Associations Between Intermediate Age-Related Macular Degeneration and Lutein and Zeaxanthin in the Carotenoids in Age-Related Eye Disease Study (CAREDS): Title and subTitle BreakAncillary Study of the Women's Health Initiative FREE

Age-related macular degeneration (AMD) is a degenerative disease of the macula, the region of the retina that is responsible for central vision. It is the leading cause of irreversible vision loss as people age in the United States.¹ In its later stages, AMD affects about 7% of people between the ages of 75 and 84 years.² Early and intermediate stages of AMD, which increase the risk of developing advanced AMD,³ are even more common, affecting about one quarter of people older than 65 years.² Because there is no cure for this condition and limited treatment is available to slow its progression, and there are no established means of prevention, it is important to identify modifiable risk factors that may reduce disease occurrence or prevent progression to the late stages of AMD. The pathogenesis of AMD is likely to involve a complex interaction of cellular and vascular factors, which may be promoted by light damage,⁴ oxidative stress,⁵ and inflammation.⁶

Several epidemiologic studies have reported associations between various dietary components and AMD, including the dietary carotenoids lutein and zeaxanthin (previously reviewed^{7 - 8} ). Lutein and zeaxanthin may be obtained only through the diet in foods such as leafy green vegetables, corn, squash, broccoli, peas, and egg yolks^{9 - 10} or in supplements. They are the only carotenoids that concentrate in the macula.^{11 - 12} There is evidence of 3 mechanisms by which lutein and zeaxanthin might protect against AMD: by absorbing blue light,^{13 - 16} by quenching free radicals (reviewed by Landrum and Bone¹⁶ ), or by increasing membrane stability.¹⁷

The Carotenoids in Age-Related Eye Disease Study (CAREDS), an ancillary study of the Women's Health Initiative (WHI), was designed, in part, to evaluate the relationship between the dietary carotenoids lutein and zeaxanthin and the prevalence of AMD. Women who had high or low intake of lutein plus zeaxanthin at WHI enrollment were recruited into CAREDS to maximize extremes in intake of these carotenoids in our study sample. The present study examines whether women with high intake of lutein plus zeaxanthin have a lower prevalence of intermediate AMD and specific macular lesions than women with low intake of these carotenoids. We also report on whether these relationships are altered after accounting for other lifestyle, health, and dietary attributes that may be correlated with diets high in lutein plus zeaxanthin and AMD.

The WHI Observational Study was a prospective cohort study of the most common causes of mortality and morbidity among 93 676 postmenopausal women 50 to 79 years of age at enrollment, at 40 sites around the United States. The original cohort was recruited to the WHI at each of these sites through regional mass mailings and mass-media strategies, from among women ineligible for or uninterested in participation in the WHI Clinical Trials. Participants were followed on average 7 years after enrollment. Women were excluded if they had medical conditions that predicted survival of less than 3 years, alcoholism, drug dependency, or mental illness.^{18 - 19}

The CAREDS population consists of women who were enrolled in the observational study of the WHI (WHI-OS) at 3 of 40 sites—the University of Wisconsin (Madison), the University of Iowa (Iowa City), and the Kaiser Center for Health Research (Portland, Ore)—who had intakes of lutein plus zeaxanthin that were above the 78th or below the 28th percentiles, as assessed at baseline enrollment into the WHI from 1994 to 1998 (Figure). Of the 3143 women who fulfilled these criteria, 96 died or were lost to follow-up between selection in 2000 and enrollment in CAREDS from 2001 to 2004. Those remaining were sent letters inviting them to participate in CAREDS. A total of 1042 women declined participation and 2005 (63.8%) were enrolled. Of those enrolled, 1894 participated in study visits. Gradable fundus photographs were available for 1853 participants; an additional 4 participants were included who did not have AMD photographs but had a physician's confirmation of AMD. One participant was excluded because her lutein data were determined to be unreliable. Sixty-nine participants were further excluded because of missing covariate data. Thus, 1787 women composed the analysis data set for this investigation. All procedures conformed to the Declaration of Helsinki and were approved by the institutional review board at each university.

To evaluate potential bias due to lack of participation in CAREDS, we compared characteristics of women in the recruitment data set who did not enroll in CAREDS with those of women who did. Women with high lutein intake were more likely to join CAREDS than those with low intake (69% vs 60%, P<.001). However, there was no difference in the prevalence of self-reported AMD between the participants and nonparticipants (5% vs 4%; P = .21).

We further compared women in the recruitment data set who were included in the final analysis data set with those who were excluded. Women included in the final analysis data set were younger, had more formal education, and had generally healthier lifestyles and fewer comorbid conditions (R.B.W., unpublished data, 2005).

Diet was assessed at WHI baseline (1994-1998) and 15 years before CAREDS baseline (1986-1989) by means of a previously validated, semiquantitative food frequency questionnaire (FFQ).²⁰ The 15-year past FFQ was administered at CAREDS baseline (2001-2004); retrospective FFQs have been found to be reliable indicators of past diet.^{21 - 22} Nutrient and food group estimates in WHI and CAREDS were computed from responses to FFQs at the Fred Hutchinson Cancer Research Center. Data from the 2004 participants with reliable FFQ data (≥600 kcal and ≤5000 kcal) at WHI baseline and the 1995 participants with CAREDS 15-year past FFQ data collected retrospectively were used to generate quintiles of nutrient intake, and tertiles of food and food group intake, for each time point.

Because the FFQ did not clearly distinguish between consumption of dark vs light greens in salads, we explored the possibility of nonrandom misclassification of participants in the high or low lutein intake recruitment groups. For these analyses, estimates of lutein plus zeaxanthin at both times were modified by responses to a query about the proportion of time that dark vs light greens were consumed at the time of the CAREDS baseline examination and in an earlier period (1986-1989). This modification did not significantly change the original classification of individuals in the high (>78th percentile) and low (<28th percentile) lutein intake groups (data not shown).

Serum samples were obtained from participants at WHI baseline examinations (1994-1998) after a 10-hour or longer fast and were stored at −80°C. Serum levels of lutein and zeaxanthin were determined at Tufts University, Boston, Mass (2004-2005) by a reverse-phase high-performance liquid chromatography analysis.²³ Levels of lutein and zeaxanthin represent the sum of all-trans isomers. Blind duplicates were analyzed in each batch of serum analyses (in a total of 57 subjects). Mean coefficients of variation were 7.0% for all-trans lutein and 9.6% for all-trans zeaxanthin. Total triglyceride levels were determined with an automated chemistry analyzer based on an enzymatic method with a glycerol blank,²⁴ and total cholesterol levels were determined enzymatically using the cholesterol oxidase method.²⁵

Stereoscopic color fundus photographs were taken at CAREDS baseline (2001-2004) examinations and graded by the University of Wisconsin Fundus Photography Reading Center. Methods for taking and grading the photographs were based on those used in the Age-Related Eye Disease Study and have been previously described.²⁶ The primary outcome was the presence of intermediate AMD in at least 1 eye, defined as the presence of either large drusen (≥1 large druse [≥125 μm] or extensive intermediate drusen [area, ≥360 μm when soft indistinct drusen are present, or ≥650 μm when soft indistinct drusen are absent]) or pigmentary abnormalities (an increase or decrease in pigmentation accompanied by at least 1 druse [≥63 μm]). The nondiseased referent group included women without any intermediate or advanced AMD in either eye. Because of the small number of cases, advanced AMD was an exploratory outcome defined as geographic atrophy, neovascularization, or exudation in the center subfield. In addition, women who provided questionnaire data but no photographs (n = 11) were eligible to be classified as having the AMD end points if we received a written physician confirmation of the outcome (n = 4).

Age, family history of AMD (immediate family member aged 55 years or older when diagnosed), ambient sunlight exposure (since age 18 years and in last 20 years based on outdoor activities during routine and vacation periods, living location, and use of protective gear [hats or sunglasses]), and use of nutritional supplements were queried in questionnaires submitted at the CAREDS visit. Iris color was measured from photographs taken at the CAREDS eye examination. Additional demographic, lifestyle, and health history data were available from questionnaires completed at WHI study entry (education, smoking, physical activity, height, weight, hormone replacement therapy, alcohol use, and history of diabetes, hypertension, and cardiovascular disease).

We performed unpaired, 2-tailed t tests, analysis of covariance, and χ² tests to assess the statistical significance of potential covariates in the high vs low lutein plus zeaxanthin intake (recruitment) groups and by the presence or absence of AMD. Odds ratios (ORs) and 95% confidence intervals (CIs) for specific AMD end points (large drusen, pigmentary abnormalities, any intermediate AMD, and advanced AMD) were calculated comparing the high lutein plus zeaxanthin intake group with the low intake group by means of logistic regression (PROC LOGISTIC in SAS version 8.2; SAS Institute Inc, Cary, NC). Variables that were statistically significantly related (P≤.10) to both the AMD and lutein plus zeaxanthin intake group in CAREDS, and/or that were previously suspected to be biologically plausible confounders, were tested by adding them singly to age-adjusted models. These variables included age (continuous), pack-years smoked (never, 0-7, or >7 y), history of diabetes (yes or no), family history of AMD (yes or no), iris color (blue vs nonblue), alcohol use (none, 0-<4, or ≥4 g/d), pulse pressure (millimeters of mercury), history of cardiovascular disease (yes or no), use of hormone replacement therapy (never, past, or current), physical activity (metabolic equivalents per week), body mass index (calculated as weight in kilograms divided by the square of height in meters), sunlight exposure (Maryland sun-years²⁷ ), intakes at WHI baseline of total energy (kilocalories), total fat (percentage of kilocalories), polyunsaturated fat (percentage of kilocalories), vitamin C (milligrams per day from food), vitamin E (milligrams per day from food), γ-tocopherol (milligrams per day), and zinc (milligrams per day); and use of the following supplements at CAREDS baseline: multivitamins and minerals, high-dose antioxidants (daily intake of at least 2 of the following 3 antioxidant supplements: ≥120 mg of ascorbic acid [vitamin C], ≥60 IU [40 mg] of vitamin E, and ≥10 000 μg of β-carotene), or high-dose zinc (≥15 mg/d). If the addition of the covariate changed the OR by 10% or more, the covariate was selected for addition to a larger model. For each AMD end point, the covariate that changed the OR the most was added to the model first. Other covariates were added one at a time to the model. The ORs comparing the high vs low lutein plus zeaxanthin intake recruitment group were reassessed after the addition of each potential covariate. Only covariates that still changed the OR by 10% or more in the fuller model were retained in the final model. Dietary variables that met these criteria but were strongly correlated with dietary lutein plus zeaxanthin (r≥0.50) were not included in the final model to limit problems of collinearity.

We tested for interactions to determine whether the associations between lutein plus zeaxanthin intake and AMD differed according to levels of other risk factors (age, sunlight exposure). An α level of .10 was used to determine significance. When a significant interaction was found, analyses were conducted both in the overall sample and by strata. Where applicable, quintiles of diet or serum values and tertiles of food or food group servings were regenerated within strata.

Preliminary data suggested fluctuations in the level of lutein plus zeaxanthin in the diets of women in our sample before their enrollment in the WHI. Thus, we further examined the ORs for each AMD end point after excluding women whose intake of lutein changed more than 1 quintile category between the CAREDS 15-year past FFQ and the WHI baseline FFQ. Because diet change related to the presence of other comorbidities may have occurred before diet measurement but influenced AMD risk, we further excluded women with risk factors for dietary change: history of cardiovascular disease, history of diabetes, self-reported AMD, and/or a history of hypertension.

In additional exploratory analyses, we examined the odds of intermediate AMD in women with high and low intakes of specific foods rich in lutein plus zeaxanthin, including spinach, dark green leafy vegetables, and fruits and vegetables in general. Tertiles of food intake were created by means of data from the 2004 CAREDS participants with reliable WHI baseline FFQ data, both overall and within age strata. Because of the low intake of cooked greens in our sample (spinach, mustard greens, turnip greens, and collards), comparisons were made in those eating less than 1 vs 1 or more serving per month. To guide future research, we also created a nutrient index to explore the possibility that intake of lutein plus zeaxanthin may influence the odds of AMD differently depending on intake of other nutrients that may protect against AMD.

Serum lutein, zeaxanthin, and lutein plus zeaxanthin levels were log-transformed and divided into quintile categories. The ORs and 95% CIs were calculated as described earlier in this section, using the lowest quintile category as the referent group, both in the overall sample and by age strata.

The median level of lutein plus zeaxanthin in the diet was approximately 3 times higher in women in the high lutein intake group than in those in the low intake group (Table 1). Women in the high lutein intake recruitment group were more likely to be older, nonwhite, college educated, in a higher income bracket, and more physically active and to have lower body mass index, waist-hip ratios, waist circumferences, and serum triglyceride levels than those in the low lutein intake recruitment group. They also were more likely to rate their general health status as excellent or very good. Women with lower levels of dietary lutein plus zeaxanthin intake were more likely to live in Iowa and have blue eyes. Total energy intake was higher in the high lutein intake recruitment group, although the percentages of dietary fats in the diet were less than in the low lutein intake recruitment group. Intakes of dietary fiber, fruits, vegetables, many micronutrients and carotenoids, and alcohol were also higher in the high than the low lutein intake group, as were serum levels of lutein and zeaxanthin and α-tocopherol. Despite higher intake of γ-tocopherol, women in the high lutein intake recruitment group had lower serum levels than those in the low lutein recruitment group.

The prevalence of overall intermediate AMD was not statistically different between the high and low lutein intake recruitment groups after adjusting for age (Table 1). Age-adjusted ORs for overall intermediate AMD or specific types were not associated with intake of lutein plus zeaxanthin (Table 2). Further adjustment for other AMD risk factors did not significantly influence the estimated ORs, nor did the exclusion of women reporting a previous diagnosis of AMD (data not shown). Because competing risk factors and selective mortality bias can impede the ability to accurately estimate ORs in older individuals, we examined the ORs for AMD by age groups (P for interaction for intermediate AMD = .09). The estimated ORs were in opposite directions in women younger than 75 years vs those 75 years and older (Table 2). Addition of total energy intake did not alter the significance of the models (data not shown).

No further insight was gained when the ORs for AMD were examined across quintiles of dietary lutein plus zeaxanthin levels. Estimates remained statistically nonsignificant (data not shown). Results were similar when intake of lutein plus zeaxanthin was examined as a nutrient density (micrograms per 1000 kcal per day), as well as when dietary lutein plus zeaxanthin values modified for use of dark vs light salad greens were used.

We examined the association between lutein plus zeaxanthin levels and AMD in women whose intakes had not changed more than 1 quintile category between the CAREDS 15-year past FFQ and the WHI baseline FFQ (Table 2). Estimates remained similar to those obtained in the overall sample. However, in the younger age group (<75 years), stronger inverse relationships were seen for all types of AMD. Further exclusion of women with a history of cardiovascular disease, diabetes, hypertension, and/or previously diagnosed AMD, which may have contributed to altered dietary intake of lutein- and zeaxanthin-containing foods in the years before diet estimates were available, resulted in substantially lower ORs in the overall sample and in women younger than 75 years.

Because lutein and zeaxanthin might protect, in part, by absorbing the energy of sunlight, we examined the possibility that the association between dietary lutein plus zeaxanthin and intermediate AMD differed in women whose average annual sunlight exposure during the past 20 years was above the median vs below (P for interaction = .07). The associations were in opposite directions in those above (OR, 1.13; 95% CI, 0.81-1.59) and below (OR, 0.79; 95% CI, 0.56-1.14) the median. However, in the youngest age group, the association between lutein and intermediate AMD was similar regardless of sunlight exposure (ORs and 95% CIs in those above and below the median for sun exposure, respectively: 0.80 [0.52-1.24] and 0.82 [0.53-1.29]), and did not differ from the age-adjusted estimate for women younger than 75 years. The OR in the overall sample was significantly protective after exclusion of women with recent diet instability or who were at risk for diet change, but only in those with relatively low sun exposure (OR, 0.54; 95% CI, 0.30-0.99). Results in women younger than 75 years were similar to those seen in Table 2, but the point estimate was lower in women below the median for sun exposure (OR, 0.52; 95% CI, 0.25-1.08) vs above the median (OR, 0.61; 95% CI, 0.29-1.28).

We investigated the relationship between specific foods and intermediate AMD in the youngest women at risk for AMD in our sample (<75 years) who also reported stable intake of lutein plus zeaxanthin and did not report any comorbidities that increased their risk of previous diet change (Table 3). The point estimate comparing the prevalence of AMD in the highest vs lowest tertile category of fruit and vegetable intake was similar to that observed in the age-adjusted model for dietary lutein plus zeaxanthin (Table 2), but was not statistically significant. However, the prevalence odds for intermediate AMD were substantially lower and statistically significant in women in the highest 2 tertile categories of vegetable intake compared with those in the lowest category. A similarly strong, inverse relationship was observed in women in the highest tertile category of green vegetable intake. Associations between intermediate AMD and intake of cooked greens, broccoli, and corn were protective but not statistically significant.

Since dietary intakes may not necessarily reflect the amount absorbed or used by relevant tissues and may be subject to reporting errors, we also examined the association between serum lutein plus zeaxanthin levels and intermediate AMD (Table 4). No statistically significant estimates or trends were observed in either the age- or multivariable-adjusted models. Stratifying by age group did not significantly change the results (P for interaction = .89). Results were similar for serum lutein and serum zeaxanthin levels, evaluated separately (data not shown). Similar to the observations with dietary intake, a stronger protective association was found in the subsample of the youngest women at risk for AMD (<75 years), who also reported stable intake of lutein plus zeaxanthin and did not report any comorbidities that increased their risk of previous diet change, but only in the highest quintile category of serum levels, and it was not statistically significant (Table 4).

To further explore the possible effect of other nutrients on the relationship between lutein plus zeaxanthin intake and AMD, we created a crude nutrient index to classify persons with generally healthy vs generally unhealthy intakes of other nutrients that have been associated with AMD. Analyses were conducted in women younger than 75 years (Table 5). Although results were not statistically significant, the ORs were all in a protective direction. The lowest OR was observed in women with low intake of lutein plus zeaxanthin but high intakes of vitamin E and zinc and low intake of polyunsaturated fats compared with women with poor intakes of all of the nutrients of interest (ie, low intakes of lutein, vitamin E, and zinc and high intake of polyunsaturated fats). Very few individuals reported low lutein intake with high intakes of vitamin E and zinc and low intake of polyunsaturated fats (n = 26), and even fewer reported high lutein intake and poor intakes of the other 3 nutrients (n = 19), so these exploratory analyses could not be performed within the subsample of participants younger than 75 years with stable diets or in those 75 years or older.

We conducted exploratory analyses of the associations between lutein plus zeaxanthin levels, in the diet and serum, and advanced AMD. No statistically significant associations were observed between dietary lutein plus zeaxanthin levels and advanced AMD in the overall sample, although the ORs were in a protective direction in younger women, particularly after exclusion of women who had changed their diets between the CAREDS 15-year past FFQ and the WHI baseline FFQ (Table 2). We observed direct associations between serum lutein level and advanced AMD, with statistically significant trends, although the CIs around the estimates were considerably wide (data not shown). We were unable to conduct further exploratory analyses because of the limited number of advanced AMD cases (n = 34) in CAREDS.

We did not observe the hypothesized association between dietary lutein and zeaxanthin intake and prevalence of intermediate AMD in the full study sample of women aged 50 to 79 years. However, secondary analyses disclosed a statistically significant, protective association in women younger than 75 years with stable intake of lutein plus zeaxanthin, without a history of cardiovascular disease, diabetes, hypertension, and/or previously diagnosed AMD. Similar protective associations were observed for large drusen. While not statistically significant, associations in this subsample were in the protective direction for the more advanced lesions of pigmentary abnormalities, as well as for our exploratory outcome, advanced AMD. We observed the strongest inverse associations between intermediate AMD and high intake of vegetables in general, as well as of green vegetables.

The body of evidence from experimental and observational studies supports a protective relationship between lutein and zeaxanthin levels and AMD. Primates fed diets deficient in these xanthophyll carotenoids^{28 - 29} show a loss of retinal pigment epithelial cells and increased photoreceptor cell death.³⁰ People with AMD have lower concentrations of these carotenoids in their maculae than those without AMD.^{31 - 32} Similarly, lower levels of lutein and zeaxanthin have been found in autopsy specimens of donor eyes with AMD.³³

Conversely, epidemiologic studies of early and intermediate AMD, as a whole, do not support the hypothesis of a beneficial role of lutein and zeaxanthin. Indeed, the null association we observed between intermediate AMD and dietary lutein plus zeaxanthin intake in our primary analyses is consistent with the few previous studies that have examined these relationships,^{34 - 37} despite variations in the ranges of intake, methods of quantifying intake (micrograms vs micrograms per 1000 kcal), methods of defining AMD (incident vs prevalent cases; fundus photographs in 1 eye vs 2 eyes vs physician diagnosis), and different study populations. An additional study³⁸ found no association for any AMD (early plus late) but a modest direct association for soft drusen (drusen >63 μm), which is a smaller and more common lesion than the large drusen (≥125 μm) outcome used in CAREDS. It is possible that studies of early and intermediate AMD are not in agreement with the overall literature because of inconsistencies in the measurement of, and adjustment for, other potential risk factors. Unlike some previous studies, CAREDS collected data on most known potential risk factors for AMD, including some not usually controlled for, such as family history of AMD and sunlight exposure; however, inclusion of these factors as covariates in statistical models had little effect on our estimates of risk.

Associations between dietary lutein plus zeaxanthin intake and intermediate AMD were in the hypothesized, protective direction in women younger than 75 years, although not statistically significant (until women at risk for recent dietary change were excluded). One previous study found similarly protective associations for pigmentary abnormalities in people aged 40 to 59 years, but not among those aged 60 to 79 years or 80 years or more.³⁸ Competing risk factors, selective mortality bias, and dietary changes (usually because of health concerns) are likely causes of the inability to detect associations with AMD in elderly individuals. For example, older participants in observational studies may be more likely to have eaten diets rich in fruits and vegetables during their adult lifetime than people in their birth cohort who are no longer living, making it more difficult to discern the effects of those diets on disease outcomes.

The critical window of exposure to dietary and other factors for development of AMD is unknown. Therefore, examining individuals whose exposure has been constant for many years reduces the possibility of misclassification at the critical time that influences AMD development, particularly if the critical period encompasses many years or even decades. We were able to exclude from our analyses women whose intake of lutein plus zeaxanthin changed 4 to 15 years before assessment of AMD status. This reduced the ORs slightly in the overall sample; however, in women younger than 75 years, the reduction in prevalence odds was substantial for any intermediate AMD and large drusen and was in an inverse, but not statistically significant, direction for pigmentary abnormalities. Further exclusion of women with chronic conditions that may affect diet change, possibly even in the more distant past, reduced the estimated risks even further in women younger than 75 years. In contrast, the estimates in women 75 years and older were in a harmful direction, even after the aforementioned exclusions, although they were not statistically significant. Preliminary results in one other study population also indicated a potentially deleterious association between early and any AMD and high levels of lutein intake, although the associations by age group were not reported.³⁹ Considering the current lack of scientific plausibility to support a higher risk of AMD among people with high intakes of these plant pigments, we suspect that the observed ORs above 1.0 in CAREDS may reflect selective mortality bias or other similar issues, as discussed earlier. These results suggest that future research should examine the relationship between AMD and dietary factors, such as lutein and zeaxanthin intake, in more specific subgroups than in past studies.

Sunlight is a difficult exposure to measure although some previous studies have found visible (blue) light to be positively associated with risk for various stages of AMD.^{13 - 15 ,40 - 41} To our knowledge, only 1 other study has reported the relationship between lutein plus zeaxanthin intake and AMD after adjustment for sunlight exposure.⁴² We speculated that intake of lutein plus zeaxanthin might be more protective among women with high, rather than low, sun exposure. However, in CAREDS participants with stable intake of lutein plus zeaxanthin, a protective relationship between dietary lutein plus zeaxanthin and intermediate AMD was observed only in women with low sun exposure.

In CAREDS, the strong inverse relationships between intermediate AMD and intake of vegetables in general and green vegetables were observed only in women younger than 75 years with stable intake of lutein plus zeaxanthin and without the specified comorbid conditions. We did not observe any statistically significant associations in the overall sample or in the subsample of all women younger than 75 years, which is consistent with the null or weak associations observed in other studies of early and intermediate stages of AMD.^{35 ,37 ,43} One small study in Lithuania reported that consumption of fresh, uncooked vegetables at least twice a week compared with less than twice during the winter and spring months was inversely associated with any AMD (OR, 0.37; 95% CI, 0.15-0.90) in women aged 65 to 74 years.⁴⁴ Inverse associations with intakes of vegetables and fruits throughout the rest of the year were also observed but did not reach statistical significance.⁴⁴

Serum levels of lutein and zeaxanthin were not associated with intermediate AMD, even in women younger than 75 years and after exclusion of women with unstable intake of lutein and zeaxanthin and those at risk of diet change because of comorbidities. This was observed despite stronger correlations between macular pigment density and levels of lutein plus zeaxanthin in the serum compared with the diet in the CAREDS sample (J.A.M., unpublished data, 2005). Other studies have reported a similar lack of association in serum or plasma.^{38 ,45 - 46} In contrast, 1 study reported that low plasma levels of zeaxanthin were associated with an increased risk of any AMD, with similar, but not statistically significant, associations seen with plasma levels of lutein and lutein plus zeaxanthin.⁴⁷ Another study reported a protective association with high serum lutein and zeaxanthin levels and advanced AMD.⁴⁸ The lack of consistent relationships between serum lutein and zeaxanthin levels and AMD suggests that dietary lutein and zeaxanthin intake levels may be a marker for other dietary attributes that protect against AMD.

It is possible that there are combinations of foods that provide the optimal mix of nutrients to protect against AMD. Inconsistencies among studies of dietary factors for early and intermediate AMD may reflect differing levels of these other nutrients across study samples, or influences on risk ratios that are not easily adjusted for, such as those due to measurement error and the high correlation between intakes of many foods and nutrients. We explored this possibility by using a very crude measure of intake of several of the nutrients that have been associated with AMD in previous studies. While none of the associations was statistically significant, all of the point estimates suggested that consumption above the median of any of the selected micronutrients (and/or below the median for polyunsaturated fat) might confer some benefit, as compared with those with poor intakes of all of the selected nutrients. Because the interaction of these nutrients in the pathogenesis of AMD remains unclear, it will be challenging for future studies to evaluate the individual or joint benefits of potentially protective nutrients, or whether high intakes of certain nutrients can compensate for low intakes of others.

We considered advanced AMD as an exploratory outcome, as there were only 34 cases in CAREDS. As observed for intermediate AMD, the prevalence odds of advanced AMD were in the protective direction (although statistically nonsignificant) in women younger than 75 years, particularly in those reporting stable intake of lutein plus zeaxanthin. This is consistent with reports from 2 other study populations that examined dietary intakes^{42 ,49} and serum levels.⁴⁸ In contrast, 1 study reported a nonsignificant positive association,³⁴ although the lower range of intakes in the highest quintile of that sample was only 621 μg/1000 kcal, which is less than the median intake in the low lutein intake recruitment group in CAREDS (646 μg/1000 kcal).

Nonparticipation bias is a concern in the present study, as 36% of those eligible to participate in CAREDS declined. This nonparticipation could have biased our observed associations in either direction, depending on whether the nonparticipants had more photographic evidence of AMD. Because healthier people are more likely to participate in health studies such as the WHI, we would expect our associations to be biased toward the null if those who did not participate had more AMD.

It is important to note that the estimates of dietary intake of lutein and zeaxanthin in CAREDS represent diets high or low in lutein and zeaxanthin—and therefore everything else that goes along with such diets, including other vitamins, minerals, and nutritive and nonnutritive compounds such as phytochemicals and food additives. Further studies are needed to fully understand the complexity of how food components interact as they are absorbed in the diet and metabolized in the body, as well as how they act with other environmental and genetic factors to affect disease risk.

In conclusion, the hypothesized lower prevalence of intermediate AMD among women with high lutein and zeaxanthin intake was not observed in the overall CAREDS sample. However, there was evidence that diet instability may have biased the associations and, together with the possibility of selective mortality bias, may explain our inability to detect the hypothesized association overall. High lutein plus zeaxanthin intake was associated with lower risk of intermediate AMD in women younger than 75 years who had stable intake of lutein plus zeaxanthin and were without a history of cardiovascular disease, diabetes, hypertension, and/or previously diagnosed AMD. This exploratory observation is consistent with a broad body of evidence from observational and experimental studies that suggests that these carotenoids may protect against AMD. Still, given the numerous analyses performed in this study, our results could be due to chance. More conclusive evidence from long-term prospective studies and clinical trials is needed to determine whether the intake of macular carotenoids themselves, or as markers of broader dietary patterns, can protect against intermediate AMD or delay progression in individuals who have early stages of the disease.

Correspondence: Julie A. Mares, PhD, Department of Ophthalmology and Visual Sciences, University of Wisconsin, 610 N Walnut St, 1063 WARF Bldg, Madison, WI 53726-2336 (Jmarespe@wisc.edu).

Submitted for Publication: June 13, 2005; final revision received December 7, 2005; accepted February 7, 2006.

CAREDS Research Study Group: Catherine Allen, PhD, University of Wisconsin; Matthew Davis, MD, University of Wisconsin; Karen Gehrs, MD, University of Iowa Hospital and Clinics; Larry Hubbard, MAT, University of Wisconsin; Elizabeth Johnson, PhD, Tufts University USDA Human Nutrition Research Center on Aging; Michael Klein, MD, Oregon Health Science University/Casey Eye Institute, Portland; Cheryl Ritenbaugh, PhD, MPH, University of Arizona, Tucson; D. Max Snodderly, PhD, Medical College of Georgia, Augusta.

Financial Disclosure: None reported.

Funding/Support: This research was supported by National Institutes of Health grants EY13018 and DK 07665 and by Research to Prevent Blindness.

Acknowledgment: We are grateful to the Scientific Advisory Board (Natalie Kurinji, PhD, National Eye Institute; Sheila West, PhD [chair]; Neil Bressler, MD, Johns Hopkins University; Anne Lindblad, PhD, EMMES Corp; and Susan Mayne, PhD, Yale University) for their useful discussions and critical reading of the manuscript. We also thank the CAREDS staff in Portland (Paula Smith), Iowa City (Kelly O’Berry, Heather Stockman, Steven Wallace, and Lindsey Fuhrmeister), and Madison (Jane Armstrong, Michael Neider, Hugh Wabers, Janet Rowley, Tanya Judge, Lisa Oxton, Rick Voland, PhD, Gail Ostrowski, Scott Burfield, Tara LaRowe, PhD, Julie Ewing, and Tracy Perkins) for their dedicated work.

We would also like to thank and acknowledge the WHI investigators, staff, and participants for their time and effort in obtaining the WHI data that were presented in this article. Specifically, we would like to thank those listed in the box titled “Women's Health Initiative Investigators and Staff”. Article

Program Office

National Heart, Lung, and Blood Institute, Bethesda, Md: Barbara Alving, Jacques Rossouw, Linda Pottern.

Clinical Coordinating Center

Fred Hutchinson Cancer Research Center, Seattle, Wash: Ross Prentice, Garnet Anderson, Andrea LaCroix, Charles L. Kooperberg, Ruth E. Patterson, Anne McTiernan. Wake Forest University School of Medicine, Winston-Salem, NC: Sally Shumaker. Medical Research Labs, Highland Heights, Ky: Evan Stein. University of California at San Francisco: Steven Cummings.

Clinical Centers

Albert Einstein College of Medicine, Bronx, NY: Sylvia Wassertheil-Smoller. Baylor College of Medicine, Houston, Tex: Jennifer Hays. Brigham and Women's Hospital, Harvard Medical School, Boston, Mass: JoAnn Manson. Brown University, Providence, RI: Annlouise R. Assaf. Emory University, Atlanta, Ga: Lawrence Phillips. Fred Hutchinson Cancer Research Center: Shirley Beresford. George Washington University Medical Center, Washington, DC: Judith Hsia. Harbor-UCLA Research and Education Institute, Torrance, Calif: Rowan Chlebowski. Kaiser Permanente Center for Health Research, Portland, Ore: Evelyn Whitlock. Kaiser Permanente Division of Research, Oakland, Calif: Bette Caan. Medical College of Wisconsin, Milwaukee: Jane Morley Kotchen. MedStar Research Institute/Howard University, Washington, DC: Barbara V. Howard. Northwestern University, Chicago/Evanston, Ill: Linda Van Horn. Rush-Presbyterian St Luke's Medical Center, Chicago, Ill: Henry Black. Stanford Prevention Research Center, Stanford, Calif: Marcia L. Stefanick. State University of New York at Stony Brook: Dorothy Lane. Ohio State University, Columbus: Rebecca Jackson. University of Alabama at Birmingham: Cora E. Lewis. University of Arizona, Tucson/Phoenix: Tamsen Bassford. University at Buffalo, Buffalo, NY: Jean Wactawski-Wende. University of California at Davis, Sacramento: John Robbins. University of California at Irvine, Orange: Allan Hubbell. University of California at Los Angeles: Howard Judd. University of California at San Diego, LaJolla/Chula Vista: Robert D. Langer. University of Cincinnati, Cincinnati, Ohio: Margery Gass. University of Florida, Gainesville/Jacksonville: Marian Limacher. University of Hawaii, Honolulu: David Curb. University of Iowa, Iowa City/Davenport: Robert Wallace. University of Massachusetts/Fallon Clinic, Worcester: Judith Ockene. University of Medicine and Dentistry of New Jersey, Newark: Norman Lasser. University of Miami, Miami, Fla: Mary Jo O’Sullivan. University of Minnesota, Minneapolis: Karen Margolis. University of Nevada, Reno: Robert Brunner. University of North Carolina, Chapel Hill: Gerardo Heiss. University of Pittsburgh, Pittsburgh, Pa: Lewis Kuller. University of Tennessee, Memphis: Karen C. Johnson. University of Texas Health Science Center, San Antonio: Robert Brzyski. University of Wisconsin, Madison: Gloria E. Sarto. Wake Forest University School of Medicine, Winston-Salem, NC: Denise Bonds. Wayne State University School of Medicine/Hutzel Hospital, Detroit, Mich: Susan Hendrix.