Age-related Macular Degeneration and Antioxidant Status in the POLA Study FREE

IN WESTERN countries age-related macular degeneration (AMD) is the leading cause of blindness among older people.¹ With the ageing of our population, this burden is expected to increase dramatically. Currently, the only available treatment is laser photocoagulation, which has been proven useful only in a small percentage of patients and with limited benefit.² It is therefore urgent to determine factors that may lead to prevention of this affection.

The pathogenesis of this disease is poorly understood. It is probably multifactorial, involving, among others, genetic factors,^{3 - 4} an adverse effect caused by tobacco,⁵ and a possible association with atherosclerosis.⁶ In addition, oxidative stress may play an important role in this disease's etiology. Because of intense exposure to light and oxygen and a high polyunsaturated fatty acid content that is prone to lipid peroxidation, the retina, and particularly the macular area, is highly susceptible to oxidative stress.⁷ The retina contains a range of antioxidants that can inactivate free radicals. These antioxidant defenses include enzymes, such as selenium-dependent glutathione (GSH) peroxidase and related enzymes (superoxide dismutase, catalase) and antioxidant nutrients (carotenoids, vitamins E and C, GSH, etc).⁷ It is hypothesized that people with low amounts of antioxidants may be more prone to oxidative damage of the retina, and thus to AMD.

This hypothesis has received some support from 6 recent epidemiologic studies.^{8 - 13} However, the results have been inconsistent. Three studies found an inverse association with certain, although not always the same, carotenoids.^{11 - 13} In these studies the association between AMD and α-tocopherol, ascorbic acid, and retinol was not significant, while another study found a negative association between AMD and α-tocopherol, but not with β-carotene, ascorbic acid, or retinol.⁹ Finally, in the National Health and Nutrition Examination Survey, the intake of fruits and vegetables rich in vitamin A was lower in those subjects with AMD.⁸ Clearly, these epidemiologic data are encouraging but scarce and inconsistent. Other studies are needed to obtain a better understanding of the role that antioxidants play in the etiology of AMD.

The Pathologies Oculaires Liées à l'Age (POLA) survey was undertaken to study AMD, cataracts, and their etiologies in a large sample of older subjects from the south of France. We present the associations between AMD and antioxidant nutrients.

The POLA study is a prospective study taking place in Sète, a town of 40,000 inhabitants, located on the French Mediterranean. Its objective is to study age-related eye diseases (cataracts and AMD) and their risk factors. Inclusion criteria were (1) being a resident of Sète and (2) being aged 60 years or older on the examination day. According to the 1990 population census, there were almost 12,000 eligible residents from which our objective was to recruit 3000 participants. The population was informed of the study through the local media (television, radio, and newspapers). We also contacted 4543 residents individually by mail and telephone using the electoral roll. Between June 1995 and July 1997, we recruited 2584 participants, including 1133 men and 1451 women, with an average age of 70.4 years.

The baseline examination took place in a mobile unit, equipped with ophthalmologic devices: a projector of Snellen chart, using a decimal scale (L28 IR; Luneau, Chartres, France), an autorefractometer (RM-A7000; Topcon, Tokyo, Japan), a slitlamp (SL7F; Topcon), and a retinal camera (TRC 50 XF; Topcon). The mobile unit moved from one area to another to be in the proximity of the contacted participants.

Participants gave written consent for their participation in the study. The design of this study was approved by the Ethics Committee of Montpellier's University Hospital, Montpellier, France.

Four ophthalmologists (Louis Balmelle, MD, Jacques Costeau, MD, Jean-Luc Diaz, MD, and Fabienne Robert, MD) performed the ophthalmologic examinations. This examination included a record of ophthalmologic history (in particular lens extraction and year of the extraction); a measure of best-corrected far visual acuity in both eyes; after pupil dilation, a quantitative assessment of nuclear, cortical, and posterior subcapsular lens opacities at slitlamp examination according to the Lens Opacities Classification System III¹⁴ ; and one 50° color fundus photograph (Kodak Gold 100 ASA; Eastman Kodak Company, Rochester, NY) centered on the macular area in each eye.

After film processing, retinal photographs were scanned and digitalized. They were then recorded onto compact disks (Kodak procedure). Finally, for interpretation, photographs were examined on a 17-in (43-cm) computer screen. Total magnification was approximately ×31.5.

For grading the photographs, we used the definitions and grids of an international classification system.¹⁵ We also used the standard photographs of the Wisconsin Age-related Maculopathy Grading System¹⁶ to train the ophthalmologist and the technician in charge of the interpretation. Two levels of grading were carried out on the fundus photographs. After the main grid was superimposed on the photograph, a preliminary grading was performed by an ophthalmologist (Jean-Luc Diaz, MD). Second, for all participants for whom soft drusen or pigmentary abnormalities were coded as present anywhere on the photograph by the ophthalmologist, a detailed grading was performed by a specially trained technician (Sylvie Fourrey) using the international classification system.¹⁵

All lesions that were classified as geographic atrophy or neovascular macular degeneration by the ophthalmologist interpreting the photographs were discussed and adjudicated by 2 of us (Jean-Luc Diaz, MD and C. D.). In some cases we also asked the patient's ophthalmologist for additional information about the history of the lesion. The technicians conducting the grading were unaware of the antioxidant status of the participants or of the presence of any other risk factor.

Fundus photographs were not taken in 81 cases (3.1%): in 42 because of technical failure; in 9, refusal; in 5, contraindication for dilation; in 17, poor dilation or severe opacities of the lens or cornea; and in 8, poor cooperation. In addition, for 307 participants, photographs were ungradable in both eyes because of technical failure or the presence of opacities; gradable photographs were available for at least one eye in each of 2196 participants (85%).

Late AMD and early signs of AMD were defined according to the international classification system.¹⁵ Late AMD was defined by the presence of neovascular AMD or geographic atrophy within the grid (3000 µm from the foveola). Neovascular AMD included serous or hemorrhagic detachment of the retinal pigment epithelium or sensory retina, subretinal or subretinal pigment epithelium hemorrhages, and fibrous scar tissue. Geographic atrophy was defined as a discrete area of retinal depigmentation, 175 µm or larger, characterized by a sharp border and the presence of visible choroidal vessels.

Soft intermediate drusen were defined as drusen larger than 63 µm and smaller than or equal to 125 µm. Soft distinct and indistinct drusen were defined as large drusen (>125 µm) with uniform density and sharp edges or decreasing density from the center outwards and fuzzy edges, respectively. Areas of hyperpigmentation and hypopigmentation (without visibility of choroidal vessels) were also noted.

Biological measurements were made from fasting blood samples obtained at subjects' homes the morning of the examination and included measurements of plasma (cholesterol, triglycerides, and vitamins A, E, and C) levels and red blood cell GSH. Plasma triglyceride and total cholesterol levels were measured by routine enzymatic methods with a reagent purchased from Boehringer Laboratories (Boehringer Ingelheim, Ingelheim, Germany). Retinol and α-tocopherol levels were measured by high-performance liquid chromatography according to the method previously described by Catignani and Bieri.¹⁷ One sample of the liotrol mixture was measured in each series of measurements as an internal standard to prevent any shift during the study. Ascorbic acid levels were measured by high-performance liquid chromatography according to the method described by Tessier and Birlouez-Aragon.¹⁸ Red blood cell GSH levels were measured by colorimetric assay (Bioxytech GSH-400; OXIS International Inc, Portland, Ore).

Among the 2196 subjects with gradable photographs, 37 (1.7%) refused to obtain the blood sampling. In addition, retinol and α-tocopherol levels were not measured in 2 subjects because of technical failure, leaving 2157 subjects for the statistical analysis, 38 of whom were diagnosed with late AMD. Concerning red blood cell GSH measurements, 231 (10.7%) could not be made because of technical failure, leaving 1928 subjects for the analysis, 33 of whom were diagnosed with late AMD. Ascorbic acid levels were measured only in subjects recruited after November 20, 1995, ie, 1791 subjects, 33 of whom were diagnosed with late AMD.

Data were collected by trained study personnel who were unaware of the subjects' AMD status. A standardized interview was performed to assess (1) sociodemographic variables (marital status, level of education, major lifetime occupation, etc); (2) medical history (treated hypertension, cardiovascular diseases, diabetes, knee or hip arthritis, etc); (3) all medications currently used; (4) smoking history; and (5) occupational and leisure exposure to sunlight. The interviewer then measured height, weight, waist and hip circumferences, and systolic and diastolic blood pressures. Participants were considered to be diabetic in cases of known diabetes or if their fasting blood glucose level was higher than or equal to 1.4 g/L. Body mass index was calculated as weight in kilograms divided by the square of height in meters.

We determined the 20th and 80th percentile values for antioxidant nutrients in all subjects; these form 3 groups (low, middle, and high quintile). Using the low quintile group for reference, age- and sex-adjusted odds ratios (ORs) (and 95 % confidence intervals) were obtained by logistic regression, with the pathologic characteristic (late AMD, early signs of AMD) as the dependent variable and age, sex, and the middle and high quintile groups as the independent variables. Potential confounding variables were added to the model to obtain multivariate ORs. Tests for trends were performed by entering the antioxidant nutrient in the logistic regressions as a 3-category variable instead of 2 independent variables. All statistical analyses were done with SPSS (SPSS Inc, Chicago, Ill). The α-tocopherol–lipid ratio was calculated as α-tocopherol (in micromoles per liter) to cholesterol (in millimoles per liter)+triglycerides (in millimoles per liter).

As given in Table 1, using data from the 1990 census of Sète, our sample underrepresents persons aged 80 years and older (10.1% vs 19.8%). Sex distribution was similar in our sample and in the eligible population. Concerning the professional category, for men, our sample overrepresents middle and upper social classes (21.8% vs 10.4% for intermediate professions; 20.4% vs 9.8% for managers and executives) and underrepresents the lower social classes (10.0% vs 26.4% for office workers; 34.0% vs 40.4% for laborers). For women, unemployed housewives are underrepresented (26.4% vs 42.1%).

Among the 2157 subjects with data regarding AMD and antioxidant nutrients, we identified 38 subjects (1.8%) with late AMD (Table 2). Prevalence rates of late AMD were similar for both men and women, increasing sharply with age, from 0.4% between ages 60 and 69 years to 9.7% after age 80 years. Prevalence rates of early signs of AMD were also similar for both men and women and increased with age. Prevalence rates of soft intermediate, distinct, and indistinct drusen were 22.1%, 12.7%, and 2.3%, respectively. Prevalence rates of hyperpigmentation and hypopigmentation were 9.1% and 5.7%, respectively.

Our definitions of low, middle, and high quintiles of antioxidant nutrients are given in Table 3. As given in Table 4, after adjustment for age and sex, middle and high quintiles of α-tocopherol levels were associated with at least a 50% decrease in the prevalence of late AMD (OR, 0.50 and 0.42 for middle and high quintiles, respectively). However, this relationship was not significant (P=.07). The α-tocopherol–lipid ratio had a stronger association with late AMD. The prevalence of late AMD was decreased by 82% in the highest quintile. This relationship was significant (P=.003) and remained unchanged after further adjustment for potential confounding factors (level of education, body mass index, diabetes, history of treated hypertension, coronary heart disease, stroke, angioplasty, and smoking). Plasma retinol, plasma ascorbic acid, and red blood cell GSH levels did not show any significant associations with late AMD.

The association of the α-tocopherol–lipid ratio with early signs of AMD are given in Table 5. Middle and high quintiles of the ratio showed a decreased prevalence of early signs with ORs between 0.48 and 0.86. However, most of these associations were not significant. The α-tocopherol–lipid ratio was significantly associated with a reduced prevalence of any early sign of AMD (OR, 0.71 [95% confidence interval, 0.52-0.97] for high vs low quintile; P=.03). These results remained unchanged after adjustment for potential confounding variables.

The associations of the α-tocopherol–lipid ratio with AMD were not owing to higher lipid levels in late AMD (P=.67) or early signs of AMD (P=.98). However, to exclude any influence of lipid levels on our results, we adjusted further for cholesterol and triglyceride levels. This left the results unchanged (data not shown).

Our study shows a strong association between late AMD and plasma α-tocopherol levels; the prevalence of late AMD was decreased by 82% in the highest quintile of the α-tocopherol–lipid ratio (P=.003). High levels of the α-tocopherol–lipid ratio were also associated with a decreased prevalence of early signs of AMD. No associations were found with ascorbic acid, retinol, and red blood cell GSH values.

It is now recognized that the plasma α-tocopherol level should be expressed in terms of its concentration within lipids or lipoproteins.^{19 - 24} Indeed, the metabolism of vitamin E is closely related to lipids, since it is transported by lipoproteins.²⁵ Its correlation coefficients with cholesterol and triglyceride levels are usually close to 0.50,^{23 - 24} as is also observed in our study (data not shown). In fact, in overt vitamin E deficiency with neuromuscular symptoms, vitamin E status is predicted only by lipid-standardized plasma vitamin E values but not by absolute ones.^{26 - 27}

To our knowledge, this is the first study on the relationship between AMD and lipid-standardized α-tocopherol levels. Not standardizing for lipids adds random misclassification and thus biases the estimates toward the null. This is observed in our study since the association of late AMD with plasma α-tocopherol levels does not reach statistical significance, while its association with the α-tocopherol–lipid ratio is highly significant.

Results from other studies are consistent with this finding: 2 studies found a reduction of the risk of AMD close to 50% in subjects with high levels of plasma α-tocopherol, which was significant in one study⁹ and not in the other (P=.10).¹³ The remaining 3 small case-control studies did not show any significant difference in the level of plasma α-tocopherol among AMD cases and controls, but they had limited statistical power.^{11 ,28 - 29} Moreover, 2 of these studies used nonfasting blood samples.^{11 ,28} Some of these studies provided adjustment for plasma cholesterol levels, but none for plasma triglyceride levels. Our study shows that lipid standardization using levels of cholesterol and triglycerides strengthens the association between AMD and α-tocopherol, giving rise to these significant associations.

In contrast with the above findings, among the 4 available studies, only one found an association between dietary vitamin E and the incidence of large drusen,³⁰ the others finding no associations between dietary or supplemental vitamin E and AMD.^{9 - 10 ,12} However, plasma levels of vitamin E are poorly related to vitamin E dietary intakes.^{31 - 32} The metabolism of vitamin E is complex, involving, among others, the metabolism of lipids and a specific transfer protein.²⁵ Subjects with high dietary vitamin E intakes may therefore be deficient in vitamin E, in particular because of lipid malabsorption or transfer protein deficiency.²⁵ Plasma levels of α-tocopherol are also influenced by dietary intakes of other vitamins, in particular vitamin C and carotenoids.³³ Plasma levels of α-tocopherol may therefore be more precise indicators of the vitamin E status of the retina.

Concerning the use of vitamin supplements, cross-sectional studies may be biased. Patients with late AMD may start taking supplements in the hope of slowing the progression of the disease. This is supported by the fact that, in the Eye Disease Case-Control Study,¹² starting vitamin E supplements within the preceding year was 5 times more frequent in patients with neovascular AMD than in controls. This may also have led to an underestimation of the association between late AMD and levels of plasma vitamin E in that study. Prospective studies are needed to better assess the relationship between AMD and vitamin E supplements.

Our study has several limitations. First, our sample underrepresents older persons and overrepresents the middle and upper social classes, in comparison with the whole eligible population. This may have affected the prevalence of AMD or the distribution of antioxidants. Furthermore, using self-selection may have resulted in different motivations in persons with symptomatic eye disease (in particular late AMD) and persons with no known eye disease, overrepresenting health-conscious individuals in the last group. To adjust for this possible selection bias, multivariate analyses included adjustments for level of education and body mass index. These multivariate adjustments did not modify the association between AMD and α-tocopherol levels. The fact that we also found an association with early signs of AMD favors a true association. Since early signs of AMD are asymptomatic in most people, a selection bias is unlikely. Finally, our sample is probably not highly biased, since the prevalence rates of late AMD and early signs of AMD are similar to those of previous studies using the same grading protocol.^{34 - 36} However, a selection bias cannot be ruled out, which might have led to underestimation or overestimation of the association between AMD and α-tocopherol levels.

In observational studies the concern is always about confounding factors. Since atherosclerosis is linked to α-tocopherol³⁷ and may also be associated with AMD,⁶ we adjusted for factors related to cardiovascular disease: diabetes, hypertension, coronary heart disease, stroke, and angioplasty. Adjustment for these factors left the results unchanged. The levels that appear to help prevent AMD in this study (an α-tocopherol–lipid ratio>4.1, similar to α-tocopherol–cholesterol ratio>5.0) are close to those associated with minimum cardiovascular risk (α-tocopherol–cholesterol ratio>5.2).³⁸

We also adjusted for smoking, which is strongly associated with the risk of AMD both in our study³⁹ and in previous studies.^{40 - 44} Further adjustment for smoking left the results unchanged. To our present knowledge, there is no other essential confounding factor that may explain this relationship.

Since this study is cross-sectional, we cannot assume that low α-tocopherol levels preceded the development of AMD. It is possible that participants with late AMD have modified their diet after developing visual impairment because of difficulties with shopping and cooking. Once again, the fact that we observed a similar association with the early, asymptomatic signs of the disease is in favor of a true "causal" relationship. These results must, however, be confirmed in prospective studies. This will be carried out by our study, with a follow-up visit, 3 years after baseline.

Consistent with the other studies,^{9 ,13 ,28} we found no associations between AMD and plasma ascorbic acid and retinol levels. In our study no associations were found with levels of red blood cell GSH.

In conclusion, our study shows a strong inverse association between AMD and levels of plasma α-tocopherol. Other, in particular prospective, studies are needed to confirm our results. Finally, any effect vitamin E has in preventing AMD can only be proven by randomized double-blind interventional studies.

Accepted for publication June 22, 1999.

This study was supported by the Institut National de la Santé et de la Recherche Médicale, by grants from the Fondation de France, Department of Epidemiology of Ageing, by financial support from Rhône Poulenc, Essilor, and CERIN, by grants from the Fondation pour la Rercherche Médicale, Paris, by grants from the Région Languedoc-Roussillon, Montpellier, and by grants from the Association Retina-France, Toulouse, France.

Corresponding author: Cécile Delcourt, INSERM Unité 500, 39 Ave Charles Flahault, 34093 Montpellier, CEDEX 5, France (e-mail: delcourt@u500.montp.inserm.fr).