Nutritional Epidemiology

Vitamin C Is Associated with Reduced Risk of Cataract in a Mediterranean Population1,2 Marı´a Pastor Valero,*3 Astrid E. Fletcher,* Bianca L. De Stavola,* Jesu´s Vioque† and Vicente Chaque´s Alepuz** *London School of Hygiene and Tropical Medicine, London, UK; †Dpto. Salud Pu´blica.: Universidad Miguel Herna´ndez, Elche, Spain; and **Centro de Especialidades de Burjassot, Valencia, Spain ABSTRACT Cataract is an important visual problem of older people and a substantial health care cost in many countries. Most studies investigating risk factors for cataract have been conducted in the United States, and there is less information on the possible role of dietary factors in European populations. We conducted a case– control study to investigate the association of antioxidant vitamins (vitamin C, vitamin E, vitamin A, ␤-carotene, ␣-carotene, ␤-cryptoxanthin, lycopene, zeaxanthin and lutein) and minerals (zinc and selenium) and risk of cataract in a Mediterranean population. Cases with cataract (343) and 334 age/sex frequency-matched controls aged 55 to 74 y were selected from an ophthalmic outreach clinic in Valencia, Spain. Participants were interviewed about their diet using a Food Frequency Questionnaire, and other information on potential confounders, such as smoking, alcohol, and education. Blood samples were analyzed by a colorimetric method for vitamin C and by reversed-phase HLPC for other blood antioxidants. Blood levels of vitamin C above 49 ␮mol/L were associated with a 64% reduced odds for cataract (P ⬍ 0.0001). Dietary intake of vitamins C, E and selenium were marginally associated with decreased odds (P ⫽ 0.09, P ⫽ 0.09, P ⫽ 0.07, respectively), whereas moderately high levels of blood lycopene (⬎0.30 ␮mol/L) were associated with a 46% increased odds of cataract (P ⫽ 0.04). Our results strengthen the evidence for a protective role for vitamin C on the aging lens as this effect was seen in a population characterized by high vitamin C intakes. J. Nutr. 132: 1299 –1306, 2002. KEY WORDS:



cataracts



antioxidant micronutrients

Although there is some evidence that low levels of antioxidant micronutrients are associated with an increased risk of cataract, there is considerably less agreement on the role of individual antioxidants or the benefits of supplementation (1,2). Most epidemiological studies have been conducted in the United States and far less is known for European populations with different diet and lifestyle patterns. Across Europe, there is considerable variation in dietary intakes, especially in the consumption of fresh fruit and vegetables, with Northern European populations having lower intakes compared with Mediterranean diets. Supplement use is also less common in older Europeans compared with the United States. In Finland, low levels of ␣-tocopherol and ␤-carotene were associated with an increased risk of cataract surgery (3), but no benefit was found from supplementation with these vitamins in a large randomized trial in male smokers (4). An Italian case– control study found no association with a range of nutritional factors, although there was some evidence for a protective antioxidant



vitamin C



lycopene



humans

effect as measured by an overall index (5). In contrast, a strong association of antioxidant enzymes and plasma retinol with cataracts were observed in the POLA study in southern France (6,7). We carried out a study in a rich fruit-and-vegetable-growing Mediterranean region to investigate whether higher levels of antioxidant vitamin and mineral micronutrients were protective for cataract, and to investigate whether sunlight exposure, smoking and alcohol consumption, and history of severe episodes of diarrheal illness were risk factors for cataract. This article presents the results found for the antioxidant micronutrients (vitamin C, vitamin E, vitamin A, ␤-carotene, ␣-carotene, ␤-cryptoxanthin, lycopene, zeaxanthin and lutein) and minerals (zinc and selenium) taking into account the potential confounding effects of other variables. MATERIALS AND METHODS Study design. The study was carried out in Burjassot, a town located within the province of Valencia and situated on the Mediterranean east coast of Spain. The target population consisted of the residents of the catchment area of the primary health care center of the town [Centro de Especialidades de Burjassot (CEB)].4 The center

1 Financial support for the study was received from the Nutricia Foundation, The Netherlands. Dr. Pastor was supported by an award from the Fondo de Investigaciones Sanitarias, Spanish Ministry of Health. 2 Pastor M, Fletcher A, De Stavola B, Vioque J, Chaque´s V. Vitamin C and cataract results from a Spanish population Annals of Nutrition & Metabolism Abstracts 2001, 45(suppl. 1): 1– 636. Oral presentation: 17th International Congress of Nutrition, Vienna Austria, August 27–31, 2001. 3 To whom correspondence should be addressed. E-mail: [email protected].

4 Abbreviations used: CEB, Centro de Especialidades de Burjassot; CI, confidence interval; FFQ, Food Frequency Questionnaire; FPP6, Food Processor Plus Version 6; LOCS II, Lens Opacification Classification System II; OR, odds ratio.

0022-3166/02 $3.00 © 2002 American Society for Nutritional Sciences. Manuscript received 4 October 2001. Initial review completed 29 November 2001. Revision accepted 29 November 2001. 1299

1300

VALERO ET AL.

belongs to the National Health Care System and serves 176,085 inhabitants. The CEB is a primary referral health care center for all individuals living in the same geographical area. The majority of the population in the Valencia province uses the National Health Care System as the main medical service for cataract problems. A case– control study was implemented over a 14-mo period with cases and controls drawn from patients attending the ophthalmology outpatient clinic at the CEB. Most patients were referred for ophthalmologic checkup by their general practitioners. Others, already seen by the ophthalmologists, were coming back for routine scheduled visits. Subjects. A case was any patient between the ages of 55 and 74 y, diagnosed with nuclear, cortical, posterior subcapsular or mixed cataract (any combination of these) in at least one eye and of grade ⱖ 1 according to the Lens Opacification Classification System II system version II (LOCS) (8). Controls were frequency-matched on age and gender, and both eyes had lens opacity grading LOCS II ⫽ 0. Age frequency-matching was carried out in 5-y age groups (55–59, 60 – 64, 65– 69 and 70 –74) i.e., for each case, a control was matched within the same 5-y age group. To assess the role of usual dietary habits on cataract risk, patients were excluded if they: were following special diets (e.g., patients with diabetes, ulcerative colitis, Crohn’s disease or undergoing treatment for cancer); or had intraocular pressure ⬎ 21mm Hg; or had the following conditions or medical treatments, which are known to be associated with an increased risk of cataract (e.g., primary angle glaucoma treated with pilocarpine; myopia equal or greater than 6 diopters; and treatment for psoriasis). Cases were also excluded if they had a congenital or traumatic cataract or cataract associated with systemic diseases. Patients who satisfied the basic eligibility criteria for the study were then informed about the study and consent to participate was obtained. Ethical approval was obtained from the Ethical Committee of the London School of Hygiene & Tropical Medicine and CEB. Data collection. Data were collected by a trained study team composed of ophthalmologists, interviewers, nurses and the medical coordinator of the study. The ophthalmic examination included a slit lamp assessment of the eyes and dilation of pupils with one drop of tropicamide (1%) and one drop of phenylephrine (10%). After 15 min, if the dilatation of pupil was insufficient, the same procedure was repeated once more. The LOCS II system was used to classify cases by type and grade of cataract and to ensure that controls had no evidence of lens opacities. Each participant’s lens status was independently classified by two medical observers (one of the clinic’s ophthalmologists and the coordinator) and agreement reached on classification and diagnosis. The ophthalmologists participating in the study had considerable experience in the diagnosis of cataract and were trained in the use of the LOCS II system. At a second visit, experienced fieldworkers from the Valencia Institute of Public Health interviewed the participants. Fieldworkers were trained by the study coordinator (M.P.) in the administration of the questionnaires. They were not given information on the case– control status of the participants or the study hypothesis under investigation. Data collected by the interviewers included information on occupation, education level, history of smoking and alcohol consumption, occurrence of severe episodes of diarrhea, sunlight exposure, dietary history (see below) and current and past use of vitamin supplements. All participants were weighed and measured twice by the same interviewer using standard procedures. Dietary intake was collected by means of a Food Frequency Questionnaire (FFQ), using a Spanish version of the Harvard questionnaire (9). The Spanish version was modified and validated in the same region from which the population of this investigation was drawn (10). The questionnaire was expanded to cover more fully the types of vegetables and fruits consumed by the Spanish population (11,12). The nutrition analysis system Food Processor Plus version 6 (FPP6) (13) was used to convert into nutrients the consumption of food items collected by the FFQ. Because the FPP6 did not contain information about the food content for the six carotenoids under investigation (␤-carotene, ␣-carotene, ␤-cryptoxanthin, lycopene, zeaxanthin and lutein), the original FPP6 database was expanded to include data obtained from a Spanish study that analyzed the carotenoid content of the most consumed fruits and vegetables in Spain (14).

In the morning on the same day of the interview, blood samples were taken at ambient temperature, under subdued light and overnight fasting conditions in the laboratory of the CEB. A thorough protocol was designed to collect, transport and measure the blood samples for vitamin C (ascorbic acid) analysis. Samples were wrapped in tin foil and placed in insulated dry containers at 4°C to exclude light and, therefore, avoid vitamin C degradation. The samples were quickly transported to the laboratory (⬍ 10 min from the CEB). All vitamin C biochemical analyses were completed after a maximum of 3 h from their extraction. A simple colorimetric method for the determination of ascorbic acid in blood was used (15). An equivalent protocol was developed for the extraction, handling and transportation of the blood samples for the analyses of vitamin E (␣-tocopherol), A (retinol), ␤-carotene, ␣-carotene, ␤-cryptoxanthin and lycopene. These samples were stored at ⫺70°C degrees and sent monthly to London to be analyzed using a reversed-phase HLPC method (16,17).

Statistical methods Study power. The sample size for this study was calculated on the basis of the expected distribution of blood levels for vitamin E and ␤-carotene in the control population (14,18). A total of 377 cases and an equal number of controls were required to detect: an odds ratio (OR) of 1.8 or greater with 80% power and 5% significance, if 12% of the controls had a blood vitamin E level below 20 ␮mol/L and an OR of 1.7 or greater if the control prevalence of a low ␤-carotene level (⬍ 0.15 ␮g/d) was 16%. Statistical analysis. The blood and dietary data were treated separately. Variables in both groups were logarithmically transformed because of their skewed distributions and the dietary data were adjusted according to the total energy intake model described by Willett (9). These standardized values were then categorized into quintiles according to the distribution of the controls, with the lowest quintile treated as the reference group. The association between cataract risk for each quintile was separately evaluated in terms of OR using age- and sex-adjusted logistic regression models with overall significance and linear trends examined using likelihood ratio tests (19). Where appropriate, the quintiles of each micronutrient were merged into binary or tertiary variables if the OR were similar and the 95% confidence intervals (CI) overlapped. Multiple logistic regression was carried out separately for the diet and blood antioxidants using backward and forward stepwise procedures. The significance of all variables was examined using a Wald test. If one or more variables were associated with P values ⬎ 0.15, the variable that had the highest P value was dropped from the model. The model was estimated again and the significance of the remaining variables reexamined using the same method as above. This iterative procedure continued until only variables with a significance level of at least 0.15 remained. The effect of potential confounders or effect modifiers, such as years of sunlight exposure, episodes of severe diarrhea, smoking and alcohol history, and education, was examined by including these factors sequentially into the two logistic regression models for the dietary and blood data and examining whether they changed the original OR estimates and by testing for their interaction with the antioxidant nutrients. The final models were refitted on the subsets of cases with either nuclear, posterior, cortical or mixed cataracts and all the controls, to explore whether any of the identified risk factors were important for each type of cataract. In the analyses by type of cataract we used two classifications of subtype of cataract. The first classification was any cortical, posterior or nuclear cataract in one eye, or in both, irrespective of whether it was a pure or a mixed cataract. The second classification (pure cataract) was a more restrictive one and depended on only one type of cataract being present in one eye or in both eyes (if both were affected). These analyses were only explorative and no P values are given. All statistical analyses were performed using Stata 6 software (20).

RESULTS A total of 347 cases and 345 controls, frequency-matched on sex and age, were recruited. Nobody invited to participate

VITAMIN C IS ASSOCIATED WITH REDUCED RISK OF CATARACT

1301

TABLE 1 Socio-demographic and lifestyle characteristics of cases and controls1 Cases, n ⫽ 343 n Men Age group, y 55—59 60—64 65—69 70—74 Marital status Single Married/cohabiting Widowed Divorced Occupational situation Full/part time employed Unemployed Housewife Retired/others Unknown Education None/primary school only Secondary school Higher education Smoking status Never smoked Ex-smoker Current smoker Unknown Alcohol intake, g/d None ⱕ2 ⬎2—ⱕ11 ⬎11 Unknown Sunlight years2 All daylight hours From 1000 to 1600 h

Controls, n ⫽ 334 %

n

%

P value 0.33 0.9

149

(43.4)

133

(39.9)

40 82 107 114

(11.6) (24.0) (31.2) (33.2)

40 84 98 112

(12.0) (25.2) (29.3) (33.5)

21 252 63 7

(6.1) (73.5) (18.4) (2.0)

13 253 63 5

(3.9) (75.8) (18.8) (1.5)

18 12 168 143 2

(5.3) (3.5) (49.0) (41.7) (0.6)

19 11 174 130 0

(5.7) (3.3) (52.2) (39.1) —

293 40 10

(85.4) (11.7) (2.9)

256 55 23

(76.6) (16.5) (6.9)

208 73 62 0

(60.6) (21.0) (17.9) —

206 74 47 7

(61.7) (22.1) (14.1) (2)

188 40 55 59 1

(54.8) (11.7) (16.0) (17.2) —

181 51 51 51 0

(54.2) (15.3) (15.3) (15.3) —

0.6

0.9

0.01

0.4

0.55

0.30 0.64

0.08, 0.72 0.25, 1.37

0.27 0.56

0.10, 0.67 0.22, 1.35

0.26 0.51

1 Values are proportions in each category (except for sunlight exposure which is median plus inter quartile range) and P values are given for overall test of heterogeneity between cases and controls or for differences in medians. 2 Sunlight years adjusted for protective headware and use of glasses, sunglasses or contact lenses.

refused but 4 cases and 11 controls did not attend the interviews or blood collection and, therefore, were excluded from the study. Table 1 shows the sociodemographic and lifestyle characteristics of the cases and controls included in the study. The two groups were comparable on the two matching variables with similar age distributions and mean ages (66.3 y in cases and 66.4 y in controls) and gender proportions (43% of cases were male and 40% of control were male). Cases and controls were also similar on marital status and current occupational status. We had hypothesized that other potential risk factors would differ between cases and controls and, therefore, could confound any association with antioxidant status and cataract. However, neither current nor previous smoking or alcohol intake or outdoor exposure differed significantly between cases and controls. Educational level was significantly lower in cases than in controls with fewer having attended higher education and more having received education up to primary school level only. Table 2 presents the distribution of the different types of cataract diagnosed among the cases. The most frequent type of cataract was pure nuclear (n ⫽ 101) followed by pure posterior (n ⫽ 63). Among the controls the most frequent diagnoses were: conjunctival, corneal or episcleral minor inflammation or infection (8.4%); age-related

macular degeneration (8.4%), chronic open angle glaucoma (6.3%); amaurosis (4.8%); minor noninflammatory lid disorders (3.9%); and pterygium (3.6%); lacrymal disorders (2.6%); and other diagnoses including thrombosis, amblyopia, strabism, posterior detachment of vitreous (12%). Fifty percent were diagnosis free. Of these, the majority of them were first

TABLE 2 Distribution of cases by type of cataract Type of cataract Pure total Nuclear Cortical Posterior Mixed total Nuclear-cortical Nuclear-posterior Posterior-cortical Posterior-cortical-nuclear

n

(%)

202 101 38 63 141 33 45 63 0

(58.6) (29.0) (11.3) (18.3) (41.4) (9.9) (13.2) (18.3) (–)

VALERO ET AL.

1302

visits and most of them were sent to the optometrist for spectacles for refractive errors. The univariate analyses were adjusted for age, sex and energy intake for dietary variables and season of blood collection for blood analyses (Tables 3 and 4) Increasing dietary intakes of vitamin C showed an inverse association with cataract risk (P ⫽ 0.04) with intakes of ⬎ 135 mg daily being associated with the greatest benefit. For other dietary antioxidants there was no, or only weak, evidence of a linear trend, for example, ␣-carotene (P ⫽ 0.07). Table 4 shows the equivalent results for the blood antioxidants. Analyses of blood antioxidants were carried out both including and excluding 9 cases and 23 controls who regularly took vitamin and/or mineral supplements. Because the results were similar, we present only results obtained after excluding these subjects. Levels of ascorbic acid ⬎ 49 ␮mol/L showed a strong protective effect with respect to levels ⱕ 49 ␮mol/L (P ⬍ 0.0001 for trend among the five groups). The dietary antioxidants selected by the stepwise procedures to be retained in the final model included vitamin E, selenium and vitamin C, whereas the most important blood antioxidants selected by the stepwise procedures were ascorbic acid, retinol and lycopene (Table 5). These antioxidants were selected by both the backward and forward stepwise procedures. The OR in the multiple logistic models were essentially similar to those in the univariate analyses, suggesting that the effects of the antioxidants were independent of each other. The exception to this was lycopene, where the OR for a

possible adverse effect increased in magnitude and statistical significance in the multiple logistic model. In addition, none of the effects were confounded by pack-years of cigarette consumption, patterns of alcohol consumption, sunlight years of exposure, body mass index or diarrheal illness. Education was a mild confounder for dietary antioxidants, but not for blood antioxidants. No significant interaction effects were found between any of these factors. Similar results for the associations of interest were found when data were analyzed by type of cataract (Tables 6 and 7), and when analyses were repeated by pure type of cataract (results not shown). DISCUSSION Our results from the blood analyses suggest that levels above 49 ␮mol/L of ascorbic acid were significantly associated with a 64% reduced odds of cataract. The P value for the association was ⬍ 0.0001, both in the univariate analyses and after adjustment for potential confounders including education and in a model that included other antioxidants. This association is, therefore, highly unlikely to be due to chance and is not due to differences in educational level between cases and controls. It is, however, plausible that lower educational levels might lead to poorer diets with lower antioxidant levels. Although animal experiments have shown an important role for a protective effect of vitamin C against cataract (22,23), epidemiological studies have not shown consistent evidence of a link between dietary vitamin C or serum/plasma

TABLE 3 Associations of dietary levels of antioxidants and risk of cataract in a Mediterranean population1,2 Dietary antioxidant Vitamin C, mg/d OR (95% CI) Vitamin E, mg/d OR (95% CI) Retinol, RE/d OR (95% CI) Zinc, mg/d OR (95% CI) Selenium, mg/d OR (95% CI) ␤-Carotene, ␮g/d OR (95% CI) ␣-Carotene, ␮g/d OR (95% CI) ␤-Cryptoxanthin, ␮g/d OR (95% CI) Lutein, ␮g/d OR (95% CI) Zeaxanthin, ␮g/d OR (95% CI) Lycopene, ␮g/d OR (95% CI)

P value for trend2

Quintiles ⱕ102 1 ⱕ9 1 ⱕ942 1 ⱕ8.5 1 ⱕ109 1 ⱕ1135 1 ⱕ56 1 ⱕ78 1 ⱕ443 1 ⱕ39 1 ⱕ1167 1

⬎102–135 0.88 (0.56, 1.40) ⬎9–10.4 0.54 (0.33, 0.89) ⬎942–1337 0.86 (0.54, 1.39) ⬎8.5–9.4 0.61 (0.38, 0.99) ⬎109–123 1.46 (0.92, 2.33) ⬎1135–1932 1.02 (0.65, 1.62) ⬎56–177 0.84 (0.53, 1.33) ⬎78–144 0.96 (0.60, 1.54) ⬎443–699 0.88 (0.54, 1.42) ⬎39–66 1.13 (0.70, 1.82) ⬎1167–1588 0.87 (0.54, 1.40)

⬎135–164 0.66 (0.41, 1.07) ⬎10.4–12.4 0.91 (0.57, 1.44) ⬎1337–1779 1.01 (0.63, 1.62) ⬎9.4–10.2 0.71 (0.45, 1.14) ⬎123–132 0.85 (0.51, 1.39) ⬎1932–2884 0.70 (0.43, 1.15) ⬎177–342 0.81 (0.51, 1.30) ⬎144–247 0.94 (0.59, 1.51) ⬎699–993 0.98 (0.61, 1.56) ⬎66–95 0.91 (0.56, 1.49) ⬎1588–2387 1.11 (0.69, 1.78)

1 OR adjusted for sex, age and energy intake, estimated using logistic regression. 2 ␹-squared test for linear trend among the five quintile-specific odds (1 df).

⬎164–212 0.60 (0.37, 0.97) ⬎12.4–17.7 0.75 (0.47, 1.20) ⬎1779–2282 0.63 (0.39, 1.04) ⬎10.2–11.4 0.89 (0.57, 1.42) ⬎132–143 0.98 (0.60, 1.60) ⬎2884–4251 0.69 (0.42, 1.12) ⬎342–891 0.68 (0.42, 1.11) ⬎247–305 0.76 (0.47, 1.24) ⬎993–1383 0.69 (0.40, 1.09) ⬎95–142 1.06 (0.66, 1.71) ⬎2387–3037 0.78 (0.48, 1.27)

⬎212 0.70 (0.44, 1.13) ⬎17.7 0.77 (0.48, 1.24) ⬎2282 0.82 (0.50, 1.03) ⬎11.4 0.64 (0.40, 1.04) ⬎143 0.97 (0.60, 1.58) ⬎4251 0.82 (0.51, 1.33) ⬎891 0.64 (0.39, 1.04) ⬎305 0.97 (0.61, 1.56) ⬎1383 1.00 (0.64, 1.64) ⬎142 0.99 (0.61, 1.60) ⬎3037 1.11 (0.69, 1.78)

0.04 0.60 0.21 0.31 0.34 0.15 0.07 0.41 0.78 0.85 0.81

VITAMIN C IS ASSOCIATED WITH REDUCED RISK OF CATARACT

1303

TABLE 4 Associations of blood levels of antioxidants and risk of cataract in a Mediterranean population1,2 Blood antioxidant Ascorbic acid, ␮mol/L OR (95% CI) Retinol, ␮mol/L OR (95% CI) ␣-Tocopherol, ␮mol/L OR (95% CI) ␤-Cryptoxanthin, ␮mol/L OR (95% CI) ␣-Carotene, ␮mol/L OR (95% CI) ␤-Carotene, ␮mol/L OR (95% CI) Lycopene, ␮mol/L OR (95% CI)

Quintiles ⱕ49 1 ⱕ1.9 1 ⱕ26 1 ⱕ0.18 1 ⱕ0.02 1 ⱕ0.16 1 ⱕ0.18 1

⬎49–57 0.44 (0.27, 0.70) ⬎1.9–2.21 0.69 (0.42, 1.15) ⬎26–31 0.77 (0.47, 1.28) ⬎0.18–0.32 0.80 (0.50, 1.30) ⬎0.02–0.04 0.73 (0.44, 1.22) ⬎0.16–0.28 1.11 (0.68, 1.81) ⬎0.18–0.30 1.00 (0.60, 1.66)

⬎57–62 0.40 (0.24, 0.65) ⬎2.21–2.49 0.81 (0.50, 1.32) ⬎31–36 0.74 (0.45, 1.22) ⬎0.32–0.46 0.69 (0.42, 1.13) ⬎0.04–0.05 0.95 (0.58, 1.57) ⬎0.28–0.41 1.00 (0.61, 1.64) ⬎0.30–0.48 1.47 (0.90, 2.40)

P value for trend ⬎62–71 0.31 (0.18, 0.50) ⬎2.49–2.80 0.56 (0.34, 0.95) ⬎36–41 0.99 (0.61, 1.60) ⬎0.46–0.64 0.49 (0.29, 0.83) ⬎0.05–0.10 0.94 (0.57, 1.56) ⬎0.41–0.66 1.00 (0.60, 1.66) ⬎0.48–0.71 1.17 (0.71, 1.94)

⬎71 0.30 (0.18, 0.51) ⬎2.80 1.27 (0.79, 2.06) ⬎41 0.93 (0.56, 1.52) ⬎0.64 0.91 (0.56, 1.49) ⬎0.10 0.87 (0.52, 1.46) ⬎0.66 1.05 (0.63, 1.78) ⬎0.71 1.09 (0.65, 1.83)

⬍0.0001 0.47 0.88 0.32 0.89 0.97 0.56

1 OR adjusted for sex, age and season of blood collection estimated using logistic regression. 2 ␹-squared test for linear trend among the five quintile-specific odds (1 df).

ascorbic acid and cataract risk. Studies using dietary assessments including FFQ have found an association between vitamin C intake and cataract risk (24,25), a weak one (26), or no association (27,28). Results on blood ascorbic acid are also inconsistent with no association reported in some studies TABLE 5 Associations of dietary and blood antioxidants with cataract in a Mediterranean population1,2

Dietary antioxidants Vitamin E, mg/d ⱕ9 9, 10 ⬎10 Vitamin C, mg/d ⱕ135 ⬎135 Selenium, mg/d ⱕ123 ⬎123 Blood antioxidants Ascorbic acid, ␮mol/L ⱕ49 ⬎49 Retinol, ␮mol/L ⱕ2.50 2.50, 2.80 2.80, 7.99 Lycopene, ␮mol/L ⱕ0.30 ⬎0.30

OR

95% CI

P value

1 0.64 0.85

— (0.38, 1.05) (0.57, 1.25)

— 0.09 0.41

1 0.76

— (0.55, 1.05)

— 0.09

1 0.74

— (0.54, 1.02)

— 0.07

1 0.34

— (0.23, 0.50)

— P ⬍ 0.0001

1 0.64 1.52

— (0.39, 0.99) (1.02, 2.35)

— 0.05 0.04

1 1.43

— (1.02, 2.06)

— 0.04

1 Odds ratios (OR) and 95% confidence intervals (CI) obtained from multiple logistic regression models which also included age, sex, smoking, energy intake, alcohol intake and education and, additionally for the blood analyses, season of blood collection. 2 Two tailed P values based on Wald tests.

(6,29), while a study in India found an increased risk with higher intakes (30). Such studies that have reported a positive association have suffered from methodological problems. An 11-fold risk of posterior subcapsular cataract with low plasma vitamin C was described in one small study but was not significant (31). The NHANES II study (32) found a strong inverse association between serum ascorbic acid and self-reported cataract extraction. However, none of the other large cohort studies (26,27) have analyzed blood vitamin C. Longterm supplement use of vitamin C has been inversely associated with cataract risk in the Nurses study (33) and with decreased risk of cataract incidence in the Beaver Dam Eye Study (34), although no effect of vitamin C supplement was observed in the Physicians’ Study (35) or in the Age-Related Eye Disease Study (36). In this study, blood retinol levels from 2.50 to 2.80 ␮mol/L showed a decreased risk with respect to lower levels (OR ⫽ 0.64), whereas levels above 2.80 ␮mol/L showed a reverse effect (OR ⫽ 1.52). These findings are similar to those from the Baltimore Longitudinal Study on Ageing (29), showing that dietary vitamin A was not associated with risk of cataract, whereas middle and high blood retinol levels were associated with an increased risk. Our results based on the FFQ data showed that elevated intakes of vitamin E were marginally associated with a decreased risk of cataract (P ⫽ 0.09). However, the epidemiological evidence for such an association based on dietary data is inconsistent. Three studies found an association with cataract risk (25,28,31), whereas two studies found no association (27,30). Our results for selenium were similar to those for vitamin E. Although animal experiments show evidence for a protective role of selenium (37), to our knowledge, there are no epidemiological studies examining the individual role of selenium intake and risk of cataract to make comparisons. Our finding of a significant adverse effect of moderately high levels of blood lycopene (⬎ 0.30 ␮mol/L) mainly found in tomatoes, watermelon and cherries was not anticipated. The role of lycopene in the etiology of cataract

VALERO ET AL.

1304

TABLE 6 Dietary antioxidants and associations with type of cataract in a Mediterranean population1 Nuclear (n ⫽ 176) Variable Vitamin E, mg/d ⱕ9 (9, 10] ⬎10 Vitamin C, mg/d ⱕ135 ⬎135 Selenium, mg/d ⱕ123 ⬎123

Cortical (n ⫽ 131)

Posterior (n ⫽ 168)

OR1

(95% CI)

OR1

(95% CI)

OR1

(95% CI)

1 0.44 0.81

(0.23, 0.84) (0.51, 1.28)

1 0.74 1.00

(0.37, 1.45) (0.59, 1.72)

1 0.96 1.16

(0.52, 1.78) (0.71, 1.90)

1 0.56

(0.38, 0.82)

1 0.92

(0.60, 1.40)

1 0.75

(0.51, 1.05)

1 0.71

(0.48, 1.04)

1 0.88

(0.58, 2.46)

1 1.03

(0.70, 1.51)

1 OR and 95% confidence intervals (CI) obtained from a multiple logistic regression model that included age, sex and all the variables listed in the table.

has only been investigated in one epidemiological study where high dietary intake and blood levels of lycopene were found to be associated with severe nuclear sclerosis (38 –39). The lycopene levels found in this study are within the range found in other Spanish studies (40). However, carotenoids others than those with provitamin A activity (e.g., food additives, food substitutes, pesticides or products formed in the processing of tomatoes, watermelon or cherries) might have been measured through the blood lycopene levels and might have affected our results. Neither lutein nor zeaxanthin was associated with cataract in our study, although higher intakes of both these carotenoids have been shown to have a protective effect in two large cohort studies (41,42). We did not have sufficient financial support to examine the plasma levels of these carotenoids, which may have been a more reliable measure than the FFQ. In general, the results from the FFQ data showed a smaller effect of antioxidants on cataract risk compared with the blood data. The smaller magnitude of the OR based on the FFQ antioxidant levels may have resulted from the error in the measurement of usual dietary intake with the questionnaire, whereas data obtained from blood analyses are likely to have been less susceptible to error. Participants may have had difficulty in accurate classification of intakes across a wide range of food products (103 food items with 9 possible cate-

gories of frequency). The resulting error in measuring antioxidant intakes from the FFQ would be expected to lead to a dilution of the magnitude of association (so called regression dilution bias), a well-known phenomenon in epidemiological studies. Of more concern is recall bias, where errors in reporting food intakes occur because of biased reporting due to knowledge of being a case. However, recall bias of diet between cases and controls seems unlikely. Doctors in Spain do not give dietary advice to patients with cataracts, and there is no obvious population awareness of possible dietary influences on cataract risk. Cases and controls were specifically asked whether they had changed their diets recently and if so, when and why. No person reported either changes in diet due to eye problems or increasing consumption of fruits and vegetables for any other reason. Moreover, participants in the study did not know the objectives under investigation. Attempts were also made to minimize interviewer bias by masking observers to disease status and the objectives under study. Other sources of errors might result from inaccurate food composition data and/or incomplete listing of foods leading to errors in estimates of the average individual energy and nutrient intake. In our study, analyses were performed using energy adjusted and nonadjusted nutrients. Results were similar, suggesting that if errors in estimating energy intake occurred, they would not have altered our nutrient estimates.

TABLE 7 Blood antioxidants and associations with type of cataract in a Mediterranean population1 Nuclear (n ⫽ 176)

Ascorbic acid, ␮mol/L ⱕ49 ⬎49 Retinol, ␮mol/L (0.37, 2.50) (2.50, 2.80) (2.80, 7.99) Lycopene, ␮mol/L (0, 0.30) (0.30, 2.52)

Cortical (n ⫽ 131)

Posterior (n ⫽ 168)

OR1

(95% CI)

OR1

(95% CI)

OR1

(95% CI)

1 0.32

(0.21, 0.51)

1 0.38

(0.23, 0.64)

1 0.36

(0.23, 0.57)

1 0.73 1.67

(0.41, 1.27) (1.02, 2.72)

1 0.60 1.82

(0.32, 1.15) (1.09, 3.08)

1 0.80 1.22

(0.47, 1.39) (0.73, 2.03)

1 1.55

(1.00, 2.38)

1 1.20

(0.76, 1.90)

1 1.34

(0.87, 2.07)

1 OR and 95% confidence intervals (CI) obtained from a multiple logistic regression model that included age, sex and all the variables listed in the table.

VITAMIN C IS ASSOCIATED WITH REDUCED RISK OF CATARACT

Our results may also have been diluted by inclusion of 28 cases of macular degeneration among the controls because dietary factors are also considered to have a role in the etiology of this age-related eye condition. Controls were chosen to be typical of the background population from which the cases were drawn. The danger of excluding controls on the basis that they may share similar risk factors as cases is that this may bias the OR by over estimating the effect of the dietary exposures. We preferred to take a more conservative option that might have led to an underestimate. Differential referral patterns can also be a source of bias in case/control studies; for example, if our cases were selectively different from all cataract cases in the area. We think this is unlikely because the majority of the population in the Comunidad Valenciana uses the National Health Care System as the main medical service for cataract problems. Finally, a lack of association in the current study between certain antioxidants and risk of cataract might be due to a lack of statistical power. For example, although in the multiple logistic model the P value for the effect of high levels of ␤-carotene was 0.22, the estimated OR was equal to 0.81 and the 95% CI indicated a range of possible values from 0.59 to 1.13. This might apply also to the results for other antioxidants (e.g., dietary retinol and zinc). Our interpretation of our study results is that OR less than unity are indicative of a protective effect for certain antioxidants. We think it unlikely that these OR reflect possible adverse effects of higher antioxidant levels in the controls. The controls were not selected on the basis of any particular diagnosis, indeed 50% of controls were diagnosis free and only 8.4% had minor ocular inflammation or infection such as mild conjunctivitis. We conclude that a diet rich in vitamin C may be protective against cataracts. The Valencia diet is rich in citrus fruits and results in high intakes of vitamin C. The mean vitamin C intakes in our study population (mean ⫽ 157 mg/d, SD ⫽ 77.05, median ⫽ 141) and reported elsewhere for a southern Spanish population (43) are considerably higher than those reported for other populations with a similar age range (18,44,45). In this study mean blood ascorbic acid levels in our population were 54.55 ␮mol/L in men compared with the U. S. means of 36.33 to 42.59 ␮mol/L (46,47), and Japan’s means of 29.52 ␮mol/L (48). Similar comparisons apply to the values for women and for dietary intake of vitamin C. Nonetheless, even in this high intake population, those with blood levels of ascorbic acid ⬍ 49 ␮mol/L (29%) and dietary intakes of ⱕ 135 mg/d (45%) were at increased risk of cataract. People with the lowest risk of cataract (intakes ⬎ 135 mg daily) were taking twice the current recommended daily intakes for Spain and the United States of 60 mg/d (45,49). These findings have important implications for the prevention of cataracts, which are a major health service burden in many countries (15% to 20% of the older population) (50) and add support to the current debate to increase vitamin C intake (51). ACKNOWLEDGMENTS We are indebted to the ophthalmologists and nurses for their dedicated work: Marı´a Isabel Ferna´ dez de Co´ rdova Martı´nez, Consuelo Arroyo Bermu´ dez, Lucrecia Aguilar Valenzuela, Juan Marı´n Montiel, Luı´s Perez Varona, Antonio Duch Samper, and the nurses: Mercedes, Ana, and Rosa; to R Wormald (Moorfields Eye Hospital, London), for training in LOCS II; to R. Farre, A. Frigola and D. Gimeno for the vitamin C analyses (Valencia University, Spain); to the fieldworkers; to the Valencia Institute of Public Health, IVESP; to P. Shetty and the late E. Wheeler (Public Health Nutrition Unit, London School of Hygiene & Tropical Medicine) for nutritional advice. We thank the participants who made this study possible.

1305

LITERATURE CITED 1. Bunce, G. (1995) The role of nutrition in cataract. In: Diseases of the Lens: Duane’s Clinical Ophthalmology. Vol. 1. (Jaeger, E. A. & Tasman, M. D., eds.). Lippincott-Raven, Philadelphia, PA. 2. Christen, W. G., Glynn, R. J. & Hennekens, C. H. (1996) Antioxidants and age-related eye disease: current and future perspectives. Ann. Epidemiol. 6: 60 – 66. 3. Knekt, P., Heliovaara, M., Rissanen, A., Aromaa, A. & Aaran, R. (1992) Serum antioxidant vitamins and risk of cataract. Br. Med. J. 305: 1392–1394. 4. Teikari, J. M., Raulatahti, R., Haukka, J., Jarvinen, P., Hartman, A. M., Virtamo, J., Albanes, D. & Heinonen, O. (1998) Incidence of cataract operations in Finnish male smokers unaffected by ␣-tocopherol or ␤-carotene supplements. J. Epidemiol. Commun. Health. 52: 468 – 472. 5. The Italian-American Cataract Study Group. (1991) Risk factors for age-related cortical, nuclear and posterior subcapsular cataracts. Am. J. Epidemiol. 133: 541–553. 6. Delcourt, C., Cristol, J. P., Tessier, F., Leger, C. L., Michel, F. & Papoz, L. (2000) Risk factors for cortical, nuclear and posterior subcapsular cataracts: the POLA study. Am. J. Epidemiol. 151: 497–504. 7. Delcourt, C., Cristol, J. P., Leger, C. L., Descomps, B. & Papoz, L. (1999) Associations of antioxidant enzymes with cataract and age-related macular degeneration: the POLA study. Ophthalmology 106: 215–222. 8. Chylack, L., Leske, M., McCarthy, C., Khu, P., Kashiwagi, T. & Sperduto, R. (1989) Lens opacities classification system II (LOCSII). Arch. Ophthalmol. 107: 991–997. 9. Willett, W. C., Sampson, L., Stampfer, M. J., Rosner, B., Bain, C., Witschi, J., Hennekens, C. H. & Speizer, F. E. (1985) Reproducibility and validity of a semiquantitative food frequency questionnaire. Am. J. Epidemiol. 122: 51– 65. 10. Vioque, J. & Gonzalez, L. (1991) Validity of a food frequency questionnaire (preliminary results). Eur. J. Cancer 1: 19. 11. Granado, F., Olmedilla, B., Blanco, I. & Rojas-Hidalgo, E. (1992) Carotenoid composition in raw and cooked Spanish vegetables. J. Agric. Food Chem. 40: 2135–2140. 12. Olmedilla, B., Granado, F., Blanco, I. & Rojas-Hidalgo, E. (1992) Determination of nine carotenoids, retinol, retinyl palmitate and ␣-tocopherol in control human serum using two internal standards. Food Chem. 45: 205–213. 13. ESHA Research. (1994) The Food Processor Plus Computer Program, Version 6.0., Salem, OR. 14. Olmedilla, B., Granado, F., Blanco, I. & Rojas-Hidalgo, E. (1993) Quantification of provitamin-A carotenoids in the fruits most commonly consumed in Spain. In: Food and Cancer Prevention: Chemical and Biological Aspects (Waldron, K. W., Johnson, I. T. & Fenwich, G. R., eds.), pp. 141–145. Royal Society of Chemistry, Cambridge, UK. 15. Kyaw, A. (1978) A simple colorimetric method for ascorbic acid determination in blood plasma. Clin. Chim. Acta 86: 153–157. 16. Thurnham, D., Smith, E. & Flora, P. (1988) Concurrent liquid-chromatographic assay of retinol, ␣-tocopherol, ␤-carotene, ␣-carotene, lycopene and ␤-cryptoxanthin in plasma, with tocopherol, acetate as internal standard. Clin. Chem. 34: 377–381. 17. Rochi, E., Borghi, A., Paolillo, F., Pradelli, M. & Casalgrandi, G. (1991) Carotenoids and liposoluble vitamins in liver cirrhosis. J. Lab. Clin. Med. 2: 176 –185. 18. Gregory, J., Foster, K., Tyler, H. & Wiseman, M. E. (1990) The Dietary and Nutritional Survey of British Adults. Office of Population Censuses and Surveys. Social Survey Division. Commissioned jointly by the Ministry of Agriculture, Fisheries and Food and the Department of Health. HMSO, London, UK. 19. Clayton, D. & Hill, M. (1993) Statistical Methods in Epidemiology. Oxford University Press, Oxford, UK. 20. Stata Corporation (1999) Stata, Version 6.0. Stata Corporation, TX. 21. Jacques, P., Chylack, L., McGandy, R. & Hartz, S. (1988) Antioxidant status in persons with and without senile cataract. Arch. Ophthalmol. 106: 337– 340. 22. Varma, S., Srivastava, V. & Richards, R. (1982) Photoperoxidation in lens and cataract formation: preventive role of superoxide dismutaste, catalase and vitamin C. Ophthalmic Res. 14: 167–175. 23. Varma, S. & Richards, R. (1988) Ascorbic acid and the eye lens. Ophthalmic Res. 20: 164 –173. 24. Jacques, P., Chylack, L., McGandy, R. & Hartz, S. (1988) Antioxidant status in persons with and without senile cataract. Arch. Ophthalmol. 106: 337– 340. 25. Leske, M., Chylack, L. & Wu, S. (1991) The lens opacities case-control study: risk factors for cataract. Arch. Ophthalmol. 109: 244 –251. 26. Mares-Perlman, J., Klein, B., Klein, R. & Ritter, L. (1994) Relation between lens opacities and vitamin and mineral supplement use. Ophthalmology 101: 315–325. 27. Hankinson, S. E., Stampfer, M. J., Seddon, J. M., Colditz, G. A., Rosner, B., Speizer, F. E. & Willett, W. C. (1992) Nutrient intake and cataract extraction in women: a prospective study. Br. Med. J. 305: 335–339. 28. Tavani, A., Negri, E. & La Vecchia, C. (1996) Food and nutrient intake and risk of cataract. Ann. Epidemiol. 6: 41– 46. 29. Vitale, S., West, S., Hallfrisch, J., Alston, C., Wang, F., Moorman, C., Muller, D., Singh, V. & Taylor, H. R. (1993) Plasma antioxidants and risk of cortical and nuclear cataract. Epidemiology 4: 195–203. 30. Mohan, M., Sperduto, R. D., Angra, S. K., Milton, R. C., Mathur, R. L., Underwood, B. A., Jaffery, N., Pandya, C. B., Chhabra, V. K. & Vajpayee, R. B.

1306

VALERO ET AL.

(1989) India-US case-control study of age-related cataracts. Arch. Ophthalmol. 107: 670 – 676. 31. Jacques, P. & Chylack, L. (1991) Epidemiologic evidence of a role for the antioxidant vitamins and carotenoids in cataract prevention. Am. J. Clin. Nutr. 53: 352S–355S. 32. Simon, J. A. & Hudes, E. S. (1999) Serum ascorbic acid and other correlates of self reported cataract among older Americans. J. Clin. Epidemiol. 52: 1207–1211. 33. Chasan-Taber, L., Willett, W. C., Seddon, J. M., Stampfer, M. J., Rosner, B., Colditz, G. A. & Hankinson, S. E. (1999) A prospective study of vitamin supplement intake and cataract extraction among US women. Epidemiology 10: 679 – 684. 34. Mares-Perlman, J., Lyle, B., Klein, B., Fisher, A. I., Brady, W. E., Van den Langenberg, G. M., Trabulsi, J. N. & Palta, M. (2000) Vitamin supplement use and incident cataracts in a population-based study. Arch. Ophthalmol. 118: 1556 –1563. 35. Seddon, J. M., Christen, W. G., Manson, J. E., LaMotte, F. S., Glynn, R. J., Buring, J. E. & Hennekens, C. H. (1994) The use of vitamin supplements and the risk of cataract among US male physicians. Am. J. Public Health 84: 788 –792. 36. AREDS Report. (2001) A randomized, placebo-controlled, clinical trial of high-dose supplementation with vitamins C and E and ␤-carotene for agerelated cataract and vision loss. Arch. Ophthalmol. 119: 1439 –1452. 37. Langle, U., Wolf, A. & Cordier, A. (1997) Enhancement of SDZ ICT 322-induced cataracts and skin changes in rats fallowing vitamin E and selenium diet. Arch. Toxicol. 71: 283–289. 38. Mares-Perlman, J. A., Brady, W. E., Klein, B. E., Klein, R., Haus, G. J., Palta, M., Ritter, L. L. & Shoff, S. M. (1995) Diet and nuclear lens opacities. Am. J. Epidemiol. 141: 322–334. 39. Mares-Perlman, J. A., Brady, W. E., Klein, B. E., Klein, R., Palta, M., Bowen, P. & Stacewicz-Sapuntzakis, M. (1995) Serum carotenoids and tocopherols and severity of nuclear and cortical opacities. Invest. Ophthalmol. Vis. Sci. 36: 276 –288. 40. Olmedilla, B., Granado, F., Blanco, I. & Rojas-Hidalgo, E. (1994) Sea-

sonal and sex-related variations in six serum carotenoids, retinol, and ␣-tocopherol. Am. J. Clin. Nutr. 60: 106 –110. 41. Lyle, B. J., Mares-Perlman, J. A., Klein, B. E., Klein, R. & Greger, J. L. (1999) Antioxidant intake and risk of incident age-related nuclear cataracts in the Beaver Dam Eye Study. Am. J. Epidemiol. 149: 801– 809. 42. Brown, L., Rimm, E. B., Seddon, J. M., Giovannucci, E. L., Chasan-Taber, I., Spiegelman, D., Willett, W. C. & Hansinson, S. E. (1999) A prospective study of carotenoid intake and risk of cataract extraction in US men. Am. J. Clin. Nutr. 70: 517–524. 43. Vioque, J. & Quiles, J. (1995) Resultados preliminares de la Encuesta de Nutricio´ n Encuesta de Nutricion de la Comunidad Valenciana. In: Guı´as Alimentarias para la Poblacio´ n Espan˜ ola. (Serra, L. I., Aranceta, J. & Mataix, J., eds.), pp. 121–124. Documento de Consenso. SG Editore, Barcelona, Spain. 44. Block, G., Dresser, C., Hartman, A. & Carroll, M. (1985) Nutrient sources in the American diet: quantitative data from the NHANES II survey: macronutrients and fats. Am. J. Epidemiol. 122: 27– 40. 45. National Research Council. (1989) Recommended Dietary Allowances. 10th ed. National Academy Press, Washington, DC. 46. Garry, P., Vanderjagt, D. & Hunt, W. (1987) Ascorbic acid intakes and plasma levels in healthy elderly. Ann. N. Y. Acad. Sci. 498: 90 –99. 47. Blanchard, J. (1991) Depletion and repletion kinetics of vitamin C in humans. J. Nutr. 21: 170 –176. 48. Itoh, R., Yamada, K., Oka, J., Echizen, H. & Murakami, K. (1989) Sex as a factor in levels of serum ascorbic acid in a healthy elderly population. Int. J. Vitam. Nutr. Res. 59: 365–372. 49. Varela, G. (1981) Ingestas recomendadas de energia y nutrientes para la poblacion espan˜ ola. Instituto de Nutricion and CSIC, Madrid, Spain. 50. Leibowitz, H. M., Krueger, D. E., Maunder, L. R., Milton, R. C., Kini, M. M., Kahn, H. A., Nickerson, R. J., Pool, J., Colton, T. L., Ganley, J. P., Loewenstein, J. I. & Dawber, T. R. (1980) The Framingham Eye Study monograph. Survey Ophthalmol. 335– 610. 51. Levine, M., Rumsey, S., Daruwala, R., Park, J. & Wang, Y. (1999) Criteria and recommendations for vitamin C intake. J. Am. Med. Assoc. 281: 1415–1423.