Intake of Blueberries, Anthocyanins, and Risk of Eye Disease in Women

Abstract

Background

Blueberries and anthocyanins, their key bioactive component, may improve eye health. However, few long-term studies have examined blueberries and anthocyanins with cataract and age-related macular degeneration (AMD).

Objectives

To investigate the prospective association between blueberry and anthocyanin intake with incident cataract, total AMD, and visually significant AMD among middle-aged and older women.

Methods

A total of 36,653 and 35,402 women initially free of AMD and cataract, respectively, aged ≥45 y from the Women’s Health Study provided semiquantitative food frequency questionnaire data on blueberry intake categorized as none, 1–3 servings/mo, 1 serving/wk, or ≥2 servings/wk, plus a combined category of ≥1 serving/wk. Total anthocyanin intake and major subclasses were energy-adjusted and categorized into quintiles. Self-reported risk factors of eye disease were adjusted in multivariable hazard ratios (HRs) (95% confidence intervals [CIs]) of confirmed cataract, AMD, and visually significant AMD with mean follow-up of 11 y.

Results

Among the participants, 10.5% consumed ≥1 serving/wk of blueberries, with mean total anthocyanin intake of 11.2 mg/d. Compared to no blueberry intake, women consuming 1–3 servings/mo, 1 serving/wk, and ≥2 servings/wk had corresponding multivariable HRs of total AMD of 0.90 (95% CI: 0.73, 1.11), 0.71 (95% CI: 0.50, 1.00), and 0.36 (95% CI: 0.14, 0.93) (P_trend = 0.011); those consuming ≥1 servings/wk had an HR of 0.68 (95% CI: 0.47, 0.98). A similar magnitude of HRs were found for visually significant AMD (P_trend = 0.012) but not for cataract. There were no significant associations between increasing total anthocyanin quintiles and total and visually significant AMD, but there was a modest inverse association with cataract (P_trend = 0.022), driven by a 10% reduction in cataract in the upper 2 quintiles.

Conclusions

Greater blueberry intake significantly reduced total AMD, but not visually significant AMD or cataract. However, the magnitude of effect for visually significant AMD was similar to total AMD. There was a modest but significant inverse association between dietary anthocyanin intake with cataract but not AMD.

Introduction

A diet rich in fruits and vegetables has been consistently associated with reductions in total and cause-specific mortality [1], supporting current United States dietary guidelines that emphasize high intake across a diverse array of nutrient-rich foods [2]. To that end, research has sought to understand whether particular fruits and vegetables may confer greater benefits on particular health outcomes based on their nutritional content and potential mechanisms. For age-related eye disease, including cataract and age-related macular degeneration (AMD), most research has focused on the roles of dietary patterns, carotenoids, and phytochemicals [3,4].

Blueberries are a good source of fiber, vitamin C, and anthocyanins, a major subclass of flavonoids and a bioactive component that gives blueberries their characteristic blue color. Anthocyanins are the dominant flavonoid in many berries and have potent antioxidant properties, which is particularly important in vision and eye health as oxidative stress has been implicated in the pathogenesis of age-related eye diseases [5]. Animal studies have shown that a diet supplemented with blueberries results in intact anthocyanins in ocular tissues [6,7] and inhibits retinal oxidative stress and inflammation [8]. Further, small short-term interventions with anthocyanins have suggested improvements in the electrical signal generated by the retina following a light stimulus [9] and faster recovery of visual acuity after photobleaching [10].

The Women’s Health Study (WHS) is well-positioned to examine the association between blueberry intake, anthocyanin intake, and medical record-confirmed cataract and AMD in a large cohort of middle-aged and older women with more than a decade of follow-up.

Methods

Study population: WHS

The WHS [[11], [12], [13]] is a completed, randomized, double-blind, placebo-controlled trial of low-dose aspirin and vitamin E in the primary prevention of cardiovascular disease and cancer. In 1992, 39,876 female United States health professionals aged ≥45 y, postmenopausal or not intending to become pregnant, and free from prior myocardial infarction, stroke, transient ischemic attack, and cancer (except nonmelanoma skin cancer) were randomized. Written informed consent was provided by all participants, and the institutional review board at Brigham and Women’s Hospital approved the research protocol.

We excluded women with self-reported prerandomization angina, coronary artery bypass graft surgery, or percutaneous transluminal coronary artery angioplasty. Cataract or AMD was also excluded for their respective analyses. At baseline, a 131-item validated [14] semiquantitative food frequency questionnaire (SFFQ) was completed by 39,310 women [15], of whom 829 were excluded due to insufficient completion of food items or energy intake of either <2510 or ≥14,644 kJ/d (<600 or ≥3500 kcal/d). Finally, we excluded participants with missing data on blueberry intake on the SFFQ. After these exclusions, 37,653 and 35,402 women were included in our analyses of AMD and cataract, respectively. The detailed participant flow chart is shown in Supplemental Figure 1.

Assessment of blueberry intake, anthocyanins, and other foods

For each individual food and beverage included on the SFFQ, a common unit or portion size was specified, and the participant was directed to fill in their average use during the past year. For seasonal foods, the instructions indicated that the participant should try to average their seasonal use of foods over the entire year. For consumption of each food, participants selected from 9 responses: never or less than once per month, 1–3 per month, once per week, 2–4 per week, 5–6 per week, once per day, 2–3 per day, 4–5 per day, and 6+ per day. Blueberry intake was defined as “blueberries, fresh, frozen, or canned (1/2 cup).” The measurement of specific nutrients from the SFFQ was based upon food tables maintained by the Department of Nutrition at the Harvard T.H. Chan School of Public Health, Boston, MA, United States. Dietary intake of total anthocyanins and its major subclasses was estimated from the database maintained by the United States Department of Agriculture [16]. The measured subclasses of dietary anthocyanins included cyanidin, delphinidin, malvidin, pelargonidin, petunidin, and peonidin. We also considered total dietary flavonoids along with flavanones, flavan-3-ols, flavonols, and flavones as subclasses. All nutrient values were energy-adjusted to account for differences in the proportion of blueberry and anthocyanin intake relative to a fixed level of total caloric intake using the residual method [17]. Dietary factors reflecting the participants’ broader dietary patterns included fruit and vegetable intake (servings/day), dairy intake (servings/day), saturated fat intake (g/day), total fiber intake (g/day), folate intake (μg/day), and total flavonoid intake (mg/day).

Other covariates

On the WHS baseline questionnaire, women provided self-reported information on potential risk factors such as age (in years), weight and height (converted to BMI, in kg/m²), smoking status (categorized as never, former, and current), alcohol use (categorized as rarely/never, 1–3 drinks/mo, 1–6 drinks/wk, and ≥1 drink/d), frequency of vigorous exercise (categorized as rarely/never, <1 times/wk, 1–3 times/wk, and ≥4 times/wk), history of an eye examination in the last 2 y (no, yes), history of hypertension (no, yes), history of hypercholesterolemia (no, yes), history of diabetes (no, yes), postmenopausal status (no, yes), and postmenopausal hormone use (categorized as never, former, and current). We additionally adjusted for the randomized WHS treatment assignments.

Outcome ascertainment

Participants completed annual questionnaires updating information on health outcomes including cataract and AMD (separate for right and left eye). Participants with new diagnoses of cataract, cataract extraction, and macular degeneration were asked to provide written consent to obtain medical records for the reported eye endpoint. Ophthalmologists/optometrists were contacted by mail and asked to provide information about the reported endpoint by completing a questionnaire or, alternatively, supplying copies of relevant medical records.

The cataract questionnaire asked about the presence of lens opacities, diagnosis date, best-corrected visual acuity, cataract extraction, and cataract type and etiology. The questionnaire also asked whether there were other ocular abnormalities and, if so, whether the cataract, by itself, was significant enough to reduce best-corrected visual acuity to 20/30 or worse. The AMD questionnaire requested information about diagnosis date, best-corrected visual acuity at diagnosis, and date when best-corrected visual acuity reached 20/30 or worse. Information was also requested about signs of AMD observed (drusen, retinal pigment epithelium [RPE] hypo/hyperpigmentation, geographic atrophy, RPE detachment, subretinal neovascular membrane, or disciform scar) when visual acuity was first noted to be 20/30 or worse, and the date exudative neovascular disease (defined by presence of RPE detachment, subretinal neovascular membrane, or disciform scar) was first noted. The questionnaire also asked whether there were other ocular abnormalities and, if so, whether the AMD, by itself, was significant enough to reduce best-corrected visual acuity to 20/30 or worse.

The primary endpoints of the original WHS trial were cataract and visually significant AMD confirmed by medical record evidence of an initial diagnosis after baseline but on or before 31 March, 2004 (scheduled end of the trial), with best-corrected visual acuity reduced to 20/30 or worse attributable to the endpoint examined (cataract or visually significant AMD) over a mean follow-up of 11 y (maximum follow-up, 12.8 y). In this observational analysis, an additional endpoint of total AMD was included for all confirmed diagnoses of AMD.

Data analyses

We categorized blueberry intake as rarely/never, 1–3 servings/mo, 1 serving/wk, and ≥2 servings/wk. We first compared mean values (analysis of variance) or proportions (chi-square tests) of lifestyle, clinical, and dietary risk factors according to categories of blueberry intake. We then calculated quintiles of total anthocyanin intake and its 6 subclasses, repeating the above descriptive analyses of baseline risk factors by quintiles of total anthocyanin intake.

We calculated hazard ratios (HRs) and 95% confidence interval (CIs) of cataract, total AMD, and visually significant AMD for increasing levels of blueberry intake, with the lowest category (rarely/never) serving as the reference group using Cox proportional hazards models. Given the relatively low levels of blueberry intake and consequent numbers of outcomes in the highest categories of blueberry intake in this population of women, we also considered a combined upper intake category of ≥1 serving/wk with the same referent (rarely/never).

For quintiles of total anthocyanins or anthocyanin subclasses (cyanidin, delphinidin, malvidin, pelargonidin, petunidin, and peonidin), the lowest quintile was the referent. All analyses were based on the time to the first confirmed eye endpoint. The proportional hazards assumption was satisfied for each of the analyses (all P > 0.05). We initially adjusted for age, total energy intake, and randomized treatment assignments; model 1 further adjusted for BMI, smoking status, frequency of exercise, alcohol intake, eye exam within the last 2 y, hypertension, hypercholesterolemia, diabetes, postmenopausal status, and postmenopausal hormone use. Model 2 then added dietary factors related to blueberry intake, and in model 3, we evaluated whether the association of blueberry or subclasses of anthocyanin intake may be mediated after adjustment for total anthocyanin or total flavonoid intake. Linear trend tests across categories of blueberry intake or quintiles of anthocyanin intake were assessed using the median value for each category as an ordinal variable. All statistical analyses were performed with SAS 9.3 (SAS Institute Inc.).

Results

Women were of mean ages of 54.5 and 53.9 y for analyses of AMD and cataract, respectively. Blueberry intake was modest, with 64.8%, 24.7%, 8.4%, and 2.1% of women consuming none, 1–3 servings/mo, 1 serving/wk, and ≥2 servings/wk of blueberries, respectively. In Table 1, higher levels of blueberry intake resulted in expected increases in total anthocyanin intake (P < 0.001), likely reflecting not only greater blueberry intake but also fruit and vegetable intake and other potential sources of anthocyanins. Even at relatively low levels of blueberry intake, there was still a 4–5-fold increase in mean dietary anthocyanins from 1–3 servings/mo to ≥2 servings/wk for blueberries (from 16.8 to 73.4 mg/d). Women consuming greater amounts of blueberries were also slightly more likely to be older, have hypertension, exercise regularly, and have an eye exam within the last 2 y, and were less likely to be a current smoker and consume ≥1 drink/d of alcohol (all P < 0.001). In turn, women consuming more blueberries also followed a healthier dietary pattern of higher fiber and folate intake (both P < 0.001) and lower saturated fat intake (P < 0.001). Individual subclasses of anthocyanins all increased markedly with increasing categories of blueberry intake as well (all P < 0.001).

TABLE 1.

Comparison of baseline risk factors by categories of blueberry intake in 37,653 women aged ≥45 y in the Women’s Health Study free of baseline age-related macular degeneration

	Categories of blueberry intake				P value1
	None (n = 24,404)	1–3 servings/mo (n = 9304)	1 serving/wk (n = 3174)	≥2 servings/wk (n = 771)
Age (y)	54.4 ± 6.92	54.5 ± 6.9	54.9 ± 7.2	55.1 ± 7.5	<0.001
BMI (kg/m2)	26.1 ± 5.1	25.8 ± 4.8	25.8 ± 4.8	26.5 ± 5.5	<0.001
History of hypertension (%)	26.3	23.9	23.9	27.6	<0.001
History of hypercholesterolemia (%)	29.8	28.1	28.7	28.7	0.022
History of diabetes (%)	2.5	2.3	2.6	3.4	0.24
History of eye exam within last 2 y (%)	82.2	83.5	84.6	83.8	<0.001
Vigorous exercise (%)					<0.001
Rarely/never	41.8	32.3	29.3	29.4
<1 times/wk	19.5	21.2	19.9	17.3
1–3 times/wk	29.2	34.6	36.8	37.3
≥4 times/wk	9.6	11.8	14.0	16.1
Smoking status (%)					<0.001
Never	49.8	53.0	53.6	51.5
Former	35.1	37.2	37.9	40.6
Current	15.1	9.8	8.5	7.9
Postmenopausal (%)	53.6	54.1	56.6	57.0	0.006
Postmenopausal hormone use (%)					0.004
Never	47.8	49.2	46.8	45.2
Former	10.4	9.1	10.3	11.2
Current	41.8	41.7	42.9	43.5
Alcohol consumption (%)					<0.001
Rarely/never	46.9	39.7	41.1	46.6
1–3 drinks/mo	13.1	13.4	13.3	13.8
1–6 drinks/wk	29.5	36.6	36.0	32.3
≥1 drink/d	10.5	10.4	9.6	7.3
Total caloric intake (kcal/d)	1660 ± 523	1825 ± 526	1883 ± 530	1997 ± 575	<0.001
Fruit/vegetable intake (servings/d)	5.4 ± 3.0	6.7 ± 3.1	7.8 ± 3.3	9.9 ± 4.4	<0.001
Total fiber intake (g/d)3	18.2 ± 5.8	19.9 ± 5.7	21.2 ± 5.8	23.8 ± 6.5	<0.001
Folate intake (μg/d)3	420 ± 230	436 ± 210	454 ± 212	471 ± 217	<0.001
Saturated fat intake (g/d)3	20.2 ± 5.0	19.1 ± 4.5	18.4 ± 4.5	17.2 ± 4.6	<0.001
Total anthocyanin intake (mg/d)3	5.27 ± 5.19	16.8 ± 7.2	25.1 ± 10.2	73.4 ± 43.5	<0.001
Cyanidin intake (mg/d)3	1.68 ± 1.42	2.99 ± 1.42	3.90 ± 1.64	8.82 ± 4.45	<0.001
Delphinidin intake (mg/d)3	0.24 ± 0.39	3.44 ± 2.00	5.48 ± 3.07	17.9 ± 14.2	<0.001
Malvidin intake (mg/d)3	0.71 ± 2.49	4.92 ± 3.49	7.57 ± 4.84	23.5 ± 19.7	<0.001
Pelargonidin intake (mg/d)3	2.57 ± 3.88	3.46 ± 3.71	4.62 ± 4.29	9.95 ± 8.89	<0.001
Petunidin intake (mg/d)3	0.16 ± 0.32	1.93 ± 1.09	3.06 ± 1.65	9.93 ± 7.69	<0.001
Peonidin intake (mg/d)3	0.09 ± 0.26	0.83 ± 0.43	1.31 ± 0.63	4.25 ± 2.98	<0.001

¹We used a global analysis of variance test for continuous variables and chi-square tests (3 degrees of freedom) for categorical variables.

²Mean ± SD.

³Energy-adjusted.

We then compared baseline risk factors by quintiles of total anthocyanin intake in Table 2. The overall mean (±SD) total anthocyanin intake was 11.2 (±14.3) mg/d, primarily as malvidin, pelargonidin, cyanidin, and delphinidin. Petunidin and peonidin contributed comparatively less to total anthocyanin intake across all quintiles. Women consuming greater amounts of anthocyanins were less likely to have hypertension, hypercholesterolemia, and be a current smoker (all P < 0.05). These women were also more likely to be older, have a lower BMI, and have an eye exam in the last 2 y. Among dietary factors evaluated, fruit and vegetable, blueberry, fiber, and folate intake were all positively associated with higher intake of total anthocyanins (all P < 0.001), which also extended to each of the anthocyanin subclasses.

TABLE 2.

Comparison of baseline risk factors by quintiles of total anthocyanin intake in 37,653 women aged ≥45 y in the Women’s Health Study free of baseline age-related macular degeneration

	Quintiles of total anthocyanin intake					P value1
	1st (n= 7543)	2nd (n = 7532)	3rd (n = 7535)	4th (n = 7529)	5th (n = 7514)
Range of intake (mg/d)	<2.9	2.9 to <5.05	5.05 to <9.54	9.54 to <17.11	≥17.12
Age (y)	54.1 ± 6.82	54.2 ± 6.8	54.7 ± 7.0	54.5 ± 7.0	55.0 ± 7.1	<0.001
BMI (kg/m2)	26.6 ± 5.5	26.2 ± 5.1	25.9 ± 4.8	25.8 ± 4.8	25.6 ± 4.7	<0.001
History of hypertension (%)	28.4	26.0	25.1	24.1	23.9	<0.001
History of hypercholesterolemia (%)	30.6	29.7	30.2	27.6	28.3	<0.001
History of diabetes (%)	2.7	2.6	2.3	2.3	2.6	0.38
History of eye exam within last 2 y (%)	80.4	82.7	83.4	83.4	83.8	<0.001
Exercise (%)						<0.001
Rarely/never	50.3	40.8	36.8	32.7	30.0
<1 times/wk	19.8	20.7	19.7	20.5	18.9
1–3 times/wk	23.6	29.6	31.9	34.9	36.6
≥4 times/wk	6.3	8.9	11.5	11.9	14.5
Smoking status (%)						<0.001
Never	45.5	51.8	52.7	53.0	51.7
Former	32.4	34.8	36.0	37.6	39.0
Current	22.0	13.4	11.3	9.4	9.3
Postmenopausal (%)	50.9	53.2	55.7	54.6	55.9	<0.001
Postmenopausal hormone use (%)						<0.001
Never	50.6	48.3	45.9	49.0	46.1
Former	11.2	10.0	9.8	9.6	9.8
Current	38.2	41.6	44.3	41.3	44.1
Alcohol consumption (%)						<0.001
Rarely/never	53.4	48.2	42.9	39.1	39.5
1–3 drinks/mo	12.8	13.8	13.7	12.8	13.0
1–6 drinks/wk	23.8	29.4	34.0	36.5	35.4
≥1 drink/d	10.0	8.6	9.4	11.6	12.1
Total caloric intake (kcal/d)	1718 ± 5832	1779 ± 484	1655 ± 577	1908 ± 494	1572 ± 457	<0.001
Fruit/vegetable intake (servings/d)	4.6 ± 2.8	5.7 ± 2.9	5.9 ± 3.0	7.1 ± 3.3	6.9 ± 3.4	<0.001
Blueberry intake (servings/d)	0.0	0.0	0.01 ± 0.02	0.06 ± 0.04	0.13 ± 0.17	<0.001
Total fiber intake (g/d)3	15.8 ± 5.2	17.9 ± 5.0	19.5 ± 5,5	20.0 ± 5.8	21.7 ± 6.3	<0.001
Folate intake (μg/d)3	394 ± 231	414 ± 213	435 ± 230	435 ± 202	462 ± 237	<0.001
Saturated fat intake (g/d)3	21.7 ± 5.3	20.3 ± 4.6	19.4 ± 4.5	19.0 ± 4.5	18.1 ± 4.5	<0.001
Total anthocyanin intake (mg/d)3	1.58 ± 0.80	3.94 ± 0.61	6.90 ± 1.28	13.2 ± 2.14	30.4 ± 21.9	<0.001
Cyanidin intake (mg/d)3	0.82 ± 0.54	1.47 ± 0.83	2.22 ± 1.24	2.70 ± 1.34	4.49 ± 2.69	<0.001
Delphinidin intake (mg/d)3	0.11 ± 0.11	0.17 ± 0.14	0.39 ± 0.42	2.10 ± 1.11	6.41 ± 6.57	<0.001
Malvidin intake (mg/d)3	0.17 ± 0.28	0.30 ± 0.48	0.79 ± 1.04	3.19 ± 1.82	9.54 ± 9.52	<0.001
Pelargonidin intake (mg/d)3	0.39 ± 0.58	1.88 ± 1.08	3.23 ± 1.86	3.25 ± 2.76	6.82 ± 7.30	<0.001
Petunidin intake (mg/d)3	0.08 ± 0.08	0.11 ± 0.10	0.24 ± 0.26	1.21 ± 0.61	3.59 ± 3.59	<0.001
Peonidin intake (mg/d)3	0.03 ± 0.03	0.05 ± 0.05	0.12 ± 0.14	0.56 ± 0.27	1.56 ± 1.46	<0.001

¹We used a global analysis of variance test for continuous variables and chi-square tests (4 degrees of freedom) for categorical variables.

²Mean ± SD.

³Energy-adjusted.

During a mean follow-up of 11.0 y (maximum follow-up, 12.8 y), there was a total of 2274 new cases of cataract, including 1485 new cases of cataract extraction. There were 2087, 845, and 684 total cases of nuclear sclerotic, cortical, and posterior subcapsular cataract types, respectively, and a corresponding 1371, 567, and 550 total extractions among these, including participants with incident, recurrent, and multiple types of cataracts. In Table 3, in models adjusted for age, total caloric intake, and randomized WHS treatment assignments, we observed no association between blueberry intake and the risk of cataract (P_trend = 0.47). Similar associations were observed additionally adjusting for lifestyle, clinical, and dietary factors in models 2 and 3 (P_trend = 0.81 and 0.94, respectively). Further adjustment for total anthocyanin intake also had no material impact on the HRs in model 3, nor did adjustment for total flavonoid intake (data not shown).

TABLE 3.

Hazard ratios (95% confidence intervals) of cataract, total age-related macular degeneration, and visually significant age-related macular degeneration by categories of blueberry intake

	Categories of blueberry intake				Ptrend1	≥1 serving/wk
	None	1–3 servings/mo	1 serving/wk	≥2 servings/wk
Cataract
Number of women	22,954	8756	2973	719		3692
Number of events	1513	502	203	56		259
Initial model2	1.00 (reference)	0.86 (0.77, 0.95)	0.95 (0.82, 1.10)	1.05 (0.80, 1.37)	0.47	0.97 (0.85, 1.11)
Multivariable model 13 (+ lifestyle and clinical factors)	1.00 (reference)	0.88 (0.80, 0.98)	0.98 (0.85, 1.14)	1.08 (0.82, 1.41)	0.81	1.00 (0.88, 1.14)
Multivariable model 24 (+ dietary factors)	1.00 (reference)	0.89 (0.80, 0.98)	1.00 (0.86, 1.16)	1.08 (0.82, 1.42)	0.94	1.02 (0.89, 1.16)
Multivariable model 35 (+ total anthocyanin intake)	1.00 (reference)	0.91 (0.81, 1.02)	1.03 (0.87, 1.23)	1.21 (0.80, 1.83)	0.56	1.03 (0.87, 1.23)
Total AMD
Number of women	24,404	9304	3174	771		3945
Number of events	381	134	40	7		47
Initial model2	1.00 (reference)	0.92 (0.75, 1.12)	0.74 (0.53, 1.02)	0.50 (0.23, 1.05)	0.012	0.69 (0.51, 0.93)
Multivariable model 13 (+ lifestyle and clinical factors)	1.00 (reference)	0.93 (0.76, 1.14)	0.75 (0.54, 1.05)	0.52 (0.24, 1.09)	0.02	0.71 (0.52, 0.96)
Multivariable model 24 (+ dietary factors)	1.00 (reference)	0.95 (0.77, 1.16)	0.77 (0.55, 1.08)	0.51 (0.24, 1.10)	0.029	0.72 (0.53, 0.98)
Multivariable model 35 (+ total anthocyanin intake)	1.00 (reference)	0.90 (0.73, 1.11)	0.71 (0.50, 1.00)	0.36 (0.14, 0.93)	0.011	0.68 (0.47, 0.98)
Visually significant AMD
Number of women	24,404	9304	3174	771		3945
Number of events	162	52	19	2		21
Initial model2	1.00 (reference)	0.84 (0.61, 1.15)	0.80 (0.49, 1.29)	0.31 (0.08, 1.27)	0.04	0.70 (0.44, 1.10)
Multivariable model 13 (+ lifestyle and clinical factors)	1.00 (reference)	0.87 (0.64, 1.20)	0.83 (0.52, 1.34)	0.32 (0.08, 1.30)	0.06	0.72 (0.46, 1.15)
Multivariable model 24 (+ dietary factors)	1.00 (reference)	0.90 (0.65, 1.24)	0.86 (0.53, 1.39)	0.31 (0.08, 1.25)	0.07	0.74 (0.46, 1.17)
Multivariable model 35 (+ total anthocyanin intake)	1.00 (reference)	0.81 (0.59, 1.13)	0.72 (0.44, 1.19)	0.13 (0.02, 0.83)	0.012	0.64 (0.38, 1.10)

Abbreviation: AMD, age-related macular degeneration.

¹Trend test for linearity across 4 categories of blueberry intake, using median values for each category.

²Adjusted for age (in years), randomized treatments (aspirin, vitamin E, and β-carotene), and total energy intake (in kcal/week).

³Adjusted for age, randomized treatments, and total energy intake, plus BMI (in kg/m²), smoking status (never, former, current), exercise (rarely/never, <1 time/wk, 1–3 times/wk, ≥4 times/wk), history of eye examination within the last 2 y (no, yes), history of hypertension (no, yes), history of diabetes (no, yes), history of hypercholesterolemia (no, yes), postmenopausal status (no, yes), postmenopausal hormone use (never, past, current).

⁴Adjusted for the variables in multivariable model 1 plus alcohol (rarely/never, 1–3 drinks/mo, 1–6 drinks/wk, and ≥1 drink/d), fruit and vegetable intake (servings/day), dairy intake (servings/day), folate intake (μg/day), saturated fat intake (g/day), and fiber intake (g/day).

⁵Adjusted for the variables in multivariable model 2 plus total anthocyanin intake (in mg/day).

During follow-up, there also were 562 total cases of age-related AMD and 235 cases of visually significant AMD. There were small numbers of total AMD and visually significant AMD in the 2 highest categories of blueberry intake, reflecting both the modest intake levels in middle-aged and older women assessed in the early 1990s and the lower events rates for this outcome. We therefore considered a combined upper category of ≥1 serving/wk of blueberries. In Table 3, in an initial model adjusting for age, total caloric intake, and the randomized WHS treatment assignments, it was observed that women consuming ≥1 serving/wk of blueberries had a significant 31% (95% CI: 7, 49%) reduction in total AMD (P_trend = 0.012). As we expanded our multivariable-adjusted models, the magnitude of HRs became modestly attenuated. When further adjusting for lifestyle, clinical, and dietary factors, there was still a significant 28% (95% CI: 2, 47%) reduction in total AMD (P_trend = 0.029). Additional adjustment for total anthocyanin or total flavonoid intake did not attenuate these associations (P_trend = 0.011). A similar pattern in lower HRs was observed for visually significant AMD (P_trend = 0.012), but the HRs were not significant after adjusting for lifestyle, clinical, and dietary factors and accounting for the lower case counts.

In Table 4, we examined the association between baseline dietary intake of total anthocyanins with the risk of developing cataract and total and visually significant AMD. In initial models, women in the upper 2 quintiles of total anthocyanin intake had a significant 15% (95% CI: 3, 26%) reduction in cataract (P_trend = 0.002). However, adjustment for lifestyle and clinical risk factors attenuated the results to nonsignificance for incident cataract with a significant trend remaining across quintiles (P_trend = 0.02). The addition of dietary factors resulted in a trend across categories that became significant for cataract (P_trend = 0.022) and nonsignificant for total AMD (P_trend = 0.11). A similar pattern in HRs and impact of adjusting for potentially confounding factors was also seen for total AMD. Finally, we found no evidence for an association between increasing total anthocyanin intake and the risk of visually significant AMD.

TABLE 4.

Hazard ratios (95% confidence intervals) of cataract, total age-related macular degeneration, and visually significant age-related macular degeneration by quintiles of total anthocyanin intake

	Quintiles of total anthocyanin intake					Ptrend1
	1st	2nd	3rd	4th	5th
Range of intake (mg/d)	<2.90	2.90 to <5.05	5.05 to <9.54	9.54 to 17.12	≥17.12
Cataract
Number of women	7111	7106	7048	7089	7048
Number of events	463	472	472	419	448
Initial model2	1.00 (reference)	1.00 (0.88, 1.14)	0.93 (0.82, 1.06)	0.84 (0.74, 0.96)	0.85 (0.74, 0.97)	0.002
Multivariable model 13 (+ lifestyle and clinical factors)	1.00 (reference)	1.03 (0.91, 1.18)	0.97 (0.85, 1.11)	0.90 (0.78, 1.03)	0.90 (0.79-1.02)	0.02
Multivariable model 24 (+ dietary factors)	1.00 (reference)	1.03 (0.91, 1.18)	0.97 (0.85, 1.10)	0.90 (0.78, 1.03)	0.89 (0.78, 1.02)	0.022
Total AMD
Number of women	7543	7532	7535	7529	7514
Number of events	111	121	121	106	103
Initial model2	1.00 (reference)	1.07 (0.83, 1.38)	0.98 (0.76, 1.27)	0.88 (0.67, 1.14)	0.80 (0.61, 1.05)	0.023
Multivariable model 13 (+ lifestyle and clinical factors)	1.00 (reference)	1.11 (0.85, 1.44)	1.02 (0.79, 1.33)	0.93 (0.71, 1.21)	0.84 (0.64, 1.10)	0.047
Multivariable model 24 (+ dietary factors)	1.00 (reference)	1.15 (0.88, 1.49)	1.07 (0.82, 1.39)	0.98 (0.74, 1.29)	0.88 (0.66, 1.17)	0.11
Visually significant AMD
Number of women	7543	7532	7535	7529	7514
Number of events	49	49	46	48	43
Initial model2	1.00 (reference)	0.97 (0.66, 1.45)	0.83 (0.55, 1.23)	0.86 (0.58, 1.29)	0.72 (0.48, 1.09)	0.11
Multivariable model 13 (+ lifestyle and clinical factors)	1.00 (reference)	1.07 (0.72, 1.60)	0.91 (0.61, 1.38)	0.97 (0.65, 1.46)	0.81 (0.53, 1.23)	0.24
Multivariable model 24 (+ dietary factors)	1.00 (reference)	1.15 (0.77, 1.73)	1.00 (0.66, 1.51)	1.07 (0.70, 1.62)	0.89 (0.58, 1.38)	0.42

Abbreviation: AMD, age-related macular degeneration.

¹Trend test for linearity across quintiles of blueberry intake, using median values for each quintile.

²Adjusted for age, randomized treatments, and total energy intake.

³Adjusted for age, randomized treatments, and total energy intake, plus BMI, smoking status, exercise, history of eye examination within the last 2 y, history of hypertension, history of diabetes, history of hypercholesterolemia, postmenopausal status, and postmenopausal hormone use.

⁴Adjusted for the variables in multivariable model 1 plus alcohol, fruit and vegetable intake, dairy intake, folate intake, saturated fat intake, and fiber intake.

We then evaluated the association of individual subclasses of anthocyanins on eye endpoints in Table 5. Overall, similar to the confounding patterns observed for total anthocyanin intake, additional adjustment for lifestyle and dietary factors generally weakened or attenuated the initial HRs for the anthocyanin subclasses. This was particularly evident for HRs of cataract. For example, the initial HRs of cataract for increasing quintiles of dietary cyanidin were 1.00 (reference), 0.80 (95% CI: 0.70, 0.91), 0.82 (95% CI: 0.72, 0.94), 0.77 (95% CI: 0.68, 0.88), and 0.82 (95% CI: 0.73, 0.93) (P_trend = 0.027); upon full multivariable adjustment, the HRs remained significant for increasing quintiles of cyanidin at 1.00 (reference), 0.82, 0.87, 0.81, and 0.86 but with no significant linear trend (P_trend = 0.17) due to a potential threshold effect starting at the second quintile. For delphinidin intake, the initial HRs of cataract for increasing quintiles of dietary cyanidin were 1.00 (reference), 0.91 (95% CI: 0.80, 1.03), 0.98 (95% CI: 0.87, 1.11), 0.78 (95% CI: 0.68, 0.89), and 0.85 (95% CI: 0.75, 0.97) (P_trend = 0.007), and multivariable adjustment only modestly changed the HRs in Table 5. In multivariable-adjusted models of anthocyanin subclasses, delphinidin, malvidin, petunidin, and peonidin intake were inversely associated with cataract (all P_trend < 0.05), largely driven by 10% to 18% reductions in the upper 2 quintiles of intake. However, anthocyanin subclasses were not associated with the risk of total AMD or visually significant AMD (all P_trend > 0.05).

TABLE 5.

Multivariable¹ hazard ratios (95% confidence intervals) of cataract, total age-related macular degeneration, and visually significant age-related macular degeneration by quintiles of anthocyanin subclasses

	Quintiles of anthocyanin subclass intake					Ptrend2
	1st	2nd	3rd	4th	5th
Cyanidin intake (mg/d)
Cataract	1.00 (reference)	0.82 (0.72, 0.94)	0.87 (0.76, 0.99)	0.81 (0.71, 0.93)	0.86 (0.74, 0.99)	0.17
Total AMD	1.00 (reference)	0.96 (0.73, 1.25)	0.99 (0.75, 1.30)	0.97 (0.74, 1.29)	0.96 (0.72, 1.28)	0.85
Visually significant AMD	1.00 (reference)	1.01 (0.67, 1.53)	0.93 (0.60, 1.44)	0.96 (0.62, 1.48)	1.11 (0.72, 1.72)	0.61
Delphinidin intake (mg/d)
Cataract	1.00 (reference)	0.92 (0.81, 1.05)	1.01 (0.89, 1.14)	0.82 (0.72, 0.94)	0.89 (0.78, 1.01)	0.039
Total AMD	1.00 (reference)	0.99 (0.76, 1.30)	1.36 (1.06, 1.75)	0.93 (0.70, 1.23)	0.94 (0.71, 1.23)	0.10
Visually significant AMD	1.00 (reference)	0.75 (0.48, 1.15)	1.41 (0.98, 2.03)	0.69 (0.44, 1.07)	0.92 (0.61, 1.37)	0.31
Malvidin intake (mg/d)
Cataract	1.00 (reference)	0.98 (0.86, 1.11)	1.01 (0.89, 1.15)	0.90 (0.79, 1.03)	0.88 (0.77, 1.00)	0.02
Total AMD	1.00 (reference)	1.10 (0.85, 1.44)	1.33 (1.03, 1.73)	1.17 (0.89, 1.54)	0.97 (0.73, 1.27)	0.22
Visually significant AMD	1.00 (reference)	1.05 (0.70, 1.57)	1.42 (0.96, 2.09)	0.98 (0.64, 1.51)	0.97 (0.64, 1.47)	0.38
Pelargonidin intake (mg/d)
Cataract	1.00 (reference)	1.05 (0.92, 1.20)	1.03 (0.90, 1.17)	0.98 (0.85, 1.11)	0.97 (0.85, 1.11)	0.40
Total AMD	1.00 (reference)	1.23 (0.94, 1.62)	1.12 (0.85, 1.48)	1.21 (0.93, 1.58)	1.06 (0.81, 1.39)	0.95
Visually significant AMD	1.00 (reference)	1.34 (0.87, 2.06)	1.31 (0.86, 2.01)	1.17 (0.76, 1.79)	1.35 (0.89, 2.03)	0.30
Petunidin intake (mg/d)
Cataract	1.00 (reference)	0.94 (0.83, 1.07)	1.06 (0.93, 1.21)	0.86 (0.74, 0.98)	0.89 (0.78, 1.02)	0.018
Total AMD	1.00 (reference)	1.11 (0.85, 1.44)	1.23 (0.94, 1.60)	1.07 (0.80, 1.42)	0.91 (0.69, 1.21)	0.10
Visually significant AMD	1.00 (reference)	1.12 (0.75, 1.67)	1.25 (0.84, 1.86)	0.85 (0.54, 1.33)	0.96 (0.63, 1.46)	0.30
Peonidin intake (mg/d)
Cataract	1.00 (reference)	0.90 (0.79, 1.03)	0.99 (0.87, 1.12)	0.86 (0.75, 0.98)	0.86 (0.75, 0.98)	0.021
Total AMD	1.00 (reference)	0.84 (0.64, 1.10)	1.19 (0.92, 1.53)	0.98 (0.75, 1.28)	0.82 (0.63, 1.08)	0.12
Visually significant AMD	1.00 (reference)	1.00 (0.67, 1.49)	1.25 (0.84, 1.85)	0.91 (0.60, 1.39)	0.89 (0.59, 1.35)	0.26

Abbreviation: AMD, age-related macular degeneration.

¹Adjusted for age, randomized treatments, total energy intake, BMI, smoking status, exercise, history of eye examination within the last 2 y, history of hypertension, history of diabetes, history of hypercholesterolemia, postmenopausal status, pos-menopausal hormone use, alcohol, fruit and vegetable intake, dairy intake, folate intake, saturated fat intake, and fiber intake.

²Trend test for linearity across quintiles of intake for each subclass of anthocyanins, using median values for each quintile.

Discussion

In this cohort of middle-aged and older women, we found that even modest levels of blueberry intake (≥1 serving/wk) were significantly associated with a reduced risk of incident total AMD. There was a similar association for visually significant AMD that was only manifested in a significant linear trend across categories due to lower case counts. Blueberry intake was not associated with incident cataract. Anthocyanins, a key bioactive component of blueberries that may underlie its beneficial effects on eye health and other outcomes, did not appear to mediate the association between blueberries and AMD. Total anthocyanin intake and its major subclasses were not independently associated with either total or visually significant AMD. On the other hand, greater intake of anthocyanins and selected subgroups of anthocyanins such as delphinidin, malvidin, petunidin, and peonidin, particularly in the upper 2 quintiles of intake, were modestly and significantly associated with lower risk of developing cataracts but not AMD.

To our knowledge, this is the first epidemiologic study that specifically examined both blueberry and anthocyanin intake in the primary prevention of eye disease. Blueberries are a rich source of fiber, vitamin C, and other key nutrients as part of a healthy dietary pattern suggested to have a possible role in the prevention of cataract [18,19] and AMD [[20], [21], [22]]. Previously in the WHS, greater intake of fruits and vegetables [23], lutein/zeaxanthin [23], and vitamin E [24] have been associated with modest, significant reductions in risk of cataract. In 77,562 women from the Nurses’ Health Study and 40,866 men from the Health Professionals Follow-up Study, those consuming 3 compared to 1.5 servings/d of fruit had a significant 36% reduction in the risk of AMD, with similar associations in men and women [25]. Gopinath et al. [26] recently reported an inverse prospective association between dietary flavonoids and AMD in the Blue Mountains Eye Study but did not include data on anthocyanins. In the Age-Related Eye Disease Study, a formulation of supplemental antioxidants (vitamins C and E and β-carotene) plus zinc significantly reduced odds of advanced AMD in a higher-risk population [27]. The addition of lutein plus zeaxanthin (replacing β-carotene) to the formulation in the Age-Related Eye Disease Study 2 trial did not further reduce the risk of progression to advanced AMD in primary analyses; however, exploratory subgroup analyses demonstrated protective effects when limited to participants in the lowest quintile of dietary lutein and zeaxanthin intake [28]. Further, smaller randomized controlled trials of lutein and zeaxanthin supplementation suggest improved macular pigment and visual function in early AMD patients [[29], [30], [31], [32], [33]].

Various polyphenols in blueberries have antioxidant activity that may improve eye health [34,35], with anthocyanins favorably impacting vascular and inflammatory mechanisms. In the Framingham Offspring Study, higher anthocyanin intake in 2375 adults was associated with a lower inflammation score based on 12 biomarkers, suggesting an anti-inflammatory mechanism [36] consistent with that reported in the TwinsUK registry [37]. Among animal studies, blueberry anthocyanins protected retinal cells of the rat from diabetes-induced oxidative stress and inflammation, which are regulated through nuclear factor erythroid 2-related factor 2 and Heme oxygenase-1 signaling [8]. In addition, a blueberry-enriched diet tested in a rat model of light-induced retinopathy showed significant levels of protection of the outer nuclear layer of the retina, highlighting the neuroprotective potential of anthocyanins via their interaction with rhodopsin and phototransduction and their antioxidative capacity [7]. In vitro studies have reported that blueberry anthocyanins protect RPE cells by protecting against light-induced damage and hydrogen peroxide–induced oxidative injuries, as well as downregulation of vascular endothelial growth factor [[38], [39], [40]]. Pterostilbene, another blueberry bioactive, may also exert beneficial effects on eye health by protecting the human cornea from hyperosmolarity-induced inflammation and oxidative stress [41]. However, based on the current body of evidence, there is no clear explanation for why we observed effects from blueberry intake on AMD but not cataract, and similarly, from anthocyanin intake on cataract but not AMD. Further mechanistic investigations are needed to elucidate these findings.

Human intervention studies of blueberries or anthocyanins on eye health are especially sparse. One trial of 72 adults tested 271 mg and 7.11 mg cyanidin 3-glucoside equivalents compared with placebo during a 3-wk intervention and 3-wk washout period, after which neither dark adaptation nor night vision improved [10]. A follow-up trial tested 346 mg cyanidin over a 12-wk intervention and an 8-wk washout with a similar lack of effect. However, in both trials, anthocyanin consumption improved visual acuity recovery after photobleaching [10]. Careful investigations are needed on the short- and long-term effects of blueberries or anthocyanins on eye outcomes to identify the most clinically relevant biomarkers and assessments related to the development of AMD and cataract, particularly for an aging population in whom the prevalence of visual impairment and eye disease has risen precipitously [42].

This large-scale, long-term prospective study examined blueberry and anthocyanin intake using a validated SFFQ and a comprehensive array of risk factors in more than 37,000 female health professionals at risk of developing cataract and AMD confirmed by medical records. However, important methodological limitations temper our results. First, misclassification based upon a single baseline self-report of dietary factors may have biased our results, as multiple longitudinal dietary assessments tend to reduce measurement error and strengthen the magnitude of effect [43]. A second dietary assessment was completed in the WHS, but after the end of outcome ascertainment for cataract and AMD, which suggested that blueberry intake modestly increased over time (data not shown). Second, blueberry and anthocyanin intake may be susceptible to residual confounding by other uncontrolled clinical, lifestyle, or dietary factors associated with eye disease. For example, in a recent systematic review, the Mediterranean diet was associated with a reduced risk of AMD [44] along with its individual components such as fish [45] and lutein and zeaxanthin [23]. Additionally, we did not assess beverage phytochemicals in our SFFQ. However, when we comprehensively controlled for potential confounding by known risk factors, including other components of the Mediterranean diet, our HRs were only modestly impacted. We therefore expect residual confounding would be modest at best.

Third, it remains unclear how much blueberry or anthocyanin intake may lead to meaningful reductions in eye disease. Potential threshold effects differed by anthocyanin subtype. Fourth, we did not consider dose-response associations between blueberry intake and AMD since too few women in WHS consumed ≥2 servings/wk of blueberries, and we had limited numbers of AMD cases in the upper categories of intake. As a result, our relatively narrow comparisons of “high” compared with low blueberry intake may have limited the observed magnitude of risk reduction. Finally, generalizability may be of concern in this study limited to predominantly White female health professionals. However, we have no evidence suggesting the underlying biologic mechanisms may differ by sex, race/ethnicity, or other factors, though additional studies should evaluate these issues.

We found novel evidence for a potential role of blueberries and anthocyanins in the prevention of eye disease in a large cohort of middle-aged and older women. Though blueberry intake was modest, women consuming ≥1 serving/wk of blueberries had a significant 28% reduction in total AMD. The magnitude of effect for visually significant AMD was not statistically significant, but similar to total AMD. and comparable reductions in visually significant AMD. These associations did not extend to dietary total anthocyanins and anthocyanin subclasses, which were modestly and significantly associated with risk of cataract, but not AMD. This hypothesis warrants replication in other studies along with the identification of clinically relevant mechanisms of effect through short-term trials.

Author contributions

The authors’ responsibilities were as follows—HDS: had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis; HDS: conceptualized and designed the study; EK: performed statistical analysis; HDS, SR, SJP, EK, I-ML, RJG, JEB, WGC: acquired, analyzed, or interpreted the data; HDS: drafted the manuscript; SR, SJP, EK, I-ML, RJG, JEB, WGC: provided critical edits to the manuscript; SR, EK, I-ML, RJG, JEB, WGC: provided administrative, technical, or material support; HDS: obtained funding; and all authors: read and approved the final manuscript.

Conflict of interest

HDS received investigator-initiated research funding from the United States Highbush Blueberry Council in support of this manuscript. HDS, I-ML, JEB, and WGC report financial support from National Institutes of Health. All other authors report no conflicts of interest.

Funding

This study was supported by an investigator-initiated grant from the United States Highbush Blueberry Council and research grants EY06633, EY18820, CA047988, HL043851, HL080467, HL099355, and CA182913 from the National Institutes of Health, Bethesda, MD, United States. The funding organizations had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; or preparation, review, or approval of this manuscript.

Data availability

Data described in the manuscript, code book, and analytic code will be made available upon request pending proposal approval.

Appendix A. Supplementary data

The following is the Supplementary data to this article: