Risk of Age-Related Macular Degeneration in Eyes With Macular Drusen or Hyperpigmentation: The Blue Mountains Eye Study Cohort

Objective  To quantify the 5-year risk of age-related macular degeneration (AMD) in eyes with different macular drusen characteristics (ie, size, type, location, and total area) or hyperpigmentation in a population-based cohort.

Methods  The Blue Mountains Eye Study examined 3654 residents during 1992-1994; 2335 (75.1% of survivors) were reexamined during 1997-1999. Retinal photographs were graded using the Wisconsin Age-Related Maculopathy Grading System. Incident AMD lesions were defined by development of neovascular AMD or geographic atrophy in eyes without these lesions at baseline (eyes at risk). Age-adjusted relative risks (RRs) were determined. Generalized estimating equation models were used to estimate odds ratios, adjusting for the correlation between eyes and other AMD risk factors.

Main Outcome Measure  Incidence of AMD.

Results  Of the 4634 eyes at risk, 52 (1.1%) developed neovascular or atrophic AMD lesions over 5 years. In right eyes, presence vs absence of the following macular signs predicted AMD: drusen that were 125 µm or larger (13.9 vs 0.6%; age-adjusted RR, 5.7; 95% confidence interval [CI], 3.6-9.0), indistinct soft or reticular drusen (23.2% vs 0.4%; RR, 9.9; 95% CI, 6.4-15.4), total drusen area of half the disc area or more (31.4% vs 0.6%; RR, 13.5; 95% CI, 8.0-22.8), and hyperpigmentation (14.4% vs 0.5%; RR, 8.0; 95% CI, 5.4-11.9). After adjusting for age, sex, and smoking status, eyes with these signs at baseline had a high likelihood of developing AMD. Eyes with Age-Related Eye Disease Study categories 3 and 4 were 5 times more likely to develop AMD compared with eyes in categories 1 and 2.

Conclusion  This study quantifies the 5-year risk of AMD in eyes with macular drusen and hyperpigmentation.

LARGE OR SOFT macular drusen and retinal pigmentary abnormalities are known from case series to predict development of neovascular or atrophic lesions, the 2 late forms of age-related maculopathy (ARM), ^1-6 termed age-related macular degeneration (AMD) in the International ARM Classification.⁷ Soft macular drusen, with the histopathological correlate of basal laminar deposits, together with retinal pigmentary changes have been identified as antecedents of retinal pigment epithelial cell dysfunction and failure, which lead to either of the late forms of AMD.⁵

This finding has also been confirmed in recent population-based cohort studies of older persons.^8-10 Eyes with indistinct soft macular drusen or retinal pigmentary abnormalities were shown in the Beaver Dam Eye Study to have a substantially higher 5-year risk of incident AMD than eyes without these lesions.^8,9 Increasing drusen size, area of involvement, and the presence of indistinct soft drusen progressively increased the risk of AMD.⁸ Larger drusen area and the presence of retinal pigmentary changes predicted 2-year incident AMD in the Rotterdam Study.¹⁰

This report from the Blue Mountains Eye Study (BMES) aims to extend findings provided in these 2 studies by quantifying 5-year AMD risk from a range of baseline macular drusen characteristics and hyperpigmentation, independent of known risk factors, such as age, sex, and smoking status. These data complement our data on the 5-year incidence of ARM lesions in the Blue Mountains population.¹¹

The BMES is a population-based survey ofvision and common eye diseases in an urban population aged 49 years or older and residing in 2 postcodes of the Blue Mountains region, west of Sydney, Australia.^12,13 Of the 4433 eligible residents, baseline examinations (BMES I) were performed on 3654 subjects (82.4%) during 1992-1994. Five-year follow-up eye examinations(BMES II) were conducted during 1997-1999. After excluding 543 persons who died since BMES I, 2335 of the survivors (75.1%) participated in BMES II. The study was approved by the Western Sydney Area Health Service Human Research Ethics Committee. Signed informed consent was obtained from participants at both examinations. A detailed medical and ocular history was taken by questionnaire, and a comprehensive eye examination was conducted after pupil dilatation.

Stereoscopic retinal photographs (30°) of the macula and other retinal fields¹² of both eyes were taken using a fundus camera (Zeiss FF3; Carl Zeiss, Oberkochen, Germany). Photographs of both eyes were obtained in 98% and of at least 1 eye in 99% of participants at both examinations. The photographic grading system for ARM lesions used in the BMES has been previously reported¹² and closely follows descriptions of the Wisconsin Age-Related Maculopathy Grading System.¹⁴

Grids with a radius of 3000 µm, centered on the fovea, were used to define the macula of each eye.^12,14 These were provided by Ronald Klein, MD, MPH, University of Wisconsin–Madison. Grading identified the presence of neovascular or atrophic AMD, together with hyperpigmentation and macular drusen characteristics, including the type, maximum size, location, and total area covered by drusen. An initial masked assessment of incident AMD lesions was followed by a side-by-side grading approach, as recently described, ¹¹ similar to that described by the 5-year report from the Beaver Dam Eye Study.⁸

Neovascular AMD and geographic atrophy were defined as described by the International ARM Classification.⁷ Lesions classified as neovascular AMD included serous or hemorrhagic detachment of the retinal pigment epithelium or sensory retina, presence of subretinal or sub–retinal pigment epithelial hemorrhages, or subretinal fibrosis and photocoagulation scars in participants previously documented as having neovascular AMD. Geographic atrophy was defined as a discrete area, at least 175 µm in diameter, of retinal depigmentation characterized by a sharp border and the presence of visible choroidal vessels. In this report, cases with signs of neovascular AMD and widespread geographic atrophy were classified as positive for both subtypes.

All participants with AMD lesions at either examination were adjudicated by a retinal subspecialist (P.M.), and, for the present report, eye-specific data were analyzed. Incident neovascular AMD was defined as the appearance at follow-up of any signs of neovascular AMD⁷ in eyes at baseline without these signs and without geographic atrophy involving the foveal center. Incident geographic atrophy was defined as the appearance at follow-up of at least one geographic atrophy area 175 µm or greater in diameter at the macula (regardless of foveal involvement) in eyes in which neither AMD lesion was present at baseline.

We also assessed AMD risk by baseline ARM category, using categories developed for the Age-Related Eye Disease Study (AREDS).¹⁵ Subcategories A and B under categories 3 and 4 were amalgamated. In brief, we used the following category definitions, which were as close as possible to the AREDS categories, and the grading circles from the Wisconsin grading system:

Category 1: No drusen or drusen less than 63µm with a total area of less than 125 µm in diameter (<C-1 circle).

Category 2: Drusen less than 63 µm with a total drusen area of 125 µm or more in diameter, drusen 63 µm or more but less than 125 µm with a total drusen area of less than 372µm in diameter (<I-2 circle), drusen 63 µm or more but less than 125 µm with no indistinct soft drusen and a total drusen area of less than 660 µm in diameter (<O-2 circle), or presence of retinal pigmentary changes at the macula.

Category 3:Drusen 63 µm or more but less than 125 µm in diameter with indistinct soft drusen and a total drusen area of 372 µm or more in diameter (≥I-2 circle), drusen 63 µm or more but less than 125 µm with no indistinct soft drusen and a total drusen area of 660 µm or more in diameter (≥O-2 circle), drusen 125 µm or more, or presence of noncentral geographic atrophy.

Category 4: Second eye of participants with AMD.

Eye-specific data were analyzed using SAS statistical software (SAS Institute Inc, Cary, NC). Age-adjusted relative risks (RRs) with 95% confidence intervals (CIs) were calculated for right and left eyes separately, using the Mantel-Haenszel method across 10-year age group strata. Adjusted odds ratios (ORs) with 95% CIs were then estimated for right and left eyes combined, using generalized estimating equation (GEE) models to account for the correlation between the 2 eyes.^16-18 Known ARM risk factors or potential confounders adjusted for in the GEE models included baseline age, sex, and current smoking status.

Table 1 shows a comparison of the baseline characteristics of surviving BMES I participants examined in BMES II (n = 2335) with those who were not examined in BMES II (n = 776) or those who died before BMES II (n = 543). Participants in BMES I who had moved from the area (n = 383) or who refused to participate in BMES II (n= 393) were more likely at baseline to have been older, to have had difficulty walking, and to have been current smokers but were less likely to own their home or to have received a trade certificate or higher qualifications after leaving school. Those who died before BMES II were significantly different from those examined in BMES II for most characteristics.

Of the 4670 eyes (2335 BMES II participants), 33 had established AMD lesions that were evident during BMES I, 2 eyes (1 subject) had pseudoxanthoma elasticum with choroidal neovascularization, and 1 eye had peripapillary choroidal neovascularization. After excluding these 36 eyes, of the remaining 4634 eyes(2321 subjects) considered at risk for the development of AMD lesions, 53 eyes from 39 persons (34/2300 with both eyes at risk and 5/21 with 1 eye at risk because the first eye already had AMD at BMES I) developed these lesions during the 5-year follow-up period (1.1%). Of the 34 persons with both eyes at risk, 14 developed either AMD lesion in both eyes. Thirty of 53 eyes developed neovascular AMD, and 28 developed geographic atrophy, including 5 that developed both lesions.

Table 2 shows eye-specific incidence rates for the development of either AMD type and age-adjusted RRs for both right and left eyes. These estimates were similar for the 2 eyes. A number of baseline macular drusen characteristics and hyperpigmentation predicted the development of AMD in eyes with vs those without the following signs: large (≥125 µm) drusen (RR, 5.7), indistinct soft drusen (RR, 9.9), location of drusen within a 1500 µm radius of the foveal center(RRs, 7.0-12.7), area involved by large drusen equal to at least one half that of the optic disc (RRs, 9.2-13.5), and presence of retinal hyperpigmentation(RRs, 8.0-8.4). Using a modified AREDS classification, ¹⁵ eyes classified as having category 3 or 4 ARM at baseline were 3.9 times more likely to develop AMD lesions during the 5-year period, compared with eyes classified as having category 1 or 2 ARM.

Table 3 presents adjusted ORs for the development of either AMD type in eyes with baseline macular drusen and hyperpigmentation, after adjusting for age, sex, smoking status, and the correlation between eyes in GEE models. Many baseline macular drusen characteristics and hyperpigmentation predicted the development of AMD in eyes with vs those without these signs. Using a modified AREDS classification, ¹⁵ eyes classified as having category 3 or 4 at baseline were significantly more likely to develop AMD lesions during the 5-year period, compared with eyes classified as having category 1 or 2 (adjusted OR, 5.1). The adjusted OR for neovascular AMD was higher (7.2).

Age-related macular degeneration is the most frequent cause of irreversible visual impairment and blindness among older people in Western countries.^19-21 Identifying persons(or eyes) at high risk of developing AMD is a critical strategy in reducing blindness from this disease. Persons at high risk could be advised to undertake regular self-testing of their vision with an Amsler chart and to seek treatment within a week if any warning symptoms develop (eg, metamorphopsia, scotoma, or sudden worsening of vision in one eye). Accurate knowledge of the level of increased risk conferred by the presence of different macular drusen characteristics or retinal pigmentary abnormalities, independent of the principal AMD risk factors, would assist considerably in this regard.

In this report, we have estimated the 5-year risk of AMD in eyes with drusen or retinal hyperpigmentation, using data from an older population-based cohort. We could have underestimated or overestimated these risks, owing to a loss to follow-up of 25% of the surviving cohort. Baseline participants who did not return for the 5-year examination were older and more likely to have been current smokers, compared with those who did return. However, the baseline prevalence rates of early ARM between these 2 groups were similar(4.7% among the 2335 examined and 4.3% among the 776 not examined; P = .30). We assume that participants lost to follow-up could have been more likely to develop AMD, given that they were older and were more likely to have been current smokers.²² Therefore, we could have underreported the 5-year AMD incidence.¹¹ Given the nondifferential losses to follow-up in relation to baseline early ARM prevalence (Table 1) and adjustment for age and smoking in our statistical models, it seems unlikely that we have substantially underestimated or overestimated the risks of developing AMD in eyes with early ARM lesions.

Apart from the many clinical case series that have reported the relationship between macular drusen or retinal pigmentary abnormalities and an increased risk for development of AMD, ^1-6 the Beaver Dam Eye Study was the first comprehensive population-based study to provided detailed crude data on these links.⁸ The earlier Watermen Study report assessed incidence in a selected sample of men.²³ In the Beaver Dam report on the 5-year incidence and progression of ARM, increasing maximum drusen size, larger drusen area, presence of indistinct soft drusen, and retinal pigmentary changes at baseline were all associated with an increased incidence and RR of AMD lesions. However, the age-adjusted RR of incident AMD for eyes with compared with those without these characteristics was not provided. The Beaver Dam report documented that the incidence of large drusen, indistinct soft drusen, retinal pigmentary changes, and either AMD lesion were all strongly age-related, ⁸ as did our own report on the 5-year incidence of age-related maculopathy lesions.¹¹

Although our findings cannot be broadly compared with the Beaver Dam Eye Study incidence report, ⁸ some specific comparisons can be made using crude incidence estimates. For example, the Beaver Dam 5-year AMD incidence in right eyes with indistinct soft drusen of 6.5% compares with a much higher rate in our data of 23.2% in right eyes and 18.3% in left eyes. Similarly, the calculated Beaver Dam AMD incidence in right eyes with drusen 125 µm or more of 5.6% is also lower than our incidence of 13.9% in right eyes and 10.9% in left eyes. The 5-year AMD incidence in right eyes with retinal pigmentary changes (7.0% in the Beaver Dam Eye Study) is closer to, but still lower than, our incidence of 14.4% in right eyes and 10.8% in left eyes.

Our recent article, which pooled risk factor data for AMD lesions, ²⁴ suggests that it is unlikely that we overdiagnosed any of the eyes that developed late-stage disease. Our study population was slightly older at baseline than the Beaver Dam Eye Study population. This could explain the fact that we observed more eyes with incident AMD after 5 years in our study population (n = 52) than were reported in the larger Beaver Dam population (n = 48).⁸ For neovascular lesions in right eyes, the 5-year incidence was 0.7% in our study compared with 0.4% (95% CI, 0.2%-0.6%) in the Beaver Dam Eye Study population. After age-standardization to the Beaver Dam Eye Study population, our incidence rate was 0.5% (95% CI, 0.3%-0.9%). For geographic atrophy in right eyes, a similar difference was found: 0.7% in our study compared with 0.2% (95% CI, 0.05%-0.35%) in the Beaver Dam study. After age-standardization to the Beaver Dam Eye Study population, our incidence rate was 0.5% (95% CI, 0.3%-0.9%).¹¹ For both AMD lesions, 95% CIs from the 2 studies overlapped.

There is no ready explanation for these differences. However, the age-specific prevalence of some early ARM lesions (particularly large or soft drusen) was lower in our study¹² than in the Beaver Dam Eye Study.²⁵ It is possible that that we used tighter criteria to define the presence of soft drusen than were used in the Beaver Dam Eye Study.

The Rotterdam Study report on the 2-year incidence and progression of ARM, found that the most important predictors for progression were presence of drusen involving more than 10% of the macular area or presence of retinal pigmentary changes (either depigmentation or hyperpigmentation) at the macula.¹⁰ However, the very small number of incident AMD cases(n = 12) in this study limited its potential to examine risk associated with individual early lesions.

Because right and left eyes were highly correlated, we assessed separately the age-adjusted RRs of the development of AMD lesions associated with particular baseline lesions in right and left eyes. The Mantel-Haenszel method used for this purpose provides relatively robust estimates. However, this method is not so readily applied when adjustment for more than 2 variables is needed. To adjust for other AMD risk factors (eg, sex and smoking), we used the GEE modeling approach, after combining the data from both eyes. Although ORs estimated using GEE models can be considered somewhat equivalent to RRs in this particular situation, in that the incidence of AMD lesions is relatively rare (<2% overall in the population), ORs can more easily inflate the apparent risk as a result of small numbers, compared with RRs estimated using the Mantel-Haenszel method. Therefore, the ORs presented in Table 3 were very imprecise, as is reflected in the wide CIs, and do not indicate the magnitude of increased risk. Given the relatively small number of eyes from this older general population that developed incident AMD over 5 years, the estimated magnitudes of both the RRs and ORs are likely to be imprecise.

In this study, we assessed the 5-year risk of developing AMD in eyes with early ARM lesions, after adjusting for age, sex, and smoking. However, we did not adjust for all early ARM lesions simultaneously in the same model because they are likely to be correlated. Therefore, these estimated risk magnitudes for different types of early ARM lesions are not independent of one another, so they cannot be interpreted in an additive way for eyes with more than 1 early ARM lesion type.

Despite such limitations, these data provide some useful predictive guides for ophthalmologists. The finding of large drusen at the posterior pole increases the likelihood of AMD developing during the next 5 years by a factor of 5, compared with persons of the same age without this sign. A finding of indistinct soft drusen or the presence of any large drusen occupying an area half that of the optic disc increases the risk by 10-fold, and hyperpigmentation at the posterior pole increases the risk of AMD by 8-fold.

In summary, this study of a population-based cohort confirms previous clinical impressions that eyes with large drusen, indistinct soft drusen, a large area of the macula covered by drusen, or hyperpigmentation have a substantially higher risk of developing AMD lesions. These predictors were quantified independent of known risk factors for AMD, including age, sex, and smoking.

Corresponding author and reprints: Paul Mitchell, MD, PhD, FRACO, Department of Ophthalmology, University of Sydney, Hawkesbury Road, Westmead, Australia 2145 (e-mail: paulmi@westgate.wh.usyd.edu.au).

Submitted for publication March 19, 2002; final revision received December 23, 2002; accepted January 17, 2003.

This study was supported by grant 974159 from the Australian National Health and Medical Research Council, Canberra.