Introduction

Photoreceptors in the retina are subject to oxidative stress throughout life due to combined exposures to light and oxygen. It has been proposed that antioxidants may prevent cellular damage in the retina by reacting with free radicals produced in the process of light absorption.1

There are a number of non-experimental studies that have examined the possible association between antioxidant micronutrients, although few studies have examined supplementation specifically.2 Data on vitamin intake in observational studies should be considered cautiously as people who have a diet rich in antioxidant vitamins and minerals or who choose to take supplements regularly are different in many ways from those who do not; these differences may not be adequately controlled by statistical analysis. Inconsistent results have been found in these observational studies.

Antioxidant vitamin and mineral supplements are increasingly being marketed for use in age-related eye disease, including AMD. This article summarises the results of two comprehensive regularly updated systematic reviews published on the Cochrane Library.3, 4 Two questions are addressed: firstly, should the general population be taking routine antioxidant and vitamin mineral supplements to prevent or delay the onset of AMD in later life? Secondly, should people with AMD be taking these supplements to slow down the progression of the disease?

Methods

This review included randomised trials comparing antioxidant vitamin and/or mineral supplementation (alone or in combination) to control. Antioxidants were defined as any vitamin or mineral, which is known to have antioxidant properties in vivo or, which is known to be an important component of an antioxidant enzyme present in the retina. We considered the following: vitamin C, vitamin E, carotenoids, selenium, and zinc. AMD was defined as the presence of geographic atrophy or neovascular disease. Age-related maculopathy (ARM) was used as an overall term encompassing both early age-related macular changes (large soft drusen, hyper- and/or hypo-pigmentation) and AMD.5, 6

Two main outcome measures were considered: development and progression of ARM and AMD and loss of vision. Quality of life was considered but not reported in the included trials. Information on adverse events was collected.

Comprehensive systematic searches were done.3, 4 In brief, we searched the Cochrane Central Register of Controlled Trials (CENTRAL) (which contains the Cochrane Eyes and Vision Group Trials Register) in The Cochrane Library, MEDLINE, SIGLE, EMBASE, National Research Register, AMED and PubMed, reference lists of identified reports, and the Science Citation Index. We contacted investigators and experts in the field for details of unpublished studies. Searches were updated in August 2007.

Titles and abstracts of all reports of trials identified by the searches were assessed and full texts of possibly relevant trials were obtained. Trials were selected according to prespecified inclusion criteria.3, 4 Five parameters of trial quality were assessed: allocation concealment, method of allocation to treatment, documentation of exclusions, masking of outcome assessment, and completeness of follow-up.7 Each parameter of trial quality was graded: A, low risk of bias; B, moderate risk of bias; or C, high risk of bias. Trials scoring C on allocation concealment (ie, where allocation was not concealed properly) were excluded.

Heterogeneity was assessed by examining the forest plot to see whether the effect measures for the different studies were in the same direction and of a similar order of effect and by the I2 value.8 An I2 value of 50% or more was taken to indicate considerable inconsistency of results such that a pooled result may be inaccurate and should not be reported. A random effects model was used to pool the data, unless there were three or fewer trials in which case a fixed effect model was used. The main clinical diversity was with respect to the type of supplement. This was incorporated into the analysis strategy by considering the formulations by type. The methodological quality of the studies was reported and used to interpret the results. Currently, there are not enough published studies to enable sensitivity analysis or formal assessment of publication bias; however, the forest plots were reviewed to see whether smaller studies were reporting larger effects.

Trials in this area fall into two broad categories: those evaluating a single vitamin or mineral (eg, vitamin E or zinc) and those investigating a broad spectrum formulation (eg, Ocuguard). The following comparisons were considered in this review.

  1. 1)

    Broad-spectrum formulation vs placebo. Within this category fall all the broad-spectrum formulations, which include one or more antioxidant vitamins or minerals.

  2. 2)

    Single-component formulations vs placebo, eg, vitamin E alone.

  3. 3)

    Broad-spectrum or single component studies together. For the progression of AMD, this comparison was subject to considerably clinical, methodological, and statistical heterogeneity and is not reported here.

Results

Results are presented for the included trials. Details of the excluded studies are available in Evans and Henshaw3 and Evans.4

Prevention of AMD

Table 1 shows the trials investigating the primary prevention of AMD. In the Alpha-Tocopherol and Beta-Carotene (ATBC) Study in Finland, 29 133 male smokers were randomly allocated to α-tocopherol (vitamin E) (50 mg/day), β-carotene (20 mg/day), vitamin E or β-carotene, or placebo.9 Treatment ranged from 5 to 8 years. A random sample of 1035 men aged 65 years and above were sampled at the end of the study and fundus photographs were taken and graded for ARM and AMD. In the Physicians' Health Study in the USA, 22 071 male physicians were randomised to aspririn (325 mg every other day), β-carotene (50 mg every other day), aspirin, and β-carotene or placebo.10 Treatment duration ranged from 11.6 to 14.2 years. A total of 21 142 men provided self-reported information on AMD, which was confirmed by medical record review.11 In the Vitamin E cataract and ARM trial, 1204 men and women aged 55–80 years were randomly allocated vitamin E (335 mg/day) or placebo and followed up for 4 years.12 In all, 82% of the participants in this study had no evidence of ARM at enrolment which is why this trial is included in the prevention rather than progression of AMD results. Conversely, the Age-Related Eye Disease Study (AREDS) enrolled mostly people with signs of ARM and outcome data were not reported on the 1117 people without ARM at enrolment.13 AREDS is discussed below under the progression of AMD.

Table 1 Trials investigating antioxidant vitamin and mineral supplements in the prevention of AMD
Full size table

All three trials were of high quality and scored ‘low risk of bias’ on all quality parameters. Allocation was concealed by means of coded tablets and randomly assigned; exclusions were documented and follow-up was equal between the study groups; outcome assessment was masked to study group and analysis was intention-to-treat.

In ATBC, 728 people were randomised to any antioxidant and 213 to placebo.9 There was no association of treatment group with any sign of maculopathy. There were 216 cases of ARM in the antioxidant groups and 53 in the placebo group (risk ratio (RR)=1.19, 95% confidence interval (CI), 0.92–1.54). The majority of these cases were early ARM. There were only 14 cases of AMD. Of these, four were geographic atrophy and 10 neovascular disease. There was only one case of geographic atrophy in the placebo group. All the other cases of late stage disease occurred in the antioxidant groups. The findings are similar when each of the antioxidant groups—vitamin E, β-carotene, vitamin E, and β-carotene—are compared with placebo.

In PHS I, there were 162 cases of ARM causing visual loss of 6/9 or worse in the β-carotene group vs 170 cases in the placebo group (RR=0.96, 95% CI, 0.78–1.20).11 Secondary end points of ARM with or without vision loss (275 vs 274 cases, RR=1.01, 95% CI, 0.86–1.20) and AMD(63 vs 66 cases, RR=0.97, 95% CI, 0.69–1.37) were reported.

In VECAT, there were 92/504 in the vitamin E group with ARM compared to 92/512 in the placebo group (RR=1.02, 95% CI, 0.78–1.32).12 The majority of these cases were early ARM. There were nine cases of late AMD, five in the treatment group and four in the placebo group.

Overall 23 099 participants were randomised in the three trials. There were 583 cases of ARM in the antioxidant groups and 419 cases of ARM in the placebo groups (Figure 1). The results of the three studies were consistent (I2=0%). There was little evidence of any effect of antioxidant supplementation (RR=1.04, 95% CI, 0.92–1.18). Similarly, for AMD, the trials were consistent and indicate little evidence of any effect of supplementation (RR=1.03, 95% CI, 0.74–1.43) (Figure 2). There were fewer AMD events (81 antioxidant, 71 placebo).

Figure 1
figure 1

ARM in people given antioxidant vitamin supplements compared to people given placebo. ARM, age-related maculopathy; ATBC, vitamin E (dl-α-tocopheryl acetate 50 mg (approx 110 IU)/day) and β-carotene (20 mg/day); PHS I, β-carotene (50 mg every other day); VECAT, vitamin E (d-α tocopherol 500 IU/day).

Full size image
Figure 2
figure 2

AMD in people given antioxidant vitamin supplements compared to people given placebo. AMD, age-related macular degeneration; ATBC, vitamin E (dl-α-tocopheryl acetate 50 mg (approx 110 IU)/day) and β-carotene (20 mg/day); PHS I, β-carotene (50 mg every other day); VECAT, vitamin E (d-α tocopherol 500 IU/day).

Full size image

There was less evidence available comparing vitamin E alone vs placebo. A total of 1466 people randomised in VECAT and ATBC resulted in 167 cases of ARM in the vitamin E group and 145 in the placebo group (Figure 3). The trial results were reasonably consistent, I2=19%. There was little evidence of any effect of supplementation with vitamin E on the incidence of ARM; RR=1.11 (95% CI, 0.91–1.36). There were fewer cases of AMD—13 in the vitamin E groups and five in the placebo (Figure 4). All effect measures were in the direction of harm but were not consistent (I2>50%).

Figure 3
figure 3

ARM in people given vitamin E supplements compared to people given placebo. ARM, age-related maculopathy; ATBC, vitamin E (dl-α-tocopheryl acetate 50 mg (approx 110 IU)/day); VECAT, vitamin E (d-α tocopherol 500 IU/day).

Full size image
Figure 4
figure 4

AMD in people given vitamin E supplements compared to people given placebo. AMD, age-related macular degeneration; ATBC, vitamin E (dl-α-tocopheryl acetate 50 mg(approx 110 IU)/day); VECAT, vitamin E (d-α tocopherol 500 IU/day).

Full size image

A total of 21 589 people were randomised to β-carotene or placebo in ATBC and PHS I. There were 343 cases of ARM in the β-carotene groups and 327 in the control groups (Figure 5). The results of the trials were consistent (I2=0%) and did not indicate any benefit of supplementation (RR=1.03, 95% CI, 0.98–1.19). There were 65 cases of AMD in the β-carotene groups and 67 cases of AMD in the control (Figure 6). Again the results of the trials were consistent (I2=0%) and indicated little effect of supplementation (RR=0.97, 95% CI, 0.69–1.36).

Figure 5
figure 5

ARM in people given β-carotene supplements compared to people given placebo. ARM, age-related maculopathy; ATBC, β-carotene (20 mg/day); PHS I, β-carotene (50 mg every other day).

Full size image
Figure 6
figure 6

AMD in people given β-carotene supplements compared to people given placebo. AMD, age-related macular degeneration; ATBC, β-carotene (20 mg/day); PHS I, β-carotene (50 mg every other day).

Full size image

Progression of AMD

Table 2 shows the trials investigating the effect of antioxidant supplementation on the progression of AMD.

Table 2 Trials investigating antioxidant vitamin and mineral supplements in the progression of AMD
Full size table

The majority of evidence comes from one trial in the USA, AREDS, which compared four treatment groups: antioxidants (vitamin C 500 mg, vitamin E 400IU, and β-carotene 15 mg)/day, zinc (zinc oxide 80 mg and cupric oxide 2 mg), antioxidants plus zinc, and placebo.13 A total of 3640 people were randomised. Treatment duration averaged 6.3 years. A total of 525 people have been randomised into other trials, which have evaluated a range of supplements both broad spectrum (Ocuguard,14 Ocupower,15 and Visaline,16) and single component (zinc17, 18 and lutein15). The smaller studies in general were of shorter duration (6–24 months) and lower quality. There have been two unpublished trials on zinc supplementation.19, 20

In most trials, randomisation appeared to have been executed properly, ie, an unpredictable sequence of treatment allocation was concealed adequately from people recruiting participants into the trial. In AMDSG, more people in the placebo group withdrew (six) compared to the treatment group (one).14 The description of the tablets cannot exclude the possibility that there were detectable differences between treatment and placebo that may mean that some participants in the study were unmasked. In AREDS, four people were documented as being unmasked to study group.13 More people in the antioxidant group (8.3%) reported changes in skin colour (yellowing) than in the placebo group (6.0%, P<0.01) and more people in the zinc groups reported difficulty in swallowing the study tablets (17.8 vs 15.3%, P=0.04). However, there was little evidence of unmasking when at the end of the study participants were asked to guess their treatment assignment. The percentage of participants who guessed correctly, by treatment assignment, were: placebo 17%, antioxidants alone 16%; zinc alone 18%; and antioxidants plus zinc 16%. In the Veterans LAST study, the tablets were apparently identical in appearance but it was not clear whether taste or systemic effects differed between the different groups.15 In Stur et al,18 analysis of the main outcome measures (visual function and progression of disease) was not carried out on a strictly intention-to-treat basis as anyone experiencing the endpoint of late-stage AMD (neovascularisation) was withdrawn from the study. Contact with the trial investigator revealed that all of these participants ended up with visual acuity of 20/200 or less and that these participants were excluded because the investigators wished to detect functional changes caused by degeneration of the retinal pigment epithelium and the sensory retina, and not vision losses caused by choroidal neovascularisation.

AREDS was the only trial reporting progression of AMD and visual acuity loss in a dichotomous format. There was a beneficial effect of treatment on progression to AMD (odds ratio adjusted for age, sex, race, AMD category, and smoking at enrolment 0.68, 95% CI, 0.53–0.87) (Figure 7), and loss of 15 or more letters (adjusted odds ratio=0.77, 95% CI, 0.62–0.96; Figure 8).

Figure 7
figure 7

Progression to AMD in people with ARM given antioxidant vitamin and mineral supplements compared to placebo. ARM, age-related maculopathy; AMD, age-related macular degeneration; AREDS, vitamin C (500 mg/day), vitamin E (dl-α-tocopheryl acetate 400 IU/day), β-carotene (15 mg/day), zinc (80 mg/day).

Full size image
Figure 8
figure 8

Loss of 15 or more letters visual acuity in people with ARM given antioxidant vitamin and mineral supplements compared to placebo. ARM, age-related maculopathy; AMD, age-related macular degeneration; AREDS, vitamin C (500 mg/day), vitamin E (dl-α-tocopheryl acetate 400 IU/day), β-carotene (15 mg/day), zinc (80 mg/day).

Full size image

It was difficult to extract meaningful data on AMD progression from the smaller studies. One study reported data on the progression of AMD in a continuous format.14 There was little evidence for any benefit of treatment (mean difference −0.06, 95% CI, −0.62 to 0.50). The number of participants in this analysis was small with 35 in the treatment group and 24 in the control group. There was some information on mean visual acuity.14, 15, 16 A total of 69 people were randomised to treatment and 62 to placebo in pooled analyses of all three trials. Little effect of treatment on visual acuity was seen from these analyses. The pooled standardised mean difference was 0.16 (95% CI, −0.19 to 0.51). The results of these trials were consistent, I2=0.

Four trials have reported the effect of zinc supplementation.13, 17, 18, 19 In addition, there is one unpublished study for which we have no data.20

Three trials provided data on progression of AMD as a dichotomous outcome13, 18, 19 (Figure 9). A total of 969 people were randomised to zinc supplementation and 974 to placebo. Overall, there was a modest benefit of treatment. The pooled odds ratio was 0.73 (95% CI, 0.58–0.93). Stur et al18 had quite different results to the other two trials. Over the treatment period, nine people experienced a choroidal neovascularisation (CNV) in the study eye in the zinc group compared to five people in the placebo group. However, this may have been a chance finding . The odds ratio for that trial (2.31) had wide CI and the results are, therefore, also consistent with a protective effect of treatment (95% CI, 0.58–9.26). Overall, the I2 value was 29.0%. Holz et al19 has been published in abstract form only, so we have little information about this trial.

Figure 9
figure 9

Progression to AMD in people with ARM given zinc supplements compared to placebo. ARM, age-related maculopathy; AMD, age-related macular degeneration; AREDS, zinc oxide (80 mg/day); Holz et al19, zinc sulphate (100 mg twice daily); Stur et al18, zinc sulphate (200 mg/day).

Full size image

Two trials reported dichotomous visual acuity data.13, 17 The pooled analyses include a total of 984 people randomised to zinc supplementation and 974 to placebo. There was a modest beneficial effect of treatment on visual acuity (pooled odds ratio=0.81, 95% CI, 0.66–0.99). The trials were consistent I2=0% (Figure 10).

Figure 10
figure 10

Loss of 15 or more letters visual acuity in people with ARM given zinc supplements compared to placebo. ARM, age-related maculopathy; AMD, age-related macular degeneration; AREDS, zinc oxide (80 mg/day); Newsome et al,17 zinc sulphate (100 mg twice daily).

Full size image

Two trials provided data on mean visual acuity.17, 18 A total of 77 people were randomised to zinc supplementation and 78 to placebo in these two trials, which had a maximum treatment and follow-up duration of 24 months. The results of these trials were less consistent, I2=56.6%. Newsome et al17 found that there was more visual acuity loss in the control group than the treatment group, although this did not reach statistical significance. Stur et al18 found little difference between the two groups with respect to mean visual acuity at the end of the study. In Stur et al,18 the primary outcome was incidence of CNV in all patients. During the treatment period, a CNV developed in the study eye in 14 patients (nine in the treatment group and five in the placebo group). People who experienced a CNV were not included in the analyses of visual acuity.

There has only been one trial published to date comparing supplementation with lutein vs placebo.15 The trial was small with a total of 29 people randomised to lutein supplementation and 31 to placebo; the treatment duration and follow-up was 12 months. The only outcome of relevance to this review, for which data could be extracted, was mean visual acuity at the end of the study. This showed little evidence of any effect of treatment: mean difference logMAR acuity 0.04 (95% CI, −0.15–0.23). The power of the study was low.

The main reported adverse effect leading to withdrawal from the studies was gastrointestinal symptoms. Of 286 people randomised into trials of zinc sulphate supplementation compared to placebo, 5/146 zinc-treated people withdrew due to gastrointestinal symptoms compared to 2/140 controls.17, 18 No one developed copper-deficiency anaemia. In AMDSG, one person developed an ‘allergic reaction,’ although it was not clear whether or not this was related to the treatment.14 AREDS considered a number of safety outcomes. They conducted over 100 comparisons of zinc vs no zinc and antioxidants vs no antioxidants. Participants in the antioxidant arms more frequently reported yellow skin (8.3 vs 6.0%, P=0.008). Participants in the zinc arms reported more anaemia (13.2 vs 10.2%, P=0.004), however, serum haematocrit levels were the same. People taking zinc were more likely to require hospital admission due to genitourinary complications.21

Discussion

Routine antioxidant vitamin supplementation with vitamin E or β-carotene probably does not protect against the development of AMD in later life. This review includes three large high quality studies that have randomised over 23 000 members of the population to antioxidant supplementation or placebo and followed-up for AMD. Duration of supplementation has ranged from 4 to 12 years. This represents a substantial amount of evidence. Most of the randomised people in these trials were men. We have no reason to suppose, however, that the effect of supplementation will be different in men and women. No trials of other antioxidant vitamins or minerals supplements to prevent AMD have been reported.

This review provides evidence that people with ARM may experience delay in progression of the disease with antioxidant vitamin and mineral supplementation. This evidence comes from the AREDS trial and therefore only applies to the formulation used in that study (vitamins C and E, β-carotene, and zinc). This trial was conducted in a relatively well-nourished American population. Until it is replicated by other large-scale trials in other populations, we will not know whether these findings can be applied more generally. The other trials of broad-spectrum multivitamin preparations, eg, Ocuguard and Ocupower were too small to provide evidence either way. These trials were also of relatively short duration and lower quality. Variable methods of presenting outcome data made it difficult to pool results.

The AREDS trial provides evidence that long-term supplementation with vitamins C, E, β-carotene, and zinc, in people with ARM, reduced the risk of progression of the disease and visual acuity loss. The overall benefit is modest with a risk reduction in the order of 20–25%. However, given that treatment options for AMD are limited, and vision loss is rarely recovered, this is of interest to people with AMD. As AREDS is a large well-conducted randomised study, potential biases will have been minimised. Bias may have been introduced, if there were different systemic effects of the antioxidant and zinc supplementation (eg, yellowing of skin or difficulty in swallowing tablets), which led the participants to guess which group they were in; or alternatively, the retinal fundus photographs might have been different in some way such that the graders response was affected by treatment group. There is little evidence that this was a problem in the study.

A total of five trials investigated zinc. The AREDS study indicated that the beneficial effect of zinc supplementation was of a similar order to that of vitamin supplementation. The other trials provide more conflicting evidence. Newsome et al17 found a reduction in the risk of visual acuity loss with supplementation over 12–24 months. However, Stur et al18 found no effect of treatment. Unfortunately Stur et al,18 who was planned to recruit 500 participants, was terminated early because the results of the first 40 patients at 24 months indicated no benefit of treatment. The other two trials of zinc supplementation are as yet unpublished, although limited results from Holz et al19 were published in abstract form and are included here.

AREDS was the only study to examine in detail the question of safety. There was some evidence that zinc supplementation at the level in the study resulted in increased hospital admissions due to genitourinary complications.21 Similarly, the safety of some of the components of the AREDS formulation has been questioned in other studies. Two large randomised controlled trials have indicated that smokers who take β-carotene may be at increased risk of developing lung cancer.22, 23 The Heart Outcomes Prevention Evaluation (HOPE) Study found that, among people with vascular disease or diabetes, vitamin E supplementation was associated with a higher risk of heart failure.24

There is currently considerable interest in the potential role of lutein and zeaxanthin supplementation in AMD. This review includes only one small equivocal trial on lutein. Such supplements currently cannot be recommended. Trials are ongoing (http://www.areds2.org/, accessed 18 September 2007).

There are a number of unanswered questions in the prevention of AMD. The hypothesis that antioxidant micronutrients may protect against the disease is a reasonable one. We do not know at what stage the protective effect may be important, nor the potential interactions with genetic effects and other risk factors for the disease such as smoking. The research to date shows that we cannot extrapolate to taking vitamin supplements without good evidence of their effectiveness and safety.