Cho E, Seddon JM, Rosner B, Willett WC, Hankinson SE. Prospective Study of Intake of Fruits, Vegetables, Vitamins, and Carotenoidsand Risk of Age-Related Maculopathy. Arch Ophthalmol. 2004;122(6):883-892. doi:10.1001/archopht.122.6.883
Copyright 2004 American Medical Association. All Rights Reserved.Applicable FARS/DFARS Restrictions Apply to Government Use.2004
To examine the intake of antioxidant vitamins and carotenoids as wellas fruits and vegetables in relation to the development of age-related maculopathy(ARM).
We conducted a prospective follow-up study of women in the Nurses' HealthStudy and men in the Health Professionals Follow-up Study. We followed 77 562women and 40 866 men who were at least 50 years of age and had no diagnosisof ARM or cancer at baseline for up to 18 years for women and up to 12 yearsfor men. Fruit and vegetable intakes were assessed with a validated semiquantitativefood-frequency questionnaire up to 5 times for women and up to 3 times formen during follow-up.
A total of 464 (329 women and 135 men) incident cases of early ARM and316 (217 women and 99men) cases of neovascular ARM, all with visual loss of20/30 or worse due primarily to ARM, were diagnosed during follow-up. Fruitintake was inversely associated with the risk of neovascular ARM. Participantswho consumed 3 or more servings per day of fruits had a pooled multivariaterelative risk of 0.64 (95% confidence interval, 0.44-0.93; P value for trend = .004) compared with those who consumed less than1.5 servings per day. The results were similar in women and men. However,intakes of vegetables, antioxidant vitamins, or carotenoids were not stronglyrelated to either early or neovascular ARM.
These data suggest a protective role for fruit intake on the risk ofneovascular ARM.
Among individuals 65 years and older, late age-related maculopathy (ARM)is a leading cause of vision loss, which severely affects quality of life.1,2 Because of increasing longevity, theimpact of this disease continues to grow in the United States.
Because effective treatments do not exist to treat the late stages ofARM,3 prevention is important. Smoking is the most consistently identified modifiable risk factor,4 and antioxidant vitamin and mineral supplementationhas also been found to be beneficial.5 Severalother factors including dietary intake of antioxidants,6- 8 dietaryfats,9- 11 obesity,12 other cardiovascular disease risk factors,exposure to sunlight, and nonmodifiable factors including family history ofARM and ocular characteristics have been reported.13
Intake of antioxidant vitamins and carotenoids has been hypothesizedto reduce the risk of ARM by protecting the retina from oxidative damage.14 In a recent randomized trial, supplementation withhigh-dose vitamins C and E, beta carotene, and zinc delayed progression ofARM in participants at high risk of progression.5 Inprevious epidemiologic studies of dietary intake or blood levels of antioxidantvitamins or carotenoids, one case-control study6 ofa large population with neovascular, advanced macular degeneration showeda protective effect of dietary carotenoids, whereas other studies of primarilyearly maculopathy showed no clear associations with ARM risk.7,15- 24 However,few studies have examined intake of fruits and vegetables in relation to ARM.6,19,25
We therefore examined intake of antioxidant vitamins, carotenoids, fruits,and vegetables in relation to the incidence of ARM in 2 large prospectivestudies of women and men with up to 18 years of follow-up.
The Nurses' Health Study (NHS) enrolled 121 700 female registerednurses aged 30 to 55 years in 1976. The Health Professionals Follow-up Study(HPFS) included 51 529 male health professionals (dentists, veterinarians,pharmacists, optometrists, osteopathic physicians, and podiatrists) aged 40to 75 years in 1986. We have sent follow-up questionnaires to both cohortsbiennially to update information regarding diet and lifestyle and to ascertainnew diagnoses of major illnesses.
A semiquantitative food-frequency questionnaire (FFQ) with approximately60 food items was sent to members of the female cohort in 1980. An expandedFFQ with approximately 130 food items was administered to women in 1984, 1986,1990, and 1994 and to men in 1986, 1990, and 1994. The NHS 1980 FFQ had 5questions for fruits and 10 questions for vegetables. Subsequent FFQs in theNHS and the FFQs in the HPFS had at least 15 questions for fruits and at least22 questions for vegetables. Participants were asked how often, on average,they had consumed each type of food during the past year. Serving sizes (eg,1 banana or ½ cup broccoli) were specified for each food in the FFQ.The questionnaire had 9 possible responses, ranging from never or less thanonce per month to 6 or more times per day. Intake of total fruits and totalvegetables was calculated by multiplying the reported frequency by a givenserving size for each food item. Participants also reported their currentuse and dose of vitamin A, C, and E supplements and brands and types of multivitaminsbiennially. For current users of multivitamins or vitamin supplements at baseline,duration of use was also ascertained. A comprehensive database on multivitaminpreparations that provides the dose of the vitamins in each preparation hasbeen developed and updated biennially. Intakes of vitamins and carotenoidsfrom foods were calculated from US Department of Agriculture (USDA) sources.26 To calculate vitamin intake from food and supplementscombined, the contributions from multivitamins and other supplements wereadded to vitamin intakes from food only. Food composition data for specifictypes of carotenoids were based on the USDA–National Cancer Institutecarotenoid database developed by Chung-Ahuja et al27 andMangels et al.28 The carotenoid content oftomato-based food products was updated with values from the USDA.29
The reproducibility and validity of food intake was assessed in bothcohorts.30,31 The correlationcoefficients between diet records and the FFQ for fruits and vegetables averaged0.59 for women (range, 0.17-0.84) and 0.59 for men (range, 0.25-0.95) aftercorrection for attenuation due to random error in diet records.
The reproducibility and validity of intakes of vitamins and carotenoidswas assessed in both cohorts using diet records32,33 andplasma levels.34,35 Pearson correlationcoefficients between estimates from the FFQ and the average of 4 1-week dietrecords were 0.49 for total vitamin A (including contributions from food andsupplements) and 0.75 for total vitamin C in women.32 Similarly,the average correlation between estimates from the FFQ and the average of2 1-week diet records for vitamins and carotenoids was 0.74 (range, 0.48 to0.92) in men after correction for attenuation due to random error in dietrecords.33 Vitamin E intake estimated fromthe FFQ was positively correlated with its plasma concentration (r = 0.41 for women and 0.51 for men).34 ThePearson correlation coefficient between dietary carotenoid intake and plasmaconcentrations of carotenoids was 0.18 to 0.47 in nonsmoking women and 0.31to 0.48 in nonsmoking men.35
For this analysis, we began follow-up when diet was first measured,in 1980 for women and in 1986 for men. We excluded those who did not completea baseline FFQ, who had implausible energy intakes (<2510 or >14 644kJ/d for women and <3347 or >17 573 kJ/d for men), or who left morethan 70 items blank on the FFQ (6291 women and 1595 men). Participants whoreported a diagnosis of ARM or cancer (except nonmelanoma skin cancer) atbaseline were excluded (3625 women and 2012 men), and these exclusions wereupdated every 2 years. We also excluded participants who did not respond toany of the follow-up questionnaires asking about a diagnosis of ARM (1986-1998in the NHS and 1988-1998 in the HPFS; 2080 women and 987 men). Finally, weexcluded those who did not report having an eye examination during follow-up(7235 women and 4787 men) to minimize any influence of undiagnosed cases.Because ARM is rare in younger populations, we restricted our analysis toparticipants 50 years and older. Participants who were younger than 50 yearsat baseline were included in the cycle after they reached age 50. A totalof 30 078 women and 26 703 men were included in the analysis atbaseline. By 1996, 77 562 women and 40 866 men contributed to theanalyses. For carotenoid analysis for women, we began follow-up in 1984 becauseimportant foods contributing to carotenoid intake were not assessed in 1980.
In this report, we defined all stages of the disease as ARM. Accordingto one classification system, the late stages of ARM are also referred toas age-related macular degeneration.36
Cases were defined as incident ARM when their best-corrected visualloss was 20/30 or worse (ie, a person could recognize at 20 ft a symbol thatcould be recognized by a person with normal acuity at 30 or more feet) dueprimarily to ARM in at least one eye. We obtained data on the diagnosis ofARM beginning in 1986 for women (regarding diagnoses received from 1980-1986)and in 1988 for men. If people reported diagnoses of ARM, we requested permissionto review their medical records and contacted their ophthalmologists to eithercomplete a standardized questionnaire or to send us copies of ocular recordsto confirm the diagnosis. The questionnaire included the date of initial diagnosis,best-corrected visual acuity, signs of ARM (drusen, retinal pigment epithelialhypopigmentation or hyperpigmentation, geographic atrophy, retinal pigmentepithelial detachment, subretinal neovascular membrane, or disciform scar),and whether there was visual acuity loss due mainly to ARM.
We conducted analyses based on subgroups of ARM (early and neovascular)because these subtypes may have different etiologies and risk factors. Theearly form of ARM was defined as the presence of drusen or retinal pigmentepithelial changes.36 The neovascular formof ARM, usually associated with greater visual impairment, included retinalpigment epithelial detachment, choroidal neovascular membrane, or disciformscar. Because of limited numbers of cases with geographic atrophy, we werenot able to conduct an analysis for them. The person was used as the unitof analysis, and, if a participant had bilateral ARM with different degreesof progression, the more severe status was used.
Our case definition of ARM has been validated by 2 retinal specialistswho conducted a standardized review of fundus slides in a subset of cases(those ascertained from the 1990 follow-up in the NHS).4 Amongcases with photographs of sufficient quality to grade, 36 (86%) of 42 wereclassified as having definite ARM and 39 (93%) of 42 were classified as definiteor probable ARM by both readers. Regarding the classification of subtypesof ARM, there was 100% (23/23) concordance between the retinal specialistand the reporting ophthalmologist for early ARM and 86% (12/14) for neovascularARM.
A total of 3283 women and 1573 men reported a diagnosis of ARM duringfollow-up; 1452 women (44%) and 677 men (42%) were confirmed to have ARM bytheir ophthalmologist. For the remainder of participants reporting ARM, oneof the following was true: (1) they did not grant permission to contact theirophthalmologist (357 NHS [11%]; 176 HPFS [11%]), (2) they indicated the initialreport was in error (651 NHS [20%]; 368 HPFS [23%]), (3) they did not havethe diagnosis confirmed by their ophthalmologist (709 NHS [22%]; 275 HPFS[17%]), or (4) we were not able to contact either the participant or theirophthalmologist (114 NHS [3%]; 100 HPFS [6%]). For those whose diagnosis wasnot confirmed by their ophthalmologist, the doctor frequently indicated othermaculopathies (eg, macular hole) or other eye diseases (eg, diabetic retinopathy).We then excluded women and men who did not have visual loss of 20/30 or worse(590 NHS; 230 HPFS) or whose visual loss was not attributable to ARM (191NHS; 66 HPFS); 670 women and 357 men met our case definition. Among them,after excluding subjects without plausible dietary information, with a previousdiagnosis of cancer, or with a diagnosis given before completing the baselineFFQ or after the end of follow-up, 546 women and 234 men were included inthe analysis.
Participants were divided into absolute intake categories or quintilesaccording to their fruit, vegetable, or nutrient intake. In the primary analysis,intake data were updated according to the cumulative average of intake duringthe follow-up period examined. For example, in women, 1980 intake was usedfor 1980-1984 follow-up and the average of 1980 and 1984 intake was used for1984-1986 follow-up and so on. Baseline and most recent intake were each examinedalone in secondary analyses. Study participants contributed person-time ineach 2-year interval from the time the baseline FFQ was returned or from thetime the first questionnaire was returned after they reached 50 years of ageuntil a diagnosis of ARM or cancer, death, time of last questionnaire return,or end of the follow-up period (June 1, 1998, for women and January 1, 1998,for men), whichever came first.
We used Cox proportional hazards regression models to account for potentialeffects of other risk factors for ARM.37 Tocontrol as finely as possible for confounding by age, calendar time, and anypossible 2-way interactions between these 2 time scales, we stratified theanalysis jointly by age in months at the start of follow-up and calendar yearof the current questionnaire cycle. Multivariate models also adjusted forsmoking, body mass index, energy intake, alcohol intake, fish intake, physicalactivity (metabolic equivalents per week in quintiles in men, hours of vigorousactivity in quintiles in women), history of hypertension and high blood cholesterollevels, postmenopausal hormone use (women), and occupation (men). To adjustfor smoking, pack-years of smoking (the number of years smoked multipliedby the average number of packs of cigarettes per day) was used, since thisbest reflects the cumulative effect of smoking and is more strongly associatedwith ARM than current smoking status.4 Amongthese covariates, pack-years of smoking, body mass index, and postmenopausalhormone use were updated in every 2-year period. Dietary covariates were updatedusing cumulative averaged intake. We used SAS PROC PHREG38 (SASInstitute Inc, Cary, NC) for all analysis, and the Anderson-Gill data structure39 was used to handle time-varying covariates efficiently,with a new data record created for every questionnaire cycle at which a participantwas at risk and covariates set to their values at the time the questionnairewas returned. For all relative risks (RRs), 95% confidence intervals (CIs)were calculated. Tests for trend across categories of intake were conductedby using the median within each category as a continuous variable.40 All P values were 2-sided.
We conducted separate analyses for each cohort and pooled the 2 studiesto achieve maximum statistical power. Tests for heterogeneity between the2 studies were conducted, and meta-analytic methods using a random-effectsmodel were used to pool the RRs from the cohorts.41
To confirm the results for neovascular ARM from the primary analyses,we conducted additional restricted analyses among nonsmokers and among nonusersof multivitamins or vitamin A, C, and E supplements as well as among thosewho, at baseline, reported they had not changed their fruit intake in thepast 10 years.
To test whether the estimates for fruit intake for early vs neovascularARM were statistically different, we conducted polychotomous logistic regression42 in each cohort using fruit intake as a continuousvariable, early and neovascular ARM as 2 different outcome variables, anda likelihood ratio test with 2 df.
We documented 464 cases of early ARM (329 women and 135 men) and 316cases of neovascular ARM (217 women and 99 men) with visual loss of 20/30or worse due primarily to ARM during up to 18 years of follow-up in women(923 926 person-years) and up to 12 years of follow-up in men (345 366person-years).
Table 1 presents the distributionof potential risk factors for ARM by categories of fruit and vegetable intakein 1990. Participants with higher fruit or vegetable intake were less likelyto smoke and more likely to be physically active and to consume fish.
Total fruit intake was inversely related to neovascular ARM risk (Table 2). The pooled multivariate RR forparticipants who consumed 3 or more servings per day of fruits was 0.64 (95%CI, 0.44-0.93; P for trend = .004) compared withthose who consumed less than 1.5 servings per day. The resultswere similar for women and men (P for heterogeneityfor top vs bottom categories = .70). There was a nonsignificant inverse associationbetween fruit intake and early ARM. However, the difference in results forearly vs neovascular ARM was not statistically significant for fruit intake(P = .45). Vegetable intake was not related to earlyor neovascular ARM risk in either cohort. The results were similarafter adjusting for total fat intake.
To confirm the inverse association between fruit intake and neovascularARM, we repeated analyses examining baseline intake as well as most recentintake. The pooled multivariate RR for the highest category of intake was0.62 (95% CI, 0.40-0.96) for baseline intake and 0.71 (95% CI, 0.40-1.25)for most recent intake. We also examined whether the association remainedamong those who did not change their fruit intake in the past 10 years atbaseline (n = 214 neovascular ARM cases); the pooled multivariate RR for thehighest category of intake was 0.80 (95% CI, 0.50-1.27). In women, more comprehensiveinformation on fruit intake was collected from 1984 onward. The inverse associationbetween fruit intake and neovascular ARM was similar, although slightly weaker,when the analysis was started from 1984 in women; the multivariate RR forthe highest category of fruit intake was 0.83 (95% CI, 0.50-1.38).
Since fruit intake was related to smoking status (Table 1), we conducted analyses among never-smokers only (n = 96with neovascular ARM) to avoid any confounding by smoking habits. The pooledmultivariate RR for the highest category of fruit intake was not statisticallysignificant, probably owing to limited power (RR, 0.87; 95% CI, 0.42-1.82).We also tried adjusting for continuous pack-years of smoking instead of acategorical variable to avoid residual confounding by a categorical smokingvariable; however, the results were similar to those using a categorical smokingvariable (data not shown). We also assessed the association between fruitintake and neovascular ARM among those who did not take multivitamins andvitamin A, C, and E supplements (n = 154 with neovascular ARM) to minimizeany confounding by these supplements or other related healthy lifestyle factors.The pooled multivariate RR for the highest category of fruit intake was 0.79(95% CI, 0.46-1.33).
To examine whether fruit overall or specific fruit were related to neovascularARM risk, we examined 5 fruit questions that were asked of women at baseline(Table 3). Although all of theseitems had a suggestive inverse association, only higher intakesof oranges and bananas achieved statistical significance; the pooledmultivariate RRs for the highest intake category (≥3 servings/wk) comparedwith those in the lowest (<2 servings/mo) were 0.61 (95% CI, 0.44-0.86; P for trend = .01) for oranges and 0.63 (95% CI, 0.44-0.90; P for trend = .18) for bananas. Because consumption ofthese foods may be related, we adjusted for both items simultaneously; theresults did not change materially (data not shown). Banana intake was alsoinversely related to early ARM; the pooled multivariate RR for participantswho consumed 3 or more servings per week of banana was 0.67 (95% CI, 0.49-0.90; P for trend = .05) compared with those who consumed lessthan 2 servings per month.
None of the vegetable items appeared to be stronglyrelated to either early or neovascular ARM risks, except that carrot intakehad a weak, nonsignificant inverse association with the neovascular form (Table 3). For example, increasing intakeof spinach or other greens had pooled multivariate RRs of 1.00, 1.14, 1.10,and 1.05 (95% CI, 0.66-1.67) for neovascular ARM (Table 3).
To examine whether any specific nutrients were responsible for the inverseassociation between ARM risk and fruit intake, we examined intakes of antioxidantvitamins and carotenoids from foods, mainly fruits and vegetables (Table 4). None of the antioxidant vitaminsand carotenoids was strongly related to either early or neovascular ARM risk,although many of them, including total carotenoids, had a suggestive inverseassociation with neovascular ARM risk. Intake of β-cryptoxanthin wasrelated to a lower risk of neovascular ARM (P fortrend = .03), although the pooled multivariate RR for the highest quintileof intake was not statistically significant. Because we might have missedsome associations in quintile analyses, we also examined deciles of the nutrientintake; again, no strong association was found (data not shown). The overallresults were similar when baseline intake was examined. However, when we examinedthe most recent intake (prior to diagnosis for ARM cases), alpha caroteneappeared to be significantly inversely related to neovascular ARM risk; thepooled multivariate RR for the highest quintile of intake was 0.63 (95% CI,0.43-0.93; P for trend = .04) for alpha carotene.Other carotenoids remained similar to the results using cumulative updateintake (eg, pooled multivariate RR for top vs bottom quintile of baselinelutein/zeaxanthin intake, 1.02; 95% CI, 0.68-1.54).
We also examined use of multivitamins and vitamin C and E supplements(Table 5); because too few participantswere taking vitamin A and beta carotene supplements, we were not able to examinethem separately. None of the supplements was related to either early or neovascularARM risk. These vitamin supplements were not related to either early or neovascularARM when current users were grouped into 2 dose categories (data not shown).Participants who had taken both vitamin C and E supplements for more than4 years also did not benefit compared with never-users of both supplements;the pooled multivariate RRs were 1.16 (95% CI, 0.84-1.60) for early ARM and0.87 (95% CI, 0.56-1.34) for neovascular ARM.
In this prospective study of women and men, intakes of antioxidant vitaminsor carotenoids either from food only or from food and supplements were notstrongly related to ARM risk. Similarly, no substantial associations wereobserved between vegetable intake and ARM. However, fruit intake was inverselyrelated to ARM, particularly neovascular ARM, the form of this disease thatfrequently involves severe visual loss.
One of the etiologic hypotheses for development of ARM is related tooxidative stress, such that a deficit of important antioxidant vitamins orcarotenoids may adversely affect the retina.43 Dietaryas well as serum carotenoids were associated with a reduction in risk of neovascular,advanced ARM in a large multicenter case-control study.6,8 However,other epidemiologic studies found no clear inverse association between intakeof antioxidant vitamins and carotenoids and ARM risk, possibly owing to smallsample size and the small number of neovascular ARM cases.7,15- 24 Inthe Age-Related Eye Disease Study,5 a largemulticenter randomized trial of 3640 participants with 6.3 years of follow-up,supplementation with high-dose vitamins C and E, beta carotene, and zinc reducedprogression to advanced ARM by 25% across 5 years among those with signs ofintermediate ARM or advanced disease in one eye. The combination of vitaminsC and E and beta carotene was also associated with a reduction in risk inthis group of patients, but this finding did not achieve statistical significance(odds ratio, 0.76; 99% CI, 0.55-1.05).5 Therelationships between dietary sources of the nutrients and ARM in this studyhave not yet been published.
Fruits and vegetables are rich sources of antioxidant nutrients, butfew studies have examined intake in relation to ARM risk. A cross-sectionalstudy suggested an inverse association between intake of fruits and vegetablesrich in vitamin A and ARM (n = 178).25 Seddonet al6 examined consumption of carotenoid-richvegetables and found that intake of green leafy vegetables such as spinachor collard greens was inversely related to neovascular ARM risk. In a smallcohort study of only 103 early ARM cases, a modest inverse association wasobserved between fruit and vegetable intake and large drusen.16
We found that fruit intake was related to a reduced risk of neovascularARM but not early ARM. Early and neovascular ARM have been hypothesized tohave different etiologies and risk factors, and abnormal vascular circulationmay be related to development of neovascular ARM.44,45 NeovascularARM (or late ARM, which includes neovascular ARM and geographic atrophy) hasbeen related to cardiovascular disease risk factors45 suchas smoking,4 obesity,12 dietaryfat,9,11 elevated plasma fibrinogenlevels,46 atherosclerosis,47 andhypertension.13,48,49 Fruit(but not vegetable) intake has been related to a reduced risk of ischemicstroke in these cohorts.50 Both fruit and vegetableintake have been associated with a lower risk of cardiovascular disease.51- 53
Among types of fruit, the suggestive inverse associations were significantonly for oranges and bananas. Although the results were similar in women andmen, this may be a chance finding because many foods were considered. In addition,differences in measurement error in reporting fruit intake may also contributeto the results. However, since none of the antioxidant vitamins and carotenoidscontributed substantially to the apparent inverse association, other factorsin fruit may also contribute to the reduced risk. Other constituents of fruitswith potential health benefit include flavonoids, isothiocyanates, phenols,fiber, folate, and potassium.54
We observed no association with intake of lutein/zeaxanthin or foodsrich in lutein/zeaxanthin such as green leafy vegetables, which is inconsistentwith many recent experimental studies that have suggested the importance ofthese carotenoids as a macular pigment and as a possible preventive factorfor ARM.55,56 However, a significantinverse association was found mainly in a large case-control study of neovascularARM.6,8 Other epidemiologic studiesof lutein/zeaxanthin have been inconclusive, possibly owing to the smallerstudy sizes as well as the mixture of disease stages.16,21,23,57,58 However,some of these latter studies did find associations with other specific carotenoids.16,23,58 One possible explanationfor the null finding in our cohort is that lutein/zeaxanthin intake mightbe already high enough; for example, the median lutein/zeaxanthin intake inthe first quintile in both cohorts was higher than the median value of thesecond quintile in the Eye Disease Case-Control Study.6 However,the RRs were quite flat across quintiles, and analyses using deciles of intakealso did not show any substantial association in our study.
This is the first large-scale prospective study examining dietary intakecomprehensively in relation to ARM risk. We had repeated measures of dietand were able to examine dietary intake in different ways (baseline, cumulativeupdated, and most recent). Cumulative updated intake can minimize measurementerror due to a onetime dietary assessment and thus best reflect long-termdietary intake, which may be most relevant to chronic diseases like ARM witha long duration of development.59 However,because the period during which diet might modify the development of ARM isunclear, it is also meaningful to examine remote (baseline) and most recentintakes. Indeed, for alpha carotene intake, we observed a significant inverseassociation only with recent intake.
Study limitations also need to be considered. Becausehigher fruit intake may be related to a more healthy lifestyle, the resultsmight be confounded by other risk factors related to a healthy lifestyle,such as nonsmoking. Although we adjusted for smoking, some residual confoundingmight remain. A healthy lifestyle may also be associated with more medicalscreening, including eye examination, and thus increase the chanceof being diagnosed with ARM. To minimize this possibility, we limited ouranalysis to participants who had at least 1 eye examination during follow-upand included cases with a visual acuity loss of at least 20/30 due to ARM.In addition, the fact that an inverse association was found between neovascularARM (but not early ARM) and fruits (but not vegetables) in both cohorts suggeststhat results were minimally influenced by confounding by other aspects ofhealthy lifestyles or biased by differential disease diagnosis. Even in thislarge study, the number of cases was somewhat limited, especially for neovascularARM.
In conclusion, in these large prospective cohorts of women and men,we found that higher fruit intake was related to a reduced risk of neovascularARM. However, none of the vitamins or carotenoids examined was clearly relatedto disease. Further studies are needed to confirm our findings and to identifythe relevant compound(s) in fruits.
Corresponding author and reprints: Eunyoung Cho, ScD, Channing Laboratory,181Longwood Ave, Boston, MA 02115 (e-mail: email@example.com).
Submitted for publication August 11, 2003; final revision received January6, 2004; accepted January 6, 2004.
This study was supported by research grants CA87969, CA55075, EY09611,and HL35464 from the National Institutes of Health, Bethesda, Md.
We thank Maureen Ireland, BA, and Stacey DeCaro, BA, for data compilationand Jae Hee Kang, ScD, and Mary Louie, PhD, for computer support.