagues 2008237 included all stages of AMD (dry and wet forms with or without cataracts) and a VA ≤ 20/20. They applied low-level laser treatment: four sessions semi-conductor laser diode, continuous emission at 780 nm, spot diameter of 3 µm. They reported a statistically significant increase in VA by study end (not defined, assume 4 weeks, no data presented; p < 0.00001) for patients undergoing laser treatment, whereas there was no change in a small control group with no treatment.

In a retrospective pilot study, Luttrull and colleagues96 recruited 108 people (158 eyes) with high-risk AMD (presence of multiple large, diffuse, or bilateral macular drusen; macular pigment disturbance; extrafoveal or subfoveal GA; and/or CNV in the fellow eye). The laser used was diode micropulse laser at subthreshold levels. Follow-up was 1 month after one course of panmacular laser treatment. Snellen VA was unchanged (data not reported, p = 0.75, includes eight participants with inherited PR degeneration) but several secondary outcomes, pattern ERG (an early indicator of retinal dysfunction) and visual function (measured using central vision analyser and automated microperimetry) improved.

Huang and colleagues 2011223 allocated one eye to laser treatment and one eye to control in 10 people with bilateral soft drusen. Treatment was with argon green laser: 514 nm, 100 spots, 200-µm spot size, 0.1-second duration, at 55 m to 100mW, to give barely visible spots. After at least 8 years’ follow-up, soft drusen in the treated eyes were reduced (although new drusen had appeared), whereas soft drusen had increased ‘significantly’ in the untreated eyes (presented in figures only). BCVA did not reduce ‘significantly’ in either eye, and at 2-year follow-up there was no difference in retinal contrast sensitivity. No CNV occurred during the study.

Prahs et al. 236 treated one eye of each of six patients with bilateral GA with laser therapy (described as ‘selective retinal therapy’) while the other eye served as control. The laser used was neodymium-doped yttrium lithium fluoride (Nd:YLF), 527 nm, 200-ns pulse duration, 30 pulses at 100 Hz. After 1 year, the mean GA area and progression rate (mm² per year) were slightly higher in the treated eyes, but statistical analyses were not reported.

Merry and colleagues 2016238 included 42 eyes of 24 people with dry AMD, with AREDS grades 2 (21%), 3 (48%) or 4 (with GA but no CNV, 31%) and a BCVA letter score of 50 (logMAR 1.0, Snellen 20/200) or better. They used low-level laser therapy, with three wavelengths: yellow 590 nm, red 670 nm and near-infrared 790 nm. They describe their approach as ‘photobiomodulation’, thought to act by biostimulation of PRs, increasing blood flow and stimulating cellular functions. It differs from earlier forms of laser therapy that created thermal effects. NCT00940407

Three months after laser treatment in all affected eyes, there was an increase in mean BCVA letter score (5.14 letters; p < 0.001) and an increase in contrast sensitivity that was statistically significant at 3.0 and 6.0 (but not 1.5) cycles per degree (see Table 10). A reduction in drusen volume was also found (p = 0.021), but there was no difference in other outcomes. No new wet AMD or GA developed during the study.

An abstract from the RANZCO 2011 conference by Beaumont and colleagues241 reported a study of subthreshold laser photocoagulation of large drusen in AMD in 121 patients in which one eye was treated and the other acted as control. After a mean of 5.5 years’ follow-up, wet AMD had developed in 4% of treated eyes and 8% of untreated eyes. GA developed in 10% of treated eyes and 7% of control eyes. In eyes which developed neither wet AMD nor GA, drusen had resolved in 57% of treated eyes and partially resolved in 34%. In control eyes, most drusen were unchanged. However, the overall results did not reach statistical significance.

One small study by Scalinci and colleagues242 from Bologna, Milan and Rome reported that photobiomodulation with a device called mnemosline (Mnemosline, Vicenza, Italy), which provided pulsed light at 650 nm for 10 minutes twice a day, 5 days a week for 3 months, was associated with modest improvements in BCVA, pattern ERG and retinal sensitivity. Only an abstract is available and no statistical significance analysis is provided. The authors call for a larger trial.

Adverse events

Guymer and colleagues 2014232 reported dot haemorrhage in one patient. Three studies reported no AEs occurred. 96,236,237

Summary

The Cochrane review of RCTs with a low risk of bias concluded that although laser treatment can reduce drusen area, it is not associated with improved outcomes for patients. Seven additional primary studies were identified: two CCTs with a high risk of bias, two studies assessed as fair quality and three assessed as poor quality. They used different forms of laser treatment. Outcome measures and length of follow-up varied. There may be some improvement in VA but it is unclear if this is maintained. Drusen appear to be reduced. Overall, the effect of laser treatment is inconclusive based on the review and primary studies.

The results in the study by Guymer et al. 232 were intriguing. A similar phenomenon was reported by Figueroa239 – most un-lasered drusen in the treated eye, distant from laser scars, regressed.

Registered research studies

NCT01799564 has the title Micropulse Laser for Geographic Atrophy and is being carried out by Mones and colleagues in Barcelona. In 15 patients, one eye will be randomised to laser to receive subthreshold micropulse spots in healthy RPE close to areas of GA. Estimated completion date is June 2017.

NCT02725762 is the Study of Photobiomodulation to Treat Dry AMD (LIGHTSITE 1) in Toronto. A total of 30 patients will be randomised to laser therapy or sham therapy. The laser is the LumiThera LT 300 (LumiThera, Poulsbro, WA, USA). Patients will be treated three times a week for 3 weeks, repeated at 6 months.

NCT02800356 is the Subthreshold Laser Treatment for Reticular Pseudodrusen and Geographic Atrophy Secondary to AMD. This is a non-randomised pilot study by Querques and colleagues, using a 577 nm yellow Pascal synthesis laser from Topcon (Tokyo, Japan). It aims to recruit 20 patients and is due to finish at end of 2017.

NCT01790802: Laser Intervention in Early Age-Related Macular Degeneration Study (LEAD). The purpose of this RCT is to determine whether or not 2RT nanosecond laser therapy slows the progression to advanced age-related macular degeneration. It aims to recruit 240 patients with early AMD (no GA). The principal investigator is Robyn Guymer. Final data collection should be in June 2018.

NCT00000167 is a trial of low-intensity laser treatment supported by the National Eye Institute in the USA. It aimed to recruit 1052 patients, and was due to end in 2007.

NCT02569892 is a trial of subthreshold Pascal laser versus sham laser, being carried out in Stanford University and the Bascom Palmer Eye Institute, in 56 patients with large, high-risk drusen. The aim is to see whether or not laser treatment will reduce progression to GA or wet AMD. The trial is due to end in 2018.

Quantity and quality of research

All the publications come from the same group in Sienna, Italy.

Results

Borrelli and colleagues 2012248 included people with bilateral AMD, with > 10 large semi-soft and/or confluent drusen within 3 mm of centre of fovea in the study eye and BCVA between 20/32 and 20/125. After 12 months, no statistically significant difference in logMAR was reported between the treatment and control groups, but the proportion of people gaining 1 line of vision was reported to be 25% in the ozone group and zero in the control group (p < 0.05). It is not clear what change happened in the other 75% but it is reported that none of the ozone group lost ≥ 2 lines, so we could presume that they had no change or a loss of not more than 1 line. The control group had lost 0.3 ETDRS lines by 12 months whereas the ozone group gained 0.3 lines, giving a difference of 0.6 lines.

Participants in the study by Bocci 2011244 presented with dry AMD, most commonly with soft confluent drusen followed by GA. At 18 months’ follow-up, there was a slight improvement in VA in the treatment group and a slight deterioration in the control group, but a statistical comparison was not reported. Two-thirds of the treatment group had an improvement of > 2 ETDRS lines compared with less than one-third of the control group; this was described as statistically significant (p-value not reported).

Adverse events

Temporary face redness was experienced by 3%. 248 No AEs were reported in the study by Bocci. 244

Summary

Five non-systematic reviews, one RCT (low risk bias) and one CCT (high risk of bias) were identified. Both studies reported that higher proportions of people with ozone therapy gained letters of VA. All the publications came from the same group, so they reviewed their own studies.

There is a lack of convincing evidence for ozone therapy.

Quantity and quality of research

These studies were done in people with late advanced AMD, in contrast with most of the patients in the previous studies who had mainly early disease.

Results

The IMT-002 group (Hudson, 2006250 Boyer, 2015251 Hudson 2008252 and Lane 2006253) included people with bilateral, stable, central VA loss by untreatable end-stage AMD (GA, disciform scar or both); phakic with evidence of cataract in the study eye; a distance BCVA 20/80 to 20/800 (ETDRS); and at least a 5-letter improvement with an external telescope used for 3 days. The patients’ own lenses were removed (which would alone would be expected to improve vision) and replaced with the IMT-002 device (VisionCare Opthalmic Technologies, Saratoga, CA, USA). Compared with untreated fellow eyes, there was a statistically significant improvement in both near and distance mean BCVA at 12 months, with two-thirds of treated eyes gaining ≥ 3 lines of distance vision versus 12.5% of untreated eyes (p < 0.0001), and two-thirds of treated eyes gaining ≥ 3 lines of near vision versus one-third of untreated eyes (p < 0.0001) (Table 12). At 24 months, 0.6% and 7.5% of implanted and untreated eyes, respectively, had lost ≥ 3 lines of vision (p = 0.013), and all categories of gains and losses except gain of ≥ 6 lines for BCVA were statistically significant between eyes in favour of the study eye (presented in figure only). Statistically significant improvements from baseline were found on the NEIVFQ-25 and activities of daily living scale.

Qureshi and colleagues 2015254 implanted telescopes in 18 eyes of 12 participants. Inclusion criteria were bilateral, intermediate or advanced dry AMD with central scotomata, minimal cataract or pseudophakia, Snellen BCVA < 0.25 and improvement with an external telescope. A 67% improvement in VA and a 50% improvement in corrected near VA was found after 4 months, with improvement of visual impairment classification in 11 (61%) eyes (see Table 12). One eye deteriorated from severe to profound, and six eyes were unchanged in terms of classification.

Adverse events

Adverse events are summarised in Table 13 and Report Supplementary Material 2. Hudson and colleagues250–253 stated that no retinal detachments, CNV or visually significant cases of posterior capsule opacification occurred during the 2-year follow-up. However, ocular AEs and complications occurred, including inflammatory deposits, increased IOP and corneal oedema. Qureshi and colleagues254 reported no cases of clinical corneal decompensation, signs of cystoid macular oedema or active CNV.

A good-quality cost-effectiveness analysis by Brown et al. 260 concluded that when using US costs (not necessarily applicable to other countries), the IMT was cost-effective, improving quality of life by 12.5%.

The CentraSight telescope implant261 (VisionCare Opthalmic Technologies, Saratoga, CA, USA) is now being used in Manchester Royal Eye Hospital and appears to be the one used in the IMT-200 trial.

Summary

One large CCT and one small case series were included. The CCT found improvement in VA in implanted eyes compared with untreated eyes, although some AEs occurred. Compared with baseline, improvements in quality of life were found. Improvements in VA were also found by the case series but no statistical analyses were presented. In both cases, some improvement could have been due to cataract removal.

A trial is under way in the UK [the Efficacy of the telescopic mirror implant for age-related macular degeneration: the MIRROR Trial, Efficacy and Mechanism Evaluation (EME) project 13/160/03]. 262,263

A trial of the CentraSight telescope, NCT03011554, is being run in the USA in people with advanced AMD who have had cataracts removed and replaced with an IOL. The IOL will be exchanged for the telescope lens.

Qureshi and colleagues264 have reported a case series of 244 eyes into which the Eyemax IOL had been implanted, but the patients had a mix of dry and wet AMD, and their results are not reported separately, making the study an exclusion for our purposes.

Reviews

We found many reviews: > 30 in the last 3 years alone. None appeared to be true systematic reviews but given the very small number of actual primary studies, it is not difficult to be comprehensive. There appear to be rather more reviews than treated patients.

A review by Ramsden et al. 268 from London, identified three teams around the world doing cell transplants: Ocata from California, the Riken Centre in Japan in collaboration with Kobe City General Hospital (one patient at time of article, but now at least two, with five candidates enrolled), and the London Project to Cure Blindness (one patient at time of article, although with wet AMD, but another since then, also with wet AMD). 269

Ramsden et al. 268 summarise some of the problems that are being encountered, including:

• having to deal with multiple regulatory bodies

• the need to find ways to scale up production of cells

• limitation of cell transplant to advanced disease until the balance of risks from cell therapy are known

• in this advanced group, problems of assessing benefit when there is already visual loss

• determining the optimal immunosuppression regimens.

They identify three issues when assessing success:

1. safety

2. cell survival

3. visual outcome.

Dalkara and colleagues134 regard human IPSCs as an unlimited source of cells for transplantation, and report that at least 15 trials were under way in late 2015, of which 10 are in dry AMD or in mixed groups including dry AMD.

Zarbin270 gives a good account of ‘challenges to clinical translation’, which include:

• Cell production.

• Cell delivery, in suspensions or sheets on scaffolds, but with many different materials being studied as potential scaffolds.

• Cell survival and differentiation. Survival may be impaired in areas of GA because of changes in supporting structures such as Bruch’s membrane.

• Immune reactions. Although the subretinal space is less accessible to the immune system, this privilege is not complete and immune reactions can occur. Not all patients (most of whom are elderly) can tolerate immunosuppressive drugs.

• A theoretical risk of mutation to cancerous cells.

Results

The Schwartz study (Table 14) included people with advanced atrophic AMD with > 250 µm of GA involving the central fovea. 98,123 After 12 months, VA had improved in 3 of 7 eyes and was stable in 3 of 7 eyes. Quality of life, assessed by the NEIVFQ-25 improved for general vision, peripheral vision, near activities, distance activities and mental health. The possibility of a placebo effect on the VFQ-25 scores cannot be excluded.

In the study by Song et al. 271 improvements were seen in BCVA at 12 months in all patients in the study eyes, ranging from 1 to 19 letters improvement. Fellow eyes deteriorated in the two dry AMD patients but improved in the two patients with STGD. This may be chance because of the very small sample size.

Quantity and quality of research

Results

In people with BCVA 20/50 to 20/200 and category 3 or 4 AMD GA, Zhang and colleagues273 (Table 15) found a higher proportion of patients with stable VA after the implant, but this was not statistically significant. A subgroup analysis in patients with baseline BCVA 20/63 or better did show a statistically significant difference with none of the 10 in the high-dose group losing > 15 letters, compared with five out of nine in a combined sham and low-dose group. There was no significant difference in area of GA, but numbers were small and duration short. Retinal thickness increased in the high-dose group. Other outcomes can be seen in Report Supplementary Material 3.

Summary

This small RCT with an unclear risk of bias found no statistically significant overall differences in VA stabilisation or area of GA between people treated with NT-501 and those not treated. However, it was a small proof of concept trial. No further trials of NT-501 are registered with ClinicalTrials.gov (accessed 18 June 2017).

Quantity and quality of research

Results

Lipoproteins and age-related macular degeneration

High-density lipoprotein cholesterol is usually regarded as ‘good cholesterol’ because of an association with reduced cardiovascular risk. The picture in AMD is mixed. Dashti and colleagues299 identified 20 studies of AMD risk and HDL cholesterol: seven reported higher risk with higher HDL, four showed reduced risk, eight showed no effect and one showed a mixed pattern.

In a more recent review, Wang and colleagues300 included 19 studies. They carried out meta-analysis and concluded that higher HDL cholesterol was associated with an increased risk of early AMD, but not of wet AMD or GA. High LDL-C was associated with a lower risk, but only just (RR 0.93, 95% CI 0.88 to 0.99).

In a case–control study with 82 patients with AMD (early or late, with late including both wet AMD or GA), Colak and colleagues,301 from Belgrade, found no association between HDL cholesterol and AMD, but did find higher total and LDL cholesterols in the AMD group (total 6.25 mmol/l vs. 5.57 mmol/l, p < 0.001; LDL 3.99 mmol/l vs. 3.60 mmol/l, p = 0.018). However, when they examined the subfractions of HDL cholesterol, they found differences in the HLD₃ subfraction in men but not in women. The HLD₃ sub fraction is not regarded as ‘good HDL’.

Therefore, looking just at HDL and LDL may be too broad in AMD, and we may need to explore gender differences in risks. To complicate things further, it has been suggested that the lipid deposition in Bruch’s membrane and drusen may not all be derived from plasma sources such as LDL cholesterol – some may be produced in the RPE itself. 302

Problems with statin studies

An almost universal problem with the observational studies is that they record use of any statin, but do not analyse by individual statin or dose, thereby lumping low-potency and high-potency statins together. The older studies probably reflect the use of older, less potent statins, rather than the more recent high-potency ones, such as atorvastatin 80 mg or rosuvastatin. There are other problems. One is that ‘use’ may mean prescription, not consumption. Another is that statin use may only have an effect if lipid levels are reduced and we do not know how great a reduction is required to affect AMD, if there is an effect.

One study by Shalev et al. 303 did subdivide statins into:

• low efficacy – fluvastatin 40 mg or less, pravastatin 40 mg or less, simvastatin 10 mg or less, cerivastatin 0.2 mg, lovastatin 40 mg or less

• moderate efficacy – fluvastatin 80 mg, rosuvastatin < 10 mg, simvastatin 20 or 40 mg, atorvastatin 10 mg

• high efficacy – atorvastatin 20 mg or more, simvastatin 80 mg, lovostatin 80 mg, rosuvastatin > 10 mg, pravastatin 80 mg.

These bands were based on LDL lowering. Note that high-dose statins are now regarded as doses such as atorvastatin 80 mg daily.

Shalev et al. 303 noted that only about one-fifth of patients took statins on ≥ 90% of days. Only 19% were prescribed a high-efficacy dose.

One problem with some older negative studies is that they may not have enough patients on statins. Smeeth and colleagues304 used UK data from General Practice Research Database from the years 1987 to 2002, but only 2.1% of people on the database had ever been prescribed a statin. In a later study305 using The Health Improvement Network (THIN) data, 1996–2006, they found much larger numbers on statins and were able to match 129,288 statins users with > 600,000 controls who had not been prescribed statins. They also provide data on which statins were used, but not dosages. Simvastatin was used by 39% of people on the database, atorvastatin by 21% patients and more than one statin by 33% patients. They reported that statin use was associated, after adjustment, with HR of 1.17 (95% CI 1.00 to 1.38) for any AMD (as recorded on the THIN database). 302 Another problem is that people treated with statins are at high risk of cardiovascular disease and may have other confounding factors that cannot be entirely adjusted for.

Adverse events

Adverse events were recorded in the two intervention studies (Table 17). Simvastatin was well tolerated in the RCT by Guymer and colleagues283 with major illnesses reported in seven participants in the statin group and 15 in the control group. Muscle aches were reported by two recruits on placebo and five on simvastatin (see Report Supplementary Material 4 for full details.) In the before-and-after study by Vavvas and colleagues,286 three participants withdrew from study due to AEs.

Summary

The evidence from observational studies is rather mixed and probably best regarded as inconclusive overall. Some studies have attempted to adjust for confounding variables, for example by propensity scoring. Most studies gave no details of which statins were used or dosages. Patients on statins are more likely to have had cardiovascular disease, which is associated with an increased risk of AMD.

The trial by Guymer and colleagues,283 using 40 mg of simvastatin, did show benefit in terms of reduced progression in patients with non-advanced AMD. The results in the study of high-dose atorvastatin by Vavvas and colleagues286 were more impressive and visual benefits were reported, but there were no controls.

Conclusion

We recommend a trial of high efficacy statins in early and intermediate AMD.

A trial may be getting under way in Spain, called Statins4 drusen. 306

Mechanism of action

Detection of light in the PRs (rods and cones) begins with the transformation of 11-cis-retinal (a photosensitive molecule derived from vitamin A, attached to an opsin protein) into all-trans-retinal. This molecule is not light sensitive so in order to replenish the 11-cis-retinal stocks, a recycling process is required. This process is called the visual cycle and is illustrated in Figure 1.

FIGURE 1.

Visual cycle.

In AMD, this cycle malfunctions, and some all-trans-retinal goes down a different pathway into the A2E dimer, which is a major component of the toxic lipofuscin deposition. A2E is formed of two molecules of all-trans vitamin A (bisretinoid) and one molecule of ethanoloamine. 307 The aim of visual cycle inhibitors (also called modulators) is to reduce the accumulation of A2E.

Emixustat

Emixustat is an oral drug that inhibits RPE65, which is found only in the eye. So adverse effects elsewhere are unlikely.

One RCT was found. Dugel and colleagues308 included adults with a clinical diagnosis of GA, and BCVA 20/400 or better in the study eye, in a dose-ranging, Phase II RCT (NCT01002950) in 15 centres in the USA. The trial had an unclear risk of selection bias, although random generation was adequate. A total of 72 participants were randomised to one of five doses of oral emixustat or placebo for 90 days. The 7-mg and 10-mg groups were discontinued by the sponsor at an early stage owing to the frequency of adverse effects. Change in GA area is summarised in Table 18, but no statistical comparisons were presented. Two participants in the emixustat cohorts (one in the 7 mg dose arm, one in the 5 mg dose one) had a decrease in VA ≥ 15 letters, compared with none of the placebo group.

Another trial has been registered, testing three doses of emixustat (2.5 mg, 5 mg and 10 mg) and placebo, but no results have yet been posted (NCT01802866). The primary outcome was area of GA. This is also registered as EUCTR2012-004952-12-DE.

Adverse events

Overall, there were 29 discontinuations due to ocular AEs (emixustat, n = 23; placebo, n = 6) in the study by Dugel and colleagues. 308 Serious AEs were reported by three participants in the emixustat groups (one exacerbation of chronic obstructive pulmonary disease and two with chromatopsia, a disturbance of colour vision, sufficient to interfere with driving) and none of the placebo group. At least one ocular event was experienced by 93% of the combined emixustat groups and 28% of the placebo group. The adverse ocular events included chromatopsia (57% emixustat vs. 17% placebo), delayed dark adaptation (48% vs. 6%), visual impairment (26% vs. 6%), reduced VA (11% vs. 0%) and blurred vision (15% vs. 6%).

Dugel and colleagues308 considered that the adverse effects were not sufficient to prevent the drug being trialled in a longer RCT with progression of GA and VA as the outcomes.

Another study of emixustat has been reported in a EURORETINA 2016 conference abstract to have shown no difference in GA growth rates between drug and placebo. 309

Fenretinide

Fenretinide inhibits bisretinoid accumulation by binding with retinol binding protein 4 (RBP4).

Mechanism of action

The complement system is part of the body’s immune system. There are three main complement pathways: classic, alternative and mannose-binding pathway. Complement components have been found in drusen and in AMD, and it is thought that the complement system is inappropriately activated. The aim of complement inhibitors is to reduce the activity.

One drug, lampalizumab, has been shown to have some effect. For completeness, a brief account is given here, but as trials are already under way, it is not anticipated that the HTA programme would commission any further trials.

Lampalizumab is an antigen-binding fragment targeting complement factor D, which is a key component of the alternative complement pathway. The MAHALO Phase II trial317 recruited 129 patients randomised to sham injections, or to two lampalizumab regimens, 10 mg once a month or twice a month, by intravitreal injection. The primary outcome measure was spread of GA, measured by AF, and follow-up was for 18 months. Almost one-quarter of patients dropped out. There was a 20% reduction in GA area on lampalizumab compared with sham. In a subgroup with the biomarker complement factor 1, the reduction was 44%.

Two very large trials are under way, CHROMA and SPECTRI.

SPECTRI is NCT02247531 and will compare lampalizumab 10 mg given every 4 weeks and every 6 weeks with sham injections. It aims to recruit 936 people in 160 sites worldwide (so only about six per site, which suggests a marketing element), although most are in the USA. There are 14 UK sites. Data collection is due to end in November 2017.

CHROMA (NCT02247479) aims to recruit 936 patients in 144 sites, again mainly USA but four in the UK. The design is the same as SPECTRI. Data collection is due to end in September 2018.

OMASPECT (NCT024745119) will recruit patients coming out of these two RCTs.

There is also another smaller trial, NCT02288559, that will compare two doses, 10 mg every 2 weeks versus 10 mg every 4 weeks, with sham injections, in 96 patients with GA in 37 sites in the USA. Final data collection is due in 2017.

All these trials are sponsored by Genentech-Roche (South San Francisco, CA, USA).

A trial without a NCT number is registered on the WHO trials portal. It looks very similar to NCT0268658 but has a slightly different start date and a recruitment target of 900. 318

NCT02515942 is a trial of the combination of CLG561, an antiproperdin drug, with tesidolumab (formerly LFG316), a C5 blocker, in GA. CLG561 was not effective on its own (NCT01527500). The trial is due to end in November 2017.

NCT02503333 is the FILLY trial of the complement inhibitor, APL-2 from Apellis Pharmaceuticals (Crestwood, KY, USA). Early results have been posted on the Apellis website (www.apellis.com; accessed 26 December 2017). In this Phase 2 study, APL-2 slowed the growth of GA by 29% compared with sham injections, but with a greater reduction becoming apparent in later months. However, in a subgroup with wet AMD in one eye, there was a higher incidence of wet AMD.

Quantity and quality of research

Results

In the Phase II COMPLETE Study, Yehoshua and colleagues323 included people with total GA area of 1.25–18 mm² and VA 20/63 or better. A total of 30 participants were randomised to treatment with i.v. eculizimab, (high or low dose) or placebo for 24 weeks, with one eye selected as the study eye. There was no difference in change from baseline for GA or VA between eculizumab and placebo at 12 months’ follow-up (see Table 20). Yehoshua and colleagues323 suggest several possible reasons for the lack of effect. First, it may just be that the drug has no effect. Second, it may have been that the effect was smaller than detectable given the power of the trial, which could detect a reduction in GA growth rate down to 55%. It was powered to detect a 75% reduction. Third, the drug might have had more effect if given intravitreally.

Adverse events

Yehoshua et al. 323 reported no AEs.

Summary

The Cochrane review319 (last updated January 2014) found no published trials on the use of complement inhibitors in AMD.

One small RCT323 with an unclear risk of bias found no significant improvement with eculizumab. It seems unlikely that further research on eculizumab would be a priority.

Lampalizumab looks more promising but a considerable amount of research is under way.

Mechanism of action

Brilliant and colleagues324 report their discovery of a protein receptor, GPR143, that is activated by levodopa (L-dopa). This receptor is found in the RPE and their hypothesis is that GPR143 signalling might protect people from AMD, possibly by altering the balance between pigment epithelium derived factor and VEGF.

Quantity and quality of research

One study was identified. Brilliant and colleagues324 carried out a large retrospective cohort study of three registries with a total of 15,252,958 people in the USA. Data on exposure were according to L-dopa prescriptions, but the duration of follow-up was not reported. The study was funded by non-commercial grants, including two from NIH. Mean age was reported only according to L-dopa treatment and AMD diagnosis with and without L-dopa treatment, for each of the registries separately (Table 21). No baseline characteristics of the participants were reported.

Brilliant and colleagues324 found that age of AMD onset was significantly later in patients who used L-dopa than in non-users (79.3 years vs. 71.2–71.3 years; p < 0.01). L-dopa users were significantly less likely to have a diagnosis of AMD (OR 0.78, 95% CI 0.76 to 0.80; p < 0.001) after controlling for age and gender.

Adverse events were not reported.

The authors consider that the results justify clinical trials.

Summary

This huge study found that people who used L-dopa were less likely to have an AMD diagnosis and that the age of AMD onset was delayed, suggesting that L-dopa protects against AMD.

The study does not tell us whether L- DOPA would be useful in treating dry AMD. It is a very cheap drug with few adverse effects.

Studies

The pilot study by Ladewig et al. 326 recruited a series of 11 patients and compared them with 10 untreated patients, all with dry AMD (≥ 10 soft and/or hard drusen, early GA and pigment epithelial detachment without indications of CNV and BCVA ≥ 0.2 and ≤ 0.8). The distributions of stages of AMD are reported to be similar but details are not given. The study was graded as poor quality (see Report Supplementary Material 4). Participants were treated with daily i.v. infusion of 60 µg alprostadil for 3 weeks. The duration of follow-up was 6 months. Funding was from an independent source.

Visual acuity at 6 months improved in the alprostadil-treated group and deteriorated in the non-treated group; however, no analysis of the difference between groups was reported (Table 22). Other outcomes were reported only for the treated group.

The RCT by Augustin et al. 325 was carried out at six sites in Germany and Austria. The trial (NCT00619229) was sponsored by UCB Pharma SA (Brussels, Belgium) and two authors were from the company. Augustin et al. 325 recruited 36 people with dry AMD with hard drusen and possible early GA limited to the perifoveal area in one eye and VA between 0.2 and 0.7 logMAR. They received daily i.v. infusion of 60 µg of alprostadil for 3 weeks. The controls received a placebo infusion. Follow-up was for 6 months. There was an unclear risk of selection bias. Mean age was 74 years and half of the recruits were male.

The study was stopped prematurely after the first interim analysis, which led to a need to recalculate the sample size, which showed a need for greater numbers if power was to be achieved. In addition, recruitment was slow. So results were based on interim data because the study was stopped early. Augustin et al. 325 state that only exploratory results can be presented.

The BCVA (using ETDRS charts) improved in the alprostadil-treated group, with benefit seen at end of treatment and at 3 and 6 months, but with greatest improvement at 6 months (see Table 22). BCVA deteriorated in the placebo group, but no analysis of the difference between groups was reported (see Table 22). Progression of dry AMD, stabilisation of dry AMD, contrast sensitivity and colour vision appeared to be similar in each group. No participants in either group developed nAMD.

Adverse events

There were no drug-related AEs in the pilot study. 327 In the RCT325 there were fewer AEs in the alprostadil group (11.1%) than the placebo group (33.3%). No serious AEs were experienced.

Summary

The trial did suggest some benefit in BCVA but this was not statistically significant with the limited numbers recruited (no p-value reported but the CIs overlap), and there was no improvement in progression, although the duration may have been too short. Whether daily infusion for 3 weeks would be acceptable in wider populations is not known, and nor do we know how long benefit would last. The value of prostaglandin infusion is therefore unproven.

Mechanism of action

Dorzolomide (Trusopt, Chibret, Munich, Germany) is a carbonic acid or carboanhydrase inhibitor. Such agents have been used in glaucoma and the rationale of testing in AMD is that in addition to lowering IOP, they also improve choroidal perfusion.

Quantity and quality of research

Only one study was identified. A placebo-controlled RCT by Remky and colleagues328 from Germany included 20 participants in the dorzolamide group and 20 in the placebo group. Treatment with a three-times-daily eye drop (dorzolamide 0.2% eye drop, placebo an artificial tear drop) lasted for 12 weeks with follow-up immediately afterwards. The study had an unclear risk of selection bias (see Report Supplementary Material 4). Participants were aged 70 years and 60–70% were male. Baseline BCVA was 0.12 to 0.13 logMAR. The funding source for the study was not reported.

Results

Remky and colleagues328 included people with AMD with VA > 0.4 (20/50). The eye with better VA was selected. At 12 weeks, VA was the same in each groups (Table 23). Pericentral visual field analysis showed an increased retinal sensitivity (1.55 dB, comparison from baseline p = 0.04) in the dorzolomide group but no significant change in the placebo group (+0.58 dB; p = 0.1). Other outcomes were also similar between groups (see Report Supplementary Material 4). The study was small and short, so power was limited.

Adverse events

There were no severe AEs in the study. Minor AEs were reported (see Report Supplementary Material 4).

Summary

This pilot RCT with an unclear risk of bias showed no significant differences in VA or other outcomes between those treated with dorzolamide and those treated with placebo, but the duration of treatment was very short and the power of the study was low. Few AEs were reported. This group of drugs have been used for many years with no serious adverse effects.

Quantity and quality of research

Two publications from Landa et al. 329,330 The first study was a pilot CCT that included 14 participants treated with glatiramer acetate (seven participants) or placebo (seven participants). The second study was a pilot RCT that included six participants (four participants treated with glatiramer acetate and two participants treated with a placebo). The studies were conducted in the USA. Participants were treated for 12 weeks in both studies and follow-up was immediate. The intervention was by subcutaneous injection (for dosing details see Report Supplementary Material 4). The studies had an unclear risk of selection bias. Ages, male sex and baseline BCVA were not reported. The funding source was not reported.

Results

Landa and colleagues330 included people with dry AMD (no further inclusion criteria reported). At 12 weeks the CCT reported a statistically significant difference in the proportion of convex drusen disappeared or shrank (Table 24), but no difference in the proportion of total drusen or concave drusen that disappeared or shrank. In the pilot RCT the change in drusen area was reported. This appeared to be lower in the glatiramer acetate treated people than the placebo controls but no statistical analysis was undertaken.

Adverse events

The study did not report AEs. 330

Summary

Two linked studies with unclear risk of bias were included. The studies were small and of short duration, and the outcome measure used was drusen area, which was reported to have reduced. The main interest of the authors seems to have been to test an outcome measure, using high-resolution spectral domain OCT, that could predict response. The evidence is too sparse to come to any conclusions on efficacy of glatarimer.

Quantity and quality of research

One pilot RCT92 conducted in the USA with 10 participants with bilateral GA was included. Treatment and follow-up were 2 years duration and one eye of each participant was randomised to treatment or observation. The trial had unclear risk of selection bias. The funding source was non-commercial. The mean age of participants was 77 years and 40% were men. Mean baseline BCVA was 46.1–55.7 letters.

Results

Wong and colleagues92 included participants with bilateral GA. A statistically significant difference between treated eyes and untreated eyes was found for change in BCVA at 2 years (p = 0.026), but no significant difference was found for contrast sensitivity, GA area or drusen area (Table 25). No eyes progressed to nAMD.

Adverse events

No serious AEs occurred. There were 32 grade 1 and four grade 2 events (see Report Supplementary Material 4).

Summary

One small RCT with an unclear risk of bias found a statistically significant improvement in BCVA in eyes treated with OT-551 for 2 years compared with untreated eyes. There was no significant difference in other outcomes. The authors conclude that OT 551 has limited or no benefit in GA. A larger trial was carried out with 137 subjects, using the eye drops on one eye four times a day for up to 2 years, but we found no publications other than a conference abstract331 reporting no benefit.

Quantity and quality of research

Two small RCTs, Wong 2013332 and Petrou 201594 conducted by the same group in the USA were included, with a total of 17 participants (total 34 eyes, 17 eyes treated). Randomisation was by eye, with the other eye acting as control. The first study used subconjunctival sirolimus, the second intravitreal. The studies had 12 months’ and 24 months’ follow-up, respectively. Both studies had an unclear risk of selection bias, although in one332 the random sequence was generated by a computer algorithm. Both studies received non-commercial funding, although the drug was donated by a commercial entity. Mean age was 74–78 years and the majority of participants were men. Visual baseline acuity was reported by both studies (Table 26) (NCT00766649 and NCT01445548).

Results

Both studies had similar eligibility criteria: bilateral GA with area of a half or more of a disc area, one or more large drusen (≥ 125 µm) in each eye, and BCVA 20/20 to 20/400 in each eye. Eyes from each participant were randomised to receive sirolimus injection or no treatment. Wong and colleagues333 administered sirolimus by subconjunctival injection every 3 months for 2 years. Petrou and colleagues94 administered sirolimus every 2 months for 1 year.

Both studies found that treated eyes had a statistically significantly greater decrease in BCVA compared with untreated eyes (p = 0.03,333 p = 0.01394) and that a higher proportion of treated eyes lost ≥ 10 lines of vision (50% vs. 12.5%,332 80% vs. 20%;94 p-value not reported). A non-significant greater increase in GA area in treated eyes was also found in both studies. No eyes developed CNV during either study.

Adverse events

Wong and colleagues332 reported mild to moderate ocular AEs, including small haemorrhages, raised IOP, and one death unrelated to the study medication. Petrou and colleagues94 reported accelerated retinal thinning in two treated eyes and treatment was suspended.

A third trial, NCT01675947, was terminated in 2014.

Summary

Two small RCTs with an unclear risk of bias found no benefits, but rather a worsening of vision in eyes treated with sirolimus injection. Serious AEs occurred in some participants. There seems to be no place for sirolimus in AMD.

Quantity and quality of research

One prospective cohort study conducted on 475 participants (prednisolone, n = 400; control, n = 75) was identified. 334 Participants from Croatia were given either peribulbar injections of prednisolone 5 mg for 5 days, or received multivitamins. Allocation was temporal – the first 400 patients received the steroid. Follow-up was at 6 months. The study was of poor quality (see Report Supplementary Material 4). Participants were aged between 39–80 years, no further baseline characteristics were provided. The funding source was not reported. The study was published in the journal of the Croatian Anthropological Society.

Results

Vojniković and colleagues334 included people with dry AMD but no other inclusion criteria were reported. At 6 months, peripheral visual field had improved by 10% to 25% in participants in the prednisolone group but there was no improvement in the control participants (Table 27). Central visual field improved by 5% to 20% in the prednisolone group (only a subgroup of controls were reported, assume no improvement was seen in the remaining participants). No statistical analyses were reported. AEs were not reported.

Summary

One study of poor quality provided sparse evidence on the effects of prednisolone on dry AMD.

Quantity and quality of research

One three-arm RCT was identified, the Geographic Atrophy Treatment Evaluation (GATE) trial. 190 The study was undertaken in 48 sites in a number of countries including the USA, Germany, Canada and the UK (see Report Supplementary Material 4). Participants were allocated to eye drops of 1% tandospirone (n = 252), 1.75% tandospirone (n = 259) or placebo (n = 261). Treatment was given twice daily for 24 months. Participant follow-up was at 30 months. The RCT had an unclear risk of selection bias. The mean age of participants was around 78 years and between 37% and 48% of participants were male. Baseline BCVA was not reported. The study was sponsored by Alcon Research (NCT00890097 and EUCTR2008-007705–37-DE).

Results

Jaffe and colleagues190 (Table 28) included people with GA secondary to dry AMD and no evidence of CNV. BCVA was required to be at least 35 letters. At 30 months VA was reduced in all groups, but the difference between groups was not statistically significant (see Table 28). The primary outcome for the study was the annualised lesion growth rate, this was not statistically significantly different between the active groups and the placebo.

Adverse events

The study reported similar rates of any ocular AEs across the three treatment groups. There were slightly higher rates of serious ocular AEs in the placebo group (Table 29 and see Report Supplementary Material 4).

Summary

This RCT found no benefit on BCVA or GA growth from tandospirone eye drops in in people with GA. AEs were similar in the tandospirone groups and the placebo group.

Quantity and quality of research

One placebo-controlled RCT by Cohen et al. 335 from France (324 centres), Belgium (6 centres) and Spain (11 centres) compared 594 participants given trimetazidine 35 mg with 598 given a matched placebo. Treatment was one tablet twice daily for a mean of 37.6 months. Follow-up was a minimum of 3 years. The study had a low risk of selection bias (see Report Supplementary Material 4). Drop-out rates were 23% in the trimetazidine group and 27% in the placebo arm. Participants had a mean age of 73.5 years and 38% were male. Baseline distance VA was at least 0.5 (20/40) in 91.5% of the trimetazidine treated participants and 93% in the placebo group. The study was sponsored by Servier Laboratories, Suresnes, France.

Results

Cohen and colleagues335 included people with AMD and unilateral CNV, and the unaffected eye was the study eye. At follow-up, CNV incidence was similar between groups (p = 0.78) (Table 30). Five-year incidence of CNV, atrophy of greater than one-third disc diameters and atrophy of greater than one-third disc diameters were similar between groups. In atrophy greater than one-third disc diameters, the trimetazidine group did slightly better – incidences per 100 patient-years 5.11 on drug versus 6.45 on placebo; HR 0.76, 95% CI 0.56 to 1.02. In some subgroups (prespecified) the drug group did better, such as the under 75s where HR for atrophy was 0.57 (95% CI 0.38 to 0.88) based on 11% and 17% developing atrophy greater than one-third disc diameters. VA was not reported.

Adverse events

Adverse events including eye disorders were similar between groups with any AEs experienced in 75% of the trimetazidine group and 79% of the placebo group.

Summary

This RCT was essentially negative overall, with no differences in the incidence of CNV and atrophy between those treated with trimetazidine and those treated with placebo.

Further research does not seem to be indicated.

Quantity and quality of research

One study was identified. Kaiser and colleagues from Switzerland336 carried out a placebo-controlled pilot RCT with nine participants in the visaline group and 11 in the placebo group. Treatment was with two tablets twice daily, 5 days per week (buphenine HCI 1.5 mg, beta-carotine 10 mg, tocopherol acetate 10 mg and ascorbic acid 50 mg or placebo) for 6 months with follow-up immediately afterwards. The study had an unclear risk of selection bias (see Report Supplementary Material 4). Participants were aged 73 years and 55% of treated participants and 9% (1 of 9) of placebo participants were male. Baseline near VA was 0.45 – 0.57. The funding source for the study was not reported.

Results

Kaiser and colleagues 1995336 included people with early AMD with BCVA between 20/100 and 20/25. The eye with better VA was selected if both eyes were affected. At 6 months, near and far VA were similar in each group as shown in Table 31. The proportions with improved visual function (not defined) was greater in the visaline group than the placebo group (44% vs. 27%); however, there appears to be some missing data (see Report Supplementary Material 4 for unchanged and worsened categories). Other outcomes were similar between groups.

Adverse events

There were no AEs.

Summary

This RCT showed no difference in VA or other outcomes between those treated with visaline and those treated with placebo.

Quantity and quality of research

One small before-and-after study of ranibizumab by Gallego-Pinazo and colleagues338 was identified. The study was described as a pilot. It was conducted in Spain. Six participants received intravitreal ranibizumab 0.5 mg/0.05 ml at least once. Retreatment was permitted if there was persistence or recurrence of elevation of the RPE contour on OCT, or if retinal fluid was seen on OCT, or if ≥ 5 letters were lost. The mean number of injections was two and the range was one to five. Participants were then followed up at 12 months. The study was of fair quality apart from the very small size. The mean age of participants was 69 years and 33% were male. Baseline VA was 0.40 (decimal ETDRS equivalent). The study funder was not reported.

Results

Gallego-Pinazo and colleagues338 included people with BCVA < 20/30 and DPED from AMD. At 12 months the mean BCVA had improved (p = 0.046, Table 32) and the study found that 33.3% of patients gained between 19 and 21 letters, with no participants experiencing loss of BCVA. There was a decrease in central macular thickness but this was not statistically significant.

Adverse events

The study did not report AEs.

Summary

In a very small before-and-after study of fair quality, a significant improvement on BCVA was seen at 12 months in people with DPEDs. AEs were not reported. The authors recommend a long-term randomised trial. If there is any prospect of benefit in dry AMD, the manufacturers will no doubt pursue it vigorously.

AREDS 2

The AREDS 2 trial30 assessed the value of adding lutein and zeaxanthin supplements to the original AREDS formula. It recruited 4203 people aged 50 to 85 years, at high risk of progression to advanced AMD, with either bilateral large drusen or non-foveal GA, or large drusen or non-foveal GA in one eye and advanced AMD in the other eye, in 82 clinical centres. The primary outcome was the development of advanced AMD. Secondary outcomes included three or more lines of VA loss or treatment for wet AMD.

The primary randomisation included arms in which supplements of the long-chain omega-3 polyunsaturated fatty acids, DHA and EPA were given. This results in four arms:

1. placebo

2. lutein 10 mg and zeaxanthin 2 mg

3. DHA 350 mg and EPA 650 mg

4. lutein, zeaxanthin, DHA and EPA.

In the second randomisation, recruits (27% declined) were allocated to the above with:

• the original AREDS 1 supplement (vitamins C and E, beta-carotene and zinc)

• AREDS 1 without beta-carotene

• AREDS 1 with reduced zinc (25 mg instead of 80 mg)

• AREDS 1 with no beta-carotene and reduced zinc.

The quality of the AREDS 2 trial30,348 was good (Table 35).

The median follow-up was 4.9 years. The mean age at recruitment was 73 years and 9% died during follow-up. Other loss to follow-up was good at only 3%. The recruits were a somewhat select group of highly educated and well-nourished people. Over 65% were educated beyond high-school level, and only 6.7% were smokers. 30

The main results were that:

• There were no differences among the primary trial arms in progression to advanced AMD. For lutein and zeaxanthin compared with placebo, HR was 0.90 (99% CI 0.76 to 1.07).

• Neither lowering the zinc dose nor omitting beta-carotene affected progression to advanced AMD.

• Lutein and zeaxanthin supplements were more effective than beta-carotene – HR 0.82 (95% CI 0.69 to 0.96) for progression to advanced AMD, HR 0.78 (95% CI 0.64 to 0.94) for development of wet AMD and 0.94 (95% CI 0.70 to 1.26) for central GA.

• Analysis by quintiles of baseline lutein and zeaxanthin, versus no lutein and zeaxanthin, showed a significant reduction in progression to advanced AMD only in the lowest quintile; HR 0.74 (95% CI 0.59 to 0.94).

• A post hoc analysis across all arms that compared taking supplements with lutein and zeaxanthin versus those without, showed HR of 0.90 (95% CI 0.82 to 0.99; p = 0.04). The effect was seen more on wet AMD (HR 0.89, 95% CI 0.79 to 1.00) than GA (HR 0.92, 95% CI 0.78 to 1.07).

• Adding DHA and EPA to the AREDS 1 formula conferred no benefit.

• The reduction in zinc dose made no difference.

• The VA outcome of ≥ 15 letters lost did not differ among arms.

One finding was that serum levels of lutein and zeaxanthin were lower when recruits also took beta-carotene, probably because beta-carotene competes with lutein and zeaxanthin for absorption.

One weakness of the study was that recruits had a much higher baseline intake of lutein and zeaxanthin than the general population (at least as reflected by NHANES participants).

The conclusion of AREDS 2 was that lutein and zeaxanthin should replace beta-carotene in the AREDS formula, and that the zinc could be reduced to 25 mg.

One problem with taking supplements is that they are bought over the counter, not prescribed. After the AREDS 1 study, Arora and colleagues349 from Wolverhampton identified 22 eye nutrient products but only two matched the dosages used in the trial.

There was some controversy over the role of genetic testing, with a theory put forward by Awh and colleagues350 that only some genetic subgroups benefited from antioxidants and zinc, and that some were actually harmed. This theory was refuted by Chew et al. 122 However, there may be different progression rates according to genetic factors. Seddon et al. 351 reported higher rates of progression in genotypes CFH Y402H (CC) and ARMS2 (TT).

Summary

The first AREDS trial showed that the supplement reduced the risk of progression by about 25%. At 10 years’ follow-up progression to advanced AMD was reduced from 44% to 34%. The AREDS 2 trial showed that beta-carotene should be replaced by lutein and zeaxanthin.

Quantity and quality of research

Summary: other studies of lutein and zeaxanthin supplements

Twenty-two studies (15 RCTs, two CCTs, two before-and-after studies and three cohort studies) compared lutein and/or zeaxanthin either separately or in combinations of intervention with themselves or inactive controls (i.e. placebo).

Three out of five lutein-only studies showed some improvement in VA, contrast sensitivity and MPOD.

The trials of lutein + zeaxanthin showed few differences but several compared different active arms, such as lutein + zeaxanthin versus lutein alone, with no placebo groups. Several studies reported reduced progression.

The three studies of zeaxanthin alone gave mixed results. One compared zeaxanthin alone versus lutein alone and found no differences. One found that zeaxanthin halved progression to wet AMD.

Several studies of combinations of lutein plus zeaxanthin plus other supplements reported benefit in BCVA and MPOD, but this could have been due to the lutein plus zeaxanthin, with other components of unproven benefit.

Reviews

One Cochrane review of fatty acids (Lawrenson and Evans 2015388) and one non-systematic review of fatty acids (Souied 2016389) were identified. Fatty acids were also mentioned in five reviews of multiple interventions but these were generally inconclusive except for the AREDS trials. 9,358,390–392

The Cochrane review388 concluded that there is no evidence from RCTs to support increasing omega 3 intake for preventing or slowing the progression of AMD. The systematic review included two studies, the AREDS 2 trial, already described, and the Nutritional AMD Treatment 2 (NAT-2) study by Souied et al.,393 described below.

The review by Souied389 noted that there was quite strong epidemiological evidence that diets rich in omega 3 fatty acid were protective against AMD, but that there was lack of evidence from intervention studies, including AREDS. However, they make the point that perhaps the well-nourished AREDS recruits had little to gain, but that people with low omega-3 fatty acid intake might gain more.

A systematic review by Huang et al. 394 examined the relationship between serum homocysteine, folic acid and vitamin B₁₂ levels, and the risk of AMD. The authors concluded that AMD is associated with elevated homocysteine levels and decreased vitamin B₁₂ levels, and that plasma homocysteine may act as a modulator of the risk for AMD.

One systematic review by Evans 2008357 and six non-systematic reviews (Bartlett 2003,395 Broadhead 2015,359 Evans and Lawrenson 2014,390 Gregori and Goldhardt 2015,360 Johnson 2010,391 Schmidl 2015396 and Scripsema 2015341) of antioxidants and nutritional supplements were identified. The Bartlett 2003395 review has been superseded by more recent ones. The Evans and Lawrenson 2014390 review summarises and updates their two Cochrane reviews. Gregori and Goldhardt360 described mainly the AREDS trials. The Johnson review391 preceded AREDS 2.

The Broadhead review, while not a systematic review, provides a very good summary of key issues, including:

• consumption of leafy green vegetables and fish is recommended as a way of reducing the risk of AMD

• a lot of studies have shown an inverse relationship between fish consumption and AMD

• this may be related to the presence of omega-3 fatty acids in fish and carotenoids in green vegetables

• the beneficial effects of eating a diet rich in fish are not replicated in intervention trials of DHA and EPA supplements

• the role of omega-3 fatty acids in the diet may be linked with reduced intake of omega-6 fatty acids (such as linoleic) from red meats and cheese

• it may be the overall diet that matters [increased fish consumption (2–4 helpings a week) may mean reduced red meat consumption]

• carbohydrates with a low glycaemic index (meaning a slow rise in blood glucose after consumption) instead of those with a high glycaemic index, may reduce the risk of AMD

• if supplements are to be used, the AREDS 2 formula should be used

• all patients with AMD should be advised to eat fish at least twice a week, eat more green leafy vegetables, and restrict high glycaemic index foods.

The Evans review357 concluded that people with AMD, or early signs of the disease, may experience some benefit from taking supplements as used in the AREDS trial, but that current evidence does not support the use of vitamin E or beta-carotene supplements to prevent AMD. However, it did support the use of the AREDS supplement.

The present systematic review includes three studies from the Evans systematic review357 (Teikara 1998,397 Christen 2007398 and Taylor 2002399) because these studies add to the totality of the evidence base for the research question of relevance here.

Five reviews of multiple interventions (Buschini 2015,310 Girmens 2013,67 Hanus 2012,9 Querques 2014361 and Sin 2013392) covered nutritional supplements and some epidemiological relationships. They came to no firm conclusions other than those by Hanus9 and Sin392 recommending the AREDs supplement.

Studies

The AREDS 2 trial, reported above, found no advantage in adding DAH and EPA to the original AREDS formula.

Five studies evaluated fatty acids including three RCTs by Feher et al. 2005,400 Souied et al. 2013393 and Tao et al. 2016401 and two cohort studies by Reynolds et al. 2013402 and Cougnard-Gregoire et al. 2016. 403 A total of 251 participants were randomised to a fatty acid intervention and 255 to placebo, while the cohort studies assessed dietary intake in a total of 3494 healthy participants. The studies were conducted in France (n = 2), the USA (n = 1), Hungary (n = 1) and China (n = 1). Follow-up ranged from 3 months to 3 years in the RCTs, and 7–12 years in the cohort studies. The RCTs had an unclear risk of selection bias, and the cohort studies were assessed as fair quality403 and poor quality. 402 Where reported, mean age was 63–74 years and 31–56% were men.

A recent study by Wu and colleagues404 combined cohorts from the Nurses’ Health Study (75,889 women) and the Health Professionals Follow-up Study (38,961 men), in which 1589 people developed intermediate AMD (defined as at least one of intermediate drusen, pigmentary abnormalities, large drusen or non-central GA) and 1356 developed advanced (curiously 96% wet) AMD. Comparison of extreme quintiles of DHA showed a lower incidence with higher intake in intermediate AMD (HR 0.78, 95% CI 0.66 to 0.92) but not in advanced AMD (HR 1.01). The HR for intermediate AMD in those consuming five or more helpings of fatty fish was 0.61 (95% CI 0.46 to 0.81) compared with those who rarely are fatty fish. 404

Results

Summary

Results are mixed, with little difference in mean BCVA, but some evidence of reduced progression with increasing DHA and EPA intake, and a reduction in prevalence of AMD associated with olive oil intake.

The Cochrane review388 of two RCTs393,405 concluded there is no evidence from RCTs to support increasing omega-3 intake for preventing, or slowing the progression of, AMD.

Five primary studies evaluating fatty acids were identified: three RCTs with an unclear risk of bias,393,400,401 one fair-quality cohort study403 and one poor-quality cohort study. 402 Interventions, length of follow-up and outcomes measures varied, and results were inconsistent between studies. Overall, the effects of fatty acids are inconclusive.

NCT01782352 is an AMD substudy of the Vitamin D and Omega-3 Trial (VITAL) being run in Boston, Massachusetts by Christen and colleagues. The main VITAL trial is concerned with cardiovascular disease cancer and stroke (www.vitalstudy.org/). It is due to end in 2017, after recruiting almost 26,000 people.

NCT02613572 is a trial of alpha lipoic acid in GA, being carried out in Pennsylvania by Kim and colleagues. There are doubts about tolerability in older people.

NCT02379169 is a 6-month trial in Finland of sea buckthorn oil (which has beta-carotene and zeaxanthin) combined with lutein.

Summary

Folate supplementation may reduce the risk of AMD.

Vitamins

The potential antioxidant effects of vitamins were examined in five RTCs in healthy populations by Christen et al. 2007,398 Christen et al. 2010,409 Christen et al. 2014410 Taylor et al. 2002399 and Teikari et al. 1998411 and one cohort study in people with dry AMD in at least one eye by Cangemi et al. 2007. 411 A total of 20,529 participants were randomised to vitamin E/alpha-tocopherol, 10,819 to beta-carotene, 7111 to a multivitamin, 257 to alpha-tocopherol plus beta-carotene and 38,214 to placebo. The cohort study included 34 people receiving an ‘oral antioxidant and omega-3 supplement’ (see below for details). Four of the studies were conducted in the USA, one in Australia and one in Finland. Follow-up was from ≤ 4 years to 12 years in the RCTs and was 6 months in the cohort study. The RCTs all had an unclear risk of selection bias and the cohort study was assessed as fair quality. One study received commercial funding, one received both commercial and non-commercial funding, and four had non-commercial funding but the supplements were provided by the manufacturer in two of these. Mean age ranged from 52–76 years and 42–100% were men, where reported. Baseline VA was reported by three of the studies.

Christen and colleagues 2007398 assessed the incidence of ARM in healthy male physicians receiving beta-carotene (50 mg on alternate days) or placebo for 12 years in a large RCT. There was no difference between groups in the risk of visually significant ARM (RR 0.96, 95% CI 0.78 to 1.20), ARM with or without vision loss (RR 1.01, 95% CI 0.86 to 1.20) or advanced ARM (RR 0.97, 95% CI 0.69 to 1.37) (Table 45). This remained the case after excluding cases diagnosed during the first 2 or 5 years of follow-up.

Christen and colleagues 2010412 included women from the Women’s Health Study who did not have a diagnosis of AMD at recruitment. With 10 years of follow-up there were fewer cases of visually significant AMD in those randomised to vitamin E and aspirin on alternate days than those given placebo, but the difference was not statistically significant (RR 0.93, 95% CI 0.72 to 1.19) (see Table 45). The incidence of advanced AMD was not significantly different between groups (RR 1.13, 95% CI 0.67 to 1.92), nor was the incidence of all AMD with or without vision loss (RR 0.90, 95% CI 0.77 to 1.06).

Taylor and colleagues 2002399 included healthy volunteers aged 55–80 years. The 4-year incidences of early AMD were similar between those randomised to high-dose vitamin E (500 IU daily) and those treated with placebo, with RR 1.05 (95% CI 0.60 to 1.61) (see Table 45). Results for late AMD, incidence of drusen and progression of AMD were similar between groups RR 1.36 (95% CI 0.67 to 2.77) for progression. The study reported that there was a significantly lower incidence of hypopigmentation in those taking vitamin E, but the data reported show a non-significant effect. No difference in BCVA was found (data not presented by the study). Quality of life was assessed using the visual function (VF-14) scale and it was stated that there were no differences between groups, but no data were presented.

Teikari and colleagues 1998397 compared the incidence of ARM among a subgroup of participants in a Finnish RCT on lung cancer prevention. Men who were ≥ 65 years and smoked at least five cigarettes a day were randomised to receive alpha-tocopherol, beta-carotene, alpha-tocopherol + beta-carotene or placebo. After approximately 6 years, no differences in the incidence of AMD were found among groups, RR for alpha-tocopherol 1.13 (95% CI 0.8 to 1.59) and 1.04 (95% CI 0.74 to 1.47) for beta-carotene (see Table 45) and there was no association between the supplements and prevalence of ARM after controlling for potential risk factors.

Christen and colleagues 2014410 included healthy male physicians from a cancer and cardiovascular prevention study (the Physicians Health Study II) who were not diagnosed with AMD. After 11.2 years of follow-up there was no significant difference between groups in cases of visually significant AMD (HR 1.19, 95% CI 0.94 to 1.50) or cases of advanced AMD (HR 1.22, 95% CI 0.88 to 1.70). However, the number of cases with total AMD with or without vision loss was significantly higher in those given multivitamins (HR 1.22, 95% CI 1.03 to 1.44).

Cangemi 2007411 compared one arm of their RCT, which compared MCS with sham microstimulation (abandoned because of lack of effect), with the placebo arm of another study. 186 As both arms of the original study received nutritional supplements, it is not clear why only the patients in the sham arm were included in this study. It was really a before-and-after study with no local controls. Participants in Cangemi 2007411 had at least one eye with dry AMD (> 10 large soft drusen of 63 µm in diameter, within 3000 µm of the fovea centre) and BCVA in the study eyes of 20/32 to 20/125 inclusive (ETDRS). After 6 months of a nutritional supplement containing vitamins A, C and E, zinc, copper, taurine, EPA fatty acid, DHA fatty acid, lutein and zeaxanthin (with sham microstimulation), VA was reported to have improved from baseline by 0.5 ETDRS lines (p = 0.045; see Table 45) compared with a deterioration in the placebo arm of the other study (no statistical comparison). BCVA improved in 56.7% of participants. There was little change in other outcomes, including contrast sensitivity and NEIVFQ-25 (see Report Supplementary Material 5). Because of the mixture of vitamins, metals, fatty acids and lutein given, it is not possible to say which ingredient had the reported effect.

Adverse events

Four papers (Christen and colleagues 2010,412 Christen and colleagues 2007,398 Teikari and colleagues,397 Christen and colleagues 2014410) did not report AEs.

Taylor and colleagues 2002399 found no serious AEs. Ocular AEs occurred in 18% of participants in the vitamin E group and 15% in the placebo group, not a significant difference. Cangemi 2007411 reported that there were no significant systemic or ocular AEs related to the nutritional supplement.

Summary

The Huang et al. 394 review concluded that AMD is associated with elevated homocysteine levels and decreased vitamin B₁₂ levels. Four additional studies (Christen 2015,406,413 Merle 2016407 and Gopinath 2013408) were identified that assessed homocysteine levels, folic acid and B vitamins: one RCT (Christen 2009406) and three good-quality cohort studies (Christen 2015,413 Merle 2016,407 Gopinath 2013408) were identified that assessed homocysteine levels, folic acid and B vitamins. The results from the primary studies suggest an association with AMD and lower intake of B vitamins or reduced serum B₁₂ levels. Results for the effect of folic acid were inconsistent.

The Evans review357 concluded that people with AMD, or early signs of the disease, may experience some benefit from taking supplements as used in the AREDS trial, but that current evidence does not support the use of antioxidant vitamin supplements to prevent AMD. Four large RCTs found no beneficial effect of beta-carotene or vitamin E/alpha-tocopherol in healthy individuals. One large RCT found higher total AMD in people given multivitamins. It may be that, as in category 1 in the AREDS trial, the rates of progression over the durations observed are too low to show any effect.

Quantity and quality of research

Summary

Two non-systematic reviews, four small studies and one abstract for a larger but short-term trial, were identified. The Falsini crossover RCT417 of uncertain risk of selection bias reported that VA was better with saffron treatment after 90 days. After 14 months of treatment with saffron, the Piccardi420 before-and-after study of good quality reported a significant increase in VA. Two studies did not report VA outcomes. No adverse effects of saffron supplementation were recorded. Three studies and one of the reviews were from the same centre.

Overall, the evidence suggests a potential for benefit but further evidence is required. An important point made by Bisti and colleagues419 in their review is that the content of saffron may vary depending on source.

Reviews

One high-quality systematic review by Vishwanathan and colleagues 2013427 was identified so we did not go back to the primary studies. Vishwanathan and colleagues 2013427 included 10 studies: four RCTs, four prospective cohort studies and two retrospective cohort studies. Outcomes were incidence of any AMD, early AMD and late AMD, BCVA, progression to GA. The review concluded that the current evidence on zinc intake for the prevention of AMD is inconclusive. The authors state that zinc treatment may be effective in preventing progression to advanced AMD but zinc alone may not be sufficient to produce clinically meaningful changes in VA. Recommendations for research were that evaluations of different forms of zinc supplementation (e.g. zinc oxide vs. zinc sulphate vs. zinc monocysteine) are required because bioavailability varies. The AREDS 1 trial used 80 mg of zinc oxide, AREDS 2 used 25 mg, Newsome 1988342 used zinc sulphate 100 mg, Newsome 2008428 used zinc monocysteine 25 mg twice daily, and Stur 1996429 used 200 mg of zinc sulphate. The Vishwanathan review was written before the AREDS 2 results were available.

Zinc was also mentioned in several non-systematic reviews310,322 of multiple interventions for dry AMD but these referred to the Vishwanathan review or the AREDS trial papers so did not add anything.

Studies

No published studies not covered by the Vishwanathan and colleagues 2013427 review were identified.

Summary: zinc

One good-quality systematic review concluded that there is inconclusive evidence for the use of zinc to prevent AMD but that it may be effective in prevention of progression to advanced AMD. It is included for this purpose in the AREDS 2 formula.

Physical treatments

• Newer forms of laser treatment show promise but a large trial, the LEAD trial from a world centre of excellence in Melbourne is ongoing (anticipated completion date June 2018), so we suggest waiting for their results.

• Implantable telescopic lenses also show promise, but a NIHR EME study is under way (MIRROR) in advanced AMD.

• There is insufficient good-quality evidence to recommend use of, or further research in, acupuncture, MCS or ozone.

• There is some evidence on rheopheresis, but the largest trial showed no benefit, most studies reporting positive results were small with only modest effect sizes and mostly uncertain risks of bias, and treatment would be inconvenient to elderly patients. So we do not see rheopheresis as a research priority.

• The evidence for the use of blue-light-filtering IOLs after cataract extraction is currently insufficient to justify their routine use, but further research is currently under way.

Cells

• One small before-and-after study and one very small case series of cell (RPE) transplantation was identified. Improvements in VA were found in over half of treated eyes.

• The evidence base is still very sparse, but this seems a promising development, and further research is under way.

Drug treatments

• We think that there is sufficient evidence to justify a trial of a potent statin, such as atorvastatin 80 mg daily.

• Fenretinide is a visual cycle inhibitor, which may reduce the deposition of lipofuscin. We found one trial, with an unclear risk of bias and run and written up by the manufacturer’s staff, which had mixed results. Progression of GA was little different overall, but was a bit lower in the subgroup that achieved greater serum RBP levels. However progression to wet AMD was halved by fenretinide. There were higher rates of AEs with fenretinide. Overall, we think a trial in early dry AMD may be justified, but the roughly 20% drop-out rates in the active drug arms should be noted.

• An impressively large retrospective study from the USA found that people taking L-dopa were less likely to develop AMD, and that if they did develop it, it was about 7 years later than among people not taking L-dopa. Further research is needed, perhaps using the large UK general practice-based databases, THIN and CPRD, in order to assess whether a trial assessing its use in treating AMD could be justified.

• The current evidence on lampalizumab suggests benefit, but very large trials are under way (sponsored by the manufacturer), so no new research is indicated meantime.

• There is a little evidence of benefit from glatarimer acetate, but with only some shrinkage of drusen in two studies which had unclear risks of bias. The evidence is too sparse to justify NIHR research at present. Future trials can be left to the manufacturer if deemed worthwhile.

• One small study reported benefit from prednisolone but it scored poorly on quality assessment so there is insufficient evidence to justify its use. Systemic steroids have adverse effects so if steroids were to be used, a topical one would seem better. The results of a trial of an implanted steroid, fluocinolone (NCT00008515), are awaited.

• For nine drugs, there was some evidence showing no or very little benefit, or even harm, so we do not recommend further consideration. They are alprostadil, eculizumab, dorzolomide, OT 551 eye drops, sirolimus, tandospirone, trimetazidine, visaline and emixustat.

Nutrients

• There are many studies of nutritional supplements but some are too small, of poor quality, of too short duration, or are of combinations of compounds, making it difficult to assess the relative contributions of each. Supplements used include lutein and zeaxanthin, in combination or individually, or combined with a variety of other minerals and/or vitamins, the omega-3 fatty acids DHA and EPA, olive oil, folate acid, various vitamins, ginkgo biloba, curcumin (from turmeric), saffron and zinc.

• The first AREDS trial showed benefit in category 3 and 4 patients, which persisted for 7 years, with a modest but useful slowing of progression, which could mean that 30% of people expected to progress to advanced AMD over a 5-year period, would not. The trial did not have enough power to confirm, or not, effects in categories 1 and 2.

• The AREDS 2 trial showed that beta-carotene should be replaced by lutein and zeaxanthin, and that the dose of zinc could be reduced.

• We therefore think that there is good evidence that the AREDS 2 supplement should be used for patients meeting the AREDS 3 and 4 categories.

• Saffron extracts have been reported to show some benefits in VA and might justify further research.

• There is currently insufficient evidence to recommend any other nutritional supplements.