Miller JW. Legacy of the Age-Related Eye Disorder Study. Arch Ophthalmol. 2009;127(12):1680-1685. doi:10.1001/archophthalmol.2009.321
The Age-Related Eye Disorder Study (AREDS) was designed as a long-term multicenter prospective study of the clinical course of age-related macular degeneration (AMD) and age-related cataract. The primary goal of the study was to generate data progression rates and risk factors for these disorders to increase our understanding of the clinical course of AMD and cataract, generate hypotheses about etiology, and aid in the design of clinical trials for intervention. In addition, 2 clinical trials were included in the design: a high-dose vitamin and mineral trial for AMD and a high-dose vitamin trial for cataract. The AREDS certainly achieved the primary goal of the study. We obtained important natural history data, including rates of progressions for different stages of AMD, providing prognostic information for patients and facilitating the design of subsequent interventional studies. The AREDS refined useful grading schemes for both clinical trials and clinical practice. A treatment regimen for certain patients with AMD was identified and adopted, though acceptance of AREDS recommendations has been tainted by controversy over the data presentation and analysis. While there is continued discussion of the molecular mechanisms underlying the treatment benefit demonstrated for patients with AMD, the AREDS findings have spawned other clinical trials for nonneovascular AMD. These include a follow-up, AREDS 2, and other interventions to prevent progression. The AREDS generated other observations, including the interaction of cataract and AMD, the reliability of quality of life measures, and the association of AMD and mortality. There have been 24 published manuscripts from AREDS through 2008 and scores of published abstracts and proceedings.
The AREDS was undertaken at a time when there was no agreement on the pathogenesis of AMD and disagreement over the grading schemes. Epidemiological studies suggested that genetics, smoking, and cardiovascular risk factors played a role.1,2 In addition, epidemiologic and small randomized trials showed inconsistent associations between the intake of antioxidants and the risk of AMD, including 1 small, randomized, 2-year, placebo-controlled clinical trial of zinc supplementation, which demonstrated a statistically significant reduction in vision loss in patients with AMD treated with zinc.3 At the time, there was intense marketing of zinc and antioxidants for AMD and a resulting public health concern over the safety of vitamin and mineral supplementation in this population. It therefore seemed reasonable to incorporate a clinical trial for AMD into AREDS. A role for antioxidants in preventing age-related cataract was also postulated, since oxidative damage to lens proteins is a prominent feature of cataract. Similar to the experience with AMD, cross-sectional and prospective epidemiologic studies of cataract and intake or blood levels of antioxidants provided inconsistent results.4- 6 The choice of vitamins and mineral supplements was guided by the somewhat imperfect data available. Four treatment interventions were devised: antioxidants (500 mg of ascorbic acid, 400 IU of DL-alpha-tocopheryl acetate, and 15 mg of beta carotene); zinc (80 mg as zinc oxide and 2 mg of copper as cupric oxide to prevent anemia); a combination of zinc and antioxidants; or placebo. Although there was interest at the time in the carotenoids lutein and zeaxanthin, none was readily available for manufacturing in a research formulation, and they were not studied. The AREDS was an early successful example of a public/private partnership in medical research; it was funded predominantly by the National Eye Institute using intramural funds, with additional support from Storz Ophthalmic Pharmaceuticals (currently owned by Bausch & Lomb Pharmaceuticals). National Eye Institute staff proposed AREDS in 1986 and developed the protocol between 1990 and 1992; in November 1992, the first patient was enrolled.7
A total of 4757 persons aged 55 to 80 years were enrolled in the first AREDS, with recruitment of non–African Americans complete in the first 3 years of enrollment, and recruitment of African Americans continuing through January 1998. Eligible participants were enrolled into 1 of 4 categories (aiming for 1000 per category) and randomized to intervention accordingly. Category 1 was included to provide a comparison group free of age-related macular changes and were randomized to placebo or antioxidants only. Category 2 participants had mild or borderline age-related macular changes (multiple small drusen [<63 μm] or intermediate drusen and/or pigment abnormalities) in 1 or both eyes. Category 3 participants had large drusen (≥125 um), extensive intermediate drusen (≥63 um), and/or geographic atrophy not involving the center of the macula. Subjects in categories 1 through 3 had to have a visual acuity of 20/32 or better in the study eye. Category 4 subjects had advanced AMD or decreased vision of 20/32 or less in 1 eye and good vision in the second eye.7 Advanced AMD was defined as having 1 of the following: retinal pigment epithelial detachment, serous or hemorrhagic neurosensory detachment, hard exudates, subretinal or subretinal pigment epithelial hemorrhage, subretinal fibrous tissue, or laser scars; or geographic atrophy involving the center of the macula. For some analyses, category 4 was subdivided into neovascular AMD and geographic atrophy groups. This classification scheme was modified from the Wisconsin Age-Related Maculopathy Grading System, with the modifications related primarily to neovascular changes. Therefore, AREDS findings are readily comparable with those from studies using the Wisconsin scheme, such as the Beaver Dam studies, as well as those using the International Classification and Grading System for age-related maculopathy and AMD.8 The AREDS categories led to a cleaner nomenclature for stages of AMD now used by clinicians and investigators, including the term no AMD from category 1; an early stage of AMD from category 2; the intermediate stage of AMD from category 3; and 2 forms of an advanced stage of AMD, namely, an atrophic advanced stage defined as central geographic atrophy and a neovascular advanced stage defined as features of choroidal neovascularization.
It is important to note that AREDS participants were allowed to continue taking their normal vitamins; 55% of subjects were using a multivitamin or at least 1 vitamin or mineral being studied in AREDS at the time of screening. Enrolled subjects were allowed to take Centrum if they wished to continue to take a multivitamin and 66% of AREDS participants elected to take CentrumR. On learning of the current AREDS recommendations regarding vitamin and mineral supplementation, many patients ask about continuing to take a multivitamin. The AREDS data support their doing so.
The baseline characteristics of the macula in AREDS participants proved similar to those of other published results.9 Advanced disease (both groups 4 and 5 in category 4) occurred in older individuals than in the control group. Those with early disease (group 2) were more likely to be female and have arthritis and less likely to have angina. Those with intermediate disease (group 3) were more likely to be taking hydrochlorothiazide diuretic drugs and to have arthritis. Subjects in group 3 and 5 (neovascular AMD) but not group 4 (central geographic atrophy) were more likely to have hypertension, hyperopia, and lens opacities and to be white. Those in group 5 were more likely to have increased body mass indexes and those in groups 3, 4, and 5 were more likely to be smokers and have less education. Group 4 was more likely to use thyroid hormones and antacids. The association of increased age, hypertension, and smoking with more advanced disease was certainly consistent with findings from other studies. The association with thyroid hormones and antacids is interesting, though it is not clear that this has been pursued as an area of investigation.
An important contribution of AREDS was determining the rate of progression for the different stages of AMD, allowing clinicians to provide prognostic information, and facilitating further interventional studies. The probability of progression to advanced AMD in 5 years ranged from 1.3% in category 2 to 43% in category 4 (Table 1).10
The cataract portion of AREDS is often overlooked but actually represented an important rationale for the study. There was, however, no specific eligibility criteria regarding lens opacity, other than the media had to be clear enough to allow adequate fundus photographs as well as meeting the visual acuity requirements for entry. In addition, subjects were still eligible for enrollment if they had undergone prior cataract extraction except for those in category 1. A grading scheme for lens opacity was developed but may have limited applicability for subsequent clinical trials as it requires photographs on equipment that is now outdated; it is not applicable clinically.11 The baseline characteristics of the lens in AREDS showed that subjects with moderate nuclear opacities were more likely to be female, nonwhite smokers and to have large drusen.12 They were less likely to have obtained higher education, have a history of diabetes, or be taking nonsteroidal anti-inflammatory medications. Moderate cortical opacities were associated with dark iris color, large drusen, and weight change. Borderline significance was found for sunlight exposure and the use of thyroid hormones, and cortical opacities were less common in those with higher education.
The AMD outcomes chosen for the clinical trials in AREDS7 were
Progression to advanced AMD comparing antioxidants with no antioxidants.
Progression to advanced AMD comparing zinc with no zinc.
A 15-letter decrease in visual acuity score comparing antioxidants with no antioxidants.
A 15-letter decrease in visual acuity score comparing zinc with no zinc.
Progression of lens opacity or cataract surgery comparing antioxidants with no antioxidants.
Of the 4757 participants enrolled, 1117 were in category 1 with few if any drusen, all of which were smaller than 63 μm by definition. Subjects in category 1 were randomized to antioxidants or placebo, but because only 5 of the 1117 subjects progressed to advanced AMD, the numbers were insufficient to assess the effect of antioxidants on this group. The remaining 3640 participants were analyzed overall and within preplanned subgroups.10 Of the 3640 participants, 56% were women and 96% were white, with a median age of 69 years.
In the original study design, the case categories (2, 3, and 4) were to be pooled to assess the effects of the 3 different treatment regimens on progression to advanced disease and visual acuity loss. In the first published report of AREDS results, the authors moved quickly to a subgroup analysis for categories 3 and 4, for reasons given subsequently. The presentation of the results in this manner led to significant controversy and doubt regarding the study findings, including letters to the editor and heated discussions at professional meetings. However, the authors' subsequent explanations seem to address the concerns raised. In a response to a letter to the editor, the AREDS authors explained that the analysis of all 3 case categories was indeed statistically significant, when the predefined design variable of AMD category was included as a covariate in the model. In this analysis, the 3-df test for differential treatment effect was significant (P = .006), as were the zinc main effect (P = .009) and the treatment effect for 2 individual treatment arms, zinc alone (P = .006) and zinc plus antioxidants (P = .001).13 However, in category 2, only 15 events of progression to advanced AMD occurred across all treatment groups, with 3 of those occurring in the placebo group. For this reason, and because the investigators believed that it would be inappropriate to recommend supplements to patients who would rarely progress to advanced AMD, the AMD category–adjusted analysis for the full cohort was not presented in the published report.10 Instead, the published report focused on the benefit to categories 3 and 4. Subjects in categories 3 and 4 assigned to placebo had a 28% probability of progression to advanced AMD compared with 23% for antioxidants, 22% for zinc, and 20% for the combination of antioxidants and zinc.
In categories 3 and 4, risk of vision loss was 29% for those taking placebo, 26% for those taking antioxidants, 25% for those taking zinc, and 23% for those undergoing combination therapy. In 175 events of visual acuity loss in category 2, only 13 were thought to be attributable to AMD; there is no basis for recommending supplements at this stage. Although zinc and antioxidants plus zinc significantly reduced the odds of developing advanced AMD for those in category 3 and 4, only combination therapy significantly reduced the rates of at least moderate vision loss. Other secondary outcomes studied included visual acuity loss attributable to AMD, marked visual acuity loss, visual acuity loss in eyes with advanced AMD at baseline, and an analysis of development of neovascular AMD or central geographic atrophy.
Most of the AREDS safety findings were anticipated based on earlier work. Patients were warned in advance of kidney stones (secondary to ascorbic acid), genitourinary complications in men (secondary to zinc), fatigue, muscle weakness, decreased thyroid function, increased risk of hemorrhagic stroke (vitamin E), yellow skin (beta carotene), anemia, decreased high-density lipoprotein levels, and upset stomach (zinc). Supplementation in any of the groups showed no clinically or statistically significant difference in serum cholesterol or hematocrit concentrations from baseline.10 Yellow skin was seen more frequently in the antioxidant arm than in the nonantioxidant arm (8.3% vs 6%). Anemia (self-reported) occurred in subjects receiving zinc (13.2% vs 10.2%). During the course of the study, concerns were raised regarding the use of beta carotene in smokers, and the data and safety-monitoring committee recommended that smokers stop taking study medications containing beta carotene. However, none of the individual treatments reduced or increased mortality, and even when subjects were stratified by smoking status, there was no significant effect of the use of antioxidants alone for current smokers and a nonsignificant reduction in mortality for the combination. The small number of deaths from lung cancer showed no statistically significant difference by treatment.10
The AREDS developed a severity scale for use in clinical trials that uses a 9-point fundus photographic severity scale for AMD.14 The scale was refined using stereoscopic fundus photographs and grids on more than 3000 participants followed up for 5 years. It combines a 6-step drusen area scale with a 5-step pigmentary abnormality scale. The 5-year risk of advanced AMD increased progressively from less than 1% in step 1 to about 50% in step 9. The gradings were reproducible, with replicate gradings showing complete agreement in 63% and agreement within 1 step in 87% of eyes. As the investigators point out, progression along the scale may be a useful surrogate for progression to advanced AMD, creating a useful tool for epidemiologic and clinical trials. Drusen area was found to be more predictive than drusen size and forms the basis of the scale. The scale does have limitations, as its development was based primarily on non-Hispanic white individuals and it is not likely applicable to all populations.
The AREDS also developed a simplified severity scale for AMD for use by clinicians.15 Here, the investigators used drusen size rather than area, recognizing that this would be more easily adopted by clinicians. In this system, a score is developed for patients, rather than eyes. The clinician assigns to each eye 1 risk factor for 1 or more large (≥125 μm) drusen, and 1 risk factor for any pigment abnormality (hyperabnormality or hypoabnormality). Intermediate drusen in both eyes count as 1 factor. For eyes in which both are at risk, the risk factors are added, yielding a 5-step score (0-4). (The 5-year risk of progression is shown in Table 2 and ranges from 0.5% to 50%.) For patients with only 1 eye at risk, advanced AMD counts as 2 factors in the affected eye: 1 factor if the eye at risk has large drusen and 1 factor for pigment abnormality. Other risk factors for progression include smoking, family history, and supplement use but were not as important as the morphologic characteristics.
Along with analyzing the ocular risks for progression and the effect of treatment, AREDS investigators analyzed the risk of progression to neovascular AMD or central geographic atrophy in patients at risk in both eyes, looking at systemic factors and using multivariate models.16 Controlling for age, sex, and AREDS treatment group, the variables that were significantly associated with an increased incidence of neovascular AMD were white race (vs black) and a greater amount smoked (≥10 pack-years). There was an increased risk of central geographic atrophy in those with less education (≤high school graduate), higher body mass index, greater amount smoked, and antacid use. In persons at risk of developing advanced AMD in 1 eye, the incidence of neovascular AMD was associated with diabetes mellitus, and the incidence of central geographic atrophy was associated with the use of anti-inflammatory agents. Ophthalmologists can play an important role in educating patients about the risks of smoking, not only its role in disease development, but also its progression. The role of smoking and inflammation may provide important clues to the pathogenesis of AMD.
During the mean follow-up period of 6.5 years, 534 of 4752 AREDS participants (11%) died. The AREDS identified an increased risk of mortality with advanced AMD compared with those with few if any drusen (category 1), and most deaths resulted from diseases of the circulatory system.17 Advanced AMD was associated with cardiovascular deaths. In addition, those with visual acuity worse than 20/40 in 1 eye had increased mortality compared with those with good acuity in both eyes, though it is also worth noting that 91% of those with a visual acuity less than 20/40 in 1 eye at baseline had advanced AMD. Nuclear opacity and cataract surgery were associated with increased mortality (all-cause) and cancer deaths. Finally, participants randomized to zinc had lower mortality rates. These results underscore once again the interaction among cardiovascular disease, smoking, and AMD, clues to pursue in elucidating the pathogenesis of AMD.
After the publication of results that showed a benefit of supplementation in AREDS categories 3 and 4, AREDS investigators went on to estimate the impact of intervention in preventing progression of AMD and vision loss.18 Using the 2000 US Census, the Eye Disease Prevalence Study estimated the age-, sex-, and race-specific prevalence of intermediate and advanced AMD, and the AREDS investigators used the estimates for those aged 55 years and older. Other assumptions included an estimate of those with monocular advanced AMD of the total with advanced AMD based on findings from the Beaver Dam Eye Study, an estimate of the percentage with central geographic atrophy of the total with geographic atrophy, and estimates for races other than white, non-Hispanic white, and non-Hispanic black, for which direct prevalence data did not exist. Using these estimates, the study group calculated that approximately 8 million persons aged 55 years or older have monocular or binocular intermediate AMD or monocular advanced AMD and would be considered at a high risk of advanced AMD. Untreated, an estimated 1.3 million would be expected to develop advanced AMD, but with supplementation based on AREDS, more than 300 000 would avoid advanced AMD and any associated vision loss during the next 5 years. While the calculated number is based on many assumptions of prevalence and it is unlikely that all of the high-risk individuals would take the supplement because of other medical conditions, intolerance, or noncompliance, the analysis still demonstrates the potential benefit of intervention in AMD to prevent progression earlier in the disease. These findings provide important support for advocating intervention in high-risk individuals as an important aspect of health and public health measure. The findings have also fueled interest in designing clinical trials to test earlier interventions, including the Anecortave Acetate Risk-Reduction Trial (clinicaltrials.gov identifier: NCT00307398) and the Complications of Age-Related Macular Degeneration Prevention Trial.19
The cataract outcomes of AREDS are often forgotten in the discussions focused on AMD. The outcomes used for cataract evaluation were an increase in nuclear, cortical, or posterior subcapsular opacity grades or cataract surgery and at least moderate visual acuity loss from baseline. Of 4757 participants, 4629 were enrolled with at least 1 natural lens present and observed for an average of 6.3 years.20 No statistically significant effect of the antioxidant formulation was seen on the development or progression of age-related lens opacities. In addition, there was no statistically significant effect of treatment in reducing the risk of progression for any of the 3 lens opacity types or cataract surgery. In category 1, with no AMD, there was again no statistically significant difference noted between treatment groups for at least moderate visual acuity loss. Overall in AREDS, the risk of a lens event at 5 years was 30% regardless of treatment. An analysis of the effect of Centrum use similarly showed that there was no impact on the development of any lens opacity.21 There was no statistically serious adverse effect of treatment.20 It is possible that because the eligibility for enrollment in AREDS was an age of 55 years or older and the median age was 68 years, that the intervention was too late for the prevention of cataract progression.20 The risk of a lens event at 5 years is 30% within 5 years, regardless of treatment. Because two-thirds of study participants took Centrum as a multivitamin, this also could be analyzed; there was no impact on the development of any lens opacity.21
The AREDS investigated the risk of AMD progression secondary to cataract surgery in an indirect way and showed no increased risk of progression.22
The AREDS demonstrated the utility of the National Eye Institute Visual Functioning Questionnaire (NEI-VFQ) for monitoring progression of disease and vision loss.23,24 Those with advanced AMD in 1 or both eyes, severe nuclear sclerosis, reduced visual acuity, or cataract surgery had lower scores in general than those without eye disease. Interestingly, the NEI-VFQ score was unresponsive to lens opacity progression. Overall, the findings of the visual functioning questionnaire confirmed that it is a useful tool to assess the effect of AMD and vision loss and to follow the progression of both. In addition to the NEI-VFQ, the AREDS Cognitive Function Battery was administered to 2946 participants and demonstrated that after adjusting for variables such as age, sex, education, and other medical problems, reduced vision correlated with reduced mean cognitive function scores, even with as little visual acuity impairment as 20/40 OU.24
While subjects were still being enrolled into AREDS, results became available for 3 large randomized clinical trials of beta carotene: the Alpha-Tocopherol, Beta Carotene Lung Cancer Prevention Study (ATBC)25; the Physicians Health Study (PHS)26; and the Beta-Carotene and Retinol Efficacy Trial (CARET).27 The ATBC and CARET showed increased risk of lung cancer and mortality in smokers taking beta carotene; PHS showed no difference. After results for ATBC became available, with recommendation of the data safety monitoring committee, all AREDS participants were informed of the findings by newsletter, which told them that they were free to stop taking the study medication or to continue without change. Subsequently 9 of 11 centers gave their participants the choice of continuing, stopping, or being reassigned to a medication that did not contain antioxidants. The other 2 centers said that smokers had to stop beta carotene use or drop out of the study. Indeed there was a bump in the study “drop-out rate” during this 12-month period. In the recommendations arising from AREDS, the complete supplement is recommended to those who do not smoke or who have quit for 1 year or more.10 This instance points out the complicated oversight of a large, long-term prospective trial and the effect of other concurrent studies, as well as the importance of an independent data safety monitoring committee.
The AREDS has provided important information to clinicians, investigators, and patients. Grading systems and assessment tools, including quality of life assessment, were developed and tested in a large population. Natural history data that provide important prognostic information for patients and clinicians and formed the basis for new trials examining earlier intervention were produced for AMD and cataract. With the power of large numbers and careful design, AREDS clearly demonstrated smoking as a key risk factor for the development of AMD and for its progression. These same natural history data provide insight into the mechanism of disease and clues for further investigation. Mortality data reinforced the overlap of cardiovascular disease and AMD. Compiling the data that were current at the time of study design, investigators selected treatment regimens that demonstrated a benefit for patients with AMD and no effect on cataract. The AREDS treatment results provide guidance for subsequent trials of interventions, including vitamins and supplements as well as other agents.
Correspondence: Joan W. Miller, MD, Department of Ophthalmology, Harvard Medical School, Retina Service, Massachusetts Eye and Ear Infirmary, 243 Charles St, Boston, MA 02114 (firstname.lastname@example.org).
Submitted for Publication: January 9, 2008; final revision received June 29, 2009; accepted July 20, 2009.
Financial Disclosure: None reported.