- Email: [email protected]
Dietary Patterns and Their Associations with Age-Related Macular Degeneration
Dietary Patterns and Their Associations with Age-Related Macular Degeneration The Melbourne Collaborative Cohort Study Fakir M. Amirul Islam, PhD,1,* Elaine W. Chong, MBBS, PhD,1,* Allison M. Hodge, PhD,2 Robyn H. Guymer, MBBS, PhD,1 Khin Zaw Aung, MBBS,1 Galina A. Makeyeva, MBBS, PhD,1 Paul N. Baird, PhD,1 John L. Hopper, PhD,2,3 Dallas R. English, PhD,2,3 Graham G. Giles, PhD,2,3 Liubov D. Robman, MBBS, PhD1 Objective: To evaluate the association between dietary patterns and age-related macular degeneration (AMD). Design: Food frequency data were collected from Melbourne Collaborative Cohort Study (MCCS) participants at the baseline study in 1990e1994. During follow-up in 2003e2007, retinal photographs were taken and evaluated for AMD. Participants: At baseline, 41 514 participants aged 40 to 70 years and born in Australia or New Zealand (69%), or who had migrated from the United Kingdom, Italy, Greece, or Malta (31%) were recruited. Of these, 21 132 were assessed for AMD prevalence at follow-up. Methods: Principal component analysis was used to identify dietary patterns (Factors F1e6) among the food items. Logistic regression was used to assess associations of dietary patterns with AMD. Main Outcome Measures: Odds ratios (ORs) for early stages and advanced AMD in association with dietary patterns. Results: A total of 2508 participants (12.8%) had early stages of AMD, and 108 participants (0.6%) had advanced AMD. Six factors characterized by predominant intakes of fruits (F1); vegetables (F2); grains, fish, steamed or boiled chicken, vegetables, and nuts (F3); red meat (F4); processed foods comprising cakes, sweet biscuits, and desserts (F5); and salad (F6) were identified. Higher F3 scores were associated with a lower prevalence of advanced AMD (fourth vs. first quartile) (OR, 0.49; 95% confidence interval [CI], 0.28e0.87), whereas F4 scores greater than the median were associated with a higher prevalence of advanced AMD (OR, 1.46; 95% CI, 1.0e2.17). Conclusions: Rather than specific individual food items, these factors represent a broader picture of food consumption. A dietary pattern high in fruits, vegetables, chicken, and nuts and a pattern low in red meat seems to be associated with a lower prevalence of advanced AMD. No particular food pattern seemed to be associated with the prevalence of the earliest stages of AMD. Ophthalmology 2014;-:1e7 ª 2014 by the American Academy of Ophthalmology.
Age-related macular degeneration (AMD) is a complex genetic disease, and the major genetic components are now well understood.1 However despite this, diet is emerging as a potentially important modifiable risk factor for AMD.2e9 To date, results from dietary studies have largely been based on analyses evaluating individual foods or food groups. The results suggest that some aspects of diet could influence the risk of AMD, but the associations found have not been consistent across studies.4,10e16 Diets high in trans fat, red meat, and alcohol have been associated with an increased risk of AMD,3e5 whereas higher intakes of fish have been associated with a lower risk of AMD.16,17 In general, foods are not consumed in isolation, and intakes of many nutrients are highly correlated to common food sources. Therefore, an integrated approach to the 2014 by the American Academy of Ophthalmology Published by Elsevier Inc.
investigation of dietary impact on chronic disease where dietary patterns are assessed, rather than each individual food or nutrient, may be more informative and predictive of disease risk.18e20 Several studies have evaluated dietary combinations and AMD using predefined food groups.13,21,22 However, to date, no study has evaluated the associations of dietary patterns derived from principal components analysis (PCA) and AMD. Principal components analysis is a data-driven statistical method used to reduce a large number of intercorrelated variables (e.g., dietary items) into a few distinct factors of intercorrelated variables within each factor that are dissimilar between factors.23 In this way, foods are grouped according to how they were actually consumed in a cohort, rather than by preconceived ideas of how foods should be consumed. ISSN 0161-6420/14/$ - see front matter http://dx.doi.org/10.1016/j.ophtha.2014.01.002
1
Ophthalmology Volume -, Number -, Month 2014 These groups of foods are aggregated together into patterns or factors, which are then evaluated for associations with AMD. Associations of dietary patterns with AMD may provide more insight into how diet is associated with the risk of developing AMD than concentrating on individual food items,24 providing information on which health messages can be based. We used PCA to identify dietary patterns among Melbourne Collaborative Cohort Study (MCCS) participants and assessed the associations of these patterns with AMD.
Methods Study Population The MCCS recruited 41 514 participants (24 469 women) aged 40 to 69 years at baseline (1990e1994) to examine links between diet and chronic diseases.25,26 To broaden the range of dietary intakes, participants born in Greece and Italy were deliberately oversampled to comprise one quarter of the cohort, whereas participants born in Australia, New Zealand, and United Kingdom made up the remainder of the cohort.25 Of the total sample, 27 883 (67%) attended a follow-up examination between 2003 and 2007; 3754 (9%) had died, and 819 (2%) had left Australia or Victoria. The remaining 9058 participants (22%) were lost to follow-up or not interested in participating. Of those who were alive and living in Victoria, the participation rate was 27 883/36 941 (73%), of whom 22 405 (80%) participated in fundus photography. The Human Research Ethics Committees of the Cancer Council Victoria and Royal Victorian Eye and Ear Hospital approved the AMD study protocol. The research adhered to the tenets of the Declaration of Helsinki.
Dietary Assessment Dietary data were collected at baseline using a self-administered 121-item food frequency questionnaire (FFQ) specifically developed for this study.27 Most items were analyzed as daily equivalent frequencies, whereas intakes of olive and vegetable oils were analyzed in milliliters/week. The repeatability of the FFQ after 12 months showed intraclass correlation coefficients >0.50 for most food items.28 Energy intake was calculated using sexspecific standard portions, together with Australian food composition data.29 Participants with estimated energy intakes in the upper and lower 1% of the sex-specific distributions, which suggested that the FFQ had been improperly completed, were excluded. Almost all dietary studies use criteria related to energy intake to exclude people with reported intakes that are unlikely to be real.30 Because we recognize the limitations of FFQs in measuring absolute amounts, people at the extreme 1% of high or low energy intakes were excluded. People with these energy extremes are unlikely to have reported a real usual intake, so including their data might distort the eating patterns identified. Participants with a history of heart attack, angina, or diabetes at baseline also were excluded from our analysis because these diagnoses had the potential to result in dietary change, and we have evidence from this cohort that this was the case (unpublished data, MCCS internal validation, December 2005).
Age-related Macular Degeneration Detection At follow-up from 2003 to 2007 (on average, 13 years after the baseline assessment), digital nonstereoscopic 45 macular photographs of both eyes were taken with a Canon CR6-45NM nonmydriatic retinal camera (Canon Inc, Tokyo, Japan). Retinal
2
photographs were graded for AMD by 2 experienced graders using “OptoLite/OptoMize Pro” software (Digital HealthCare Image Management Systems, Cambridge, UK), according to the International Classification System for grading of AMD.31 Qualitycontrol procedures for the photographs in the MCCS have been described.32
Definitions of Age-related Macular Degeneration Participants were allocated to a single AMD category according to the more advanced visible changes on the macular images of the worse affected eye. Early stages of AMD were defined as the presence of 1 or more drusen 125 mm (with or without pigmentary abnormalities) or 1 or more drusen 63 to 124 mm with pigmentary abnormalities in a 6000-mm diameter grading grid centered on the fovea, in the absence of advanced AMD in either eye (geographic atrophy or neovascular AMD).
Covariates Age, sex, country of origin, smoking status, education level, total energy intake, multivitamin supplement use, body mass index, waistehip ratio, physical activity, and alcohol consumption were included in logistic regression models and retained if they changed the beta-coefficient for any of the dietary factors by more than 5%.
Statistical Analysis The PCA23 of 121 food items plus olive and vegetable oils was performed to extract dietary factors using dietary data collected at baseline. This was followed by orthogonal (varimax) rotation to ensure that the factors were uncorrelated. Factor analysis aims to evaluate which foods are correlated, looking for underlying factors within diets of the population.19 It is used as a way of reducing many variables to few and has been recently used in dietary epidemiology.24 This means that a factor has a distinct cluster of interrelated items with similarities within, but minimal similarities between the factors. Factors with eigenvalues >2.7 were considered to reflect realistic eating patterns and were used in the analysis. Food items with an absolute value of at least 0.2 for the factor loading (the correlation between dietary items and factors) were retained to define the dietary Factors 1 to 6, which is a commonly used cutoff.20 The cutoff used does not affect the calculation of the factor score. For each participant, the factor score is computed as the sum of products of the observed intake frequency values multiplied by their factor loading. Factor scores were then analyzed by quartile and median groupings. We used binary logistic regression models to estimate the odds ratio (OR) and 95% confidence intervals (CIs) for both early-stage and advanced AMD, adjusting for covariates. In the analyses for advanced AMD, early-stage AMD cases were excluded and vice versa. Factor scores were divided into groups, cut at the median or at quartiles for analysis, with the first quartile as the reference group. Tests for trend across quartiles were then performed using the medians in each group. Interactions among sex, age group, grouped country of birth (participants born in Australia and the United Kingdom compared with Southern European participants), and factor scores were tested before performing the analysis. A subanalysis was performed investigating the associations with country of birth (participants born in Australia and the United Kingdom compared with Southern European participants) and food factors for early-stage and advanced AMD. All analyses were performed using the Statistical Package for the Social Sciences version 19.0.1 (SPSS Inc, Chicago, IL).
Amirul Islam et al
Associations of Dietary Patterns with AMD
Results Of the 22 405 participants who had retinal photographs assessed, 21 132 (94%) had retinal photographs gradable for AMD. After excluding 1364 participants at the extreme 1% of energy intake at baseline (n ¼ 311) and participants with a history of heart attack/ angina/stroke (n ¼ 688) or diabetes at baseline (n ¼ 365) (with the potential for dietary modification), we included the remaining 19 768 participants in the analysis. Of the 19 768 participants, 2508 (12.7%) had early stages of AMD and 108 (0.55%) had advanced AMD. Compared with the MCCS participants who were not included in this analysis, those included tended to be younger (mean age at baseline, 53 vs. 56 years), female (61% vs. 59%), and more educated (high school or higher level of education 91% vs. 72%), and were more likely to have never smoked (61% vs. 56%) and less likely to have been born in Greece (4.1% vs. 17.4%) or Italy (8.3% vs. 17.1%), all with P values <0.01. The prevalence of early AMD was 12.7% in participants included in the study and 13.8% in 1364 participants who were excluded (P ¼ 0.28). The prevalence of late AMD was 0.6% in included participants and 1.2% in excluded participants (P ¼ 0.02). We identified 6 factors with minimum eigenvalues of 2.7, which accounted for 33% of the total variance that could be explained by 120 factors using PCA. Factor loadings for food items with loadings having absolute values of 0.2 or greater for any factor, sorted in descending order, are shown in Table 1 (available at http:// aaojournal.org). The factor loadings are the correlation coefficients for each variable on that factor, so that variables with higher factor loadings are those that tend to define the factor. Factor 1, “fruit,” was characterized by intakes of a variety of fruits, including stone, citrus, and other fruits and olives. Factor 2, “vegetables,” was characterized by intakes of a variety of vegetables, including pumpkin, cauliflower, green beans, and potato and an avoidance of pasta, pizza, and salami. Factor 3, “grains and fish,” was characterized by more frequent intakes of boiled rice; muesli; steamed, grilled, and canned fish; steamed or boiled chicken; certain vegetables, including mushrooms, zucchini, spinach, and broccoli; and nuts; and avoidance of white bread. Factor 4, “red meat,” was characterized by intake of red meat (beef, lamb), fried and smoked fish, eggs, and avoidance of whole wheat or rye bread. Factor 5, termed “processed foods,” was characterized by intake of cakes, sweet biscuits, confectionery, ice cream, cheddar or similar cheeses, margarine, pudding, eggs, sausages, bacon, and ham. The last Factor, “salad,” was characterized mainly by intake of uncooked salad (lettuce, cucumber, tomato) and legumes. We have abbreviated the factors with a descriptive term based on the predominant items that load onto the factors. The MCCS participants included in these analyses had lower mean scores for “fruit” (0.042 vs. 0.047), “red meat” (0.116 vs. 0.131), and “salad” (0.029 vs. 0.033), but higher scores for “vegetables” (0.015 vs. 0.018), “grains and fish” (0.117 vs. 0.131), and “processed foods” (0.155 vs. 0.173) than those who were not included, all with P values <0.01. The associations of dietary factor scores with early stages and advanced AMD are shown in Table 2. The final adjusted model included age, sex, country of origin, smoking, education, multivitamin use, and total energy intake as covariates because they were established risk factors of AMD or a potential effect modifier such as sex, or they changed the beta-coefficient for any of the dietary factors by more than 5%. Body mass index, waiste hip ratio, physical activity, and alcohol consumption did not change the associations and were not retained in the analyses. We found Factor 3, “grains and fish,” associated with a 51% decreased odds of advanced AMD after adjustment for covariates, fourth versus first quartile (OR, 0.49; 95% CI, 0.28e0.87). We also found Factor 4,
“red meat,” associated with a higher prevalence of advanced AMD with increasing ORs across quartiles 1 to 3, third versus first quartile (OR, 2.29; 95% CI, 1.37e3.84). However, the odds were not increased for the fourth quartile of the “red meat” factor. A value above or equal to the median score for the “red meat” factor was associated with a higher prevalence of advanced AMD (OR, 1.46; 95% CI, 1.00e2.17). Factor 1, “fruit,” and Factor 2, “vegetables,” scores showed no significant associations with early or advanced AMD, although adjusted ORs trended downward across quartiles for advanced AMD. Factor 5, “processed food,” and Factor 6, “salad,” showed no associations with advanced AMD. None of the dietary patterns was associated with the early stages of AMD. Given the design of the study, we wanted to determine whether there was a difference in relationships between food patterns and AMD in people grouped by the country of birth: Australia/New Zealand/United Kingdom versus Italy or Greece. In this subanalysis, we found that an association of Factor 3 above the median with lower odds of advanced AMD was similar in both groups, with an OR of 0.63 (95% CI, 0.41e0.97) for those born in Australia/New Zealand/United Kingdom and an OR of 0.12 (95% CI, 0.02e0.95) for those born in Italy or Greece, adjusted for covariates. Factor 4 above the median was associated with a higher prevalence of advanced AMD, with an OR of 1.46 (95% CI, 1.0e2.21) for those born in Australia/New Zealand/United Kingdom and a similar trend for those born in Italy or Greece, although the association in this group was not significant (OR, 1.94; 95% CI, 0.55e6.84).
Discussion The MCCS presents a unique opportunity to study links between diets and those with AMD, because it is the largest single cohort study with comprehensive data on both diet and AMD status. With the use of PCA, we identified 6 dietary factors (patterns) at baseline. Of note, fruit and vegetables ended up in different factors in our data. We observed that participants born in Australia, New Zealand, and the United Kingdom scored higher on the “vegetable” factor, whereas those born in Italy and Greece scored higher on the “fruit” factor, suggesting that in our study people who ate fruit did not always eat vegetables at a similar frequency and vice versa. We observed that a high score for Factor 3, “grains and fish,” was associated with a lower prevalence of advanced AMD, whereas Factor 4, “red meat,” score was positively associated with the prevalence of advanced AMD, adjusted for covariates. To our knowledge, no other study assessing AMD risk has evaluated dietary patterns derived from datadriven PCA. The European Eye Study reported on an association of posteriori-derived dietary patterns with retinal vessel caliber in an elderly population, but to date it has not reported on any AMD association.33 However, dietary patterns and their associations with other chronic diseases, such as cancer and cardiovascular disease,24,34,35 have been reported. In the MCCS, a dietary pattern characterized by meats and fatty foods was associated with increased diabetes risk, which supports the notion that patterns of food intake are useful to study in relation to risks of chronic disease.36 Previous studies have evaluated predefined nutrient combinations and AMD risk. These studies used presumptive cutoff levels rather than evaluating the actual food intake in their cohorts. A retrospective case-control study evaluating overall diet quality on the basis of the Alternate Healthy
3
Ophthalmology Volume -, Number -, Month 2014 Table 2. Associations between Dietary Factors and Age-related Macular Degeneration* Reference Group Factor Scores Factor 1: Fruit First quartile, <0.59 Second quartile, 0.59 to 0.26 Third quartile, 0.26 to 0.28 Fourth quartile, >0.28 P for trend Mediany Factor 2: Vegetables First quartile, <0.63 Second quartile, 0.63 to 0.0.6 Third quartile, 0.06 to 0.57 Fourth quartile, >0.57 P for trend Mediany Factor 3: Grains and Fish First quartile, <0.69 Second quartile, 0.69 to 0.15 Third quartile, 0.15 to 0.55 Fourth quartile, >0.55 P for trend Mediany Factor 4: Red Meat First quartile, <0.63 Second quartile, 0.63 to 0.18 Third quartile, 0.18 to 0.40 Fourth quartile, >0.40 P for trend Mediany Factor 5: Processed Foods First quartile, <0.69 Second quartile, 0.69 to 0.08 Third quartile, 0.08 to 0.58 Fourth quartile, >0.58 P for trend Mediany Factor 6: Salad First quartile, <0.65 Second quartile, 0.65 to 0.14 Third quartile, 0.14 to 0.45 Fourth quartile, >0.45 P for trend Mediany
Early Stages of AMD
n
n
%
OR
4443 4454 4315 3937
630 640 653 585
12.4 12.6 13.1 12.9
8252
1238
13.0
1.00 0.93 0.96 0.93 0.40 0.98
3936 4443 4599 4171
573 583 660 692
12.7 11.6 12.5 14.2
8770
1352
13.4
3589 3961 4559 5040
547 603 671 687
13.2 13.2 12.8 12.0
9599
1358
12.4
4708 4553 4236 3652
709 658 631 510
13.1 12.6 13.0 12.3
7888
1141
12.6
2947 4394 4899 4909
470 647 637 754
13.8 12.8 11.5 13.3
9808
1391
12.4
4175 4465 4479 4030
606 662 667 573
12.7 12.9 13.0 12.4
8509
1240
12.7
1.00 0.90 0.92 0.97 0.77 0.99 1.00 1.04 1.07 1.04 0.60 1.03 1.00 0.98 1.02 0.95 0.54 1.0 1.00 1.05 0.95 1.08 0.41 0.97 1.00 1.06 1.07 0.99 0.84 1.0
Advanced AMD 95% CI
n
%
OR
0.82e1.05 0.85e1.08 0.82e1.05
29 21 29 29
0.6 0.5 0.7 0.7
0.90e1.07
58
0.7
1.00 0.49 0.69 0.81 0.97 1.08
0.80e1.03 0.81e1.05 0.85e1.11
20 17 25 46
0.5 0.4 0.5 1.1
0.91e1.09
71
0.8
0.92e1.18 0.94e1.21 0.92e1.18
43 28 18 19
1.2 0.7 0.4 0.4
0.94e1.13
37
0.4
0.87e1.10 0.91e1.15 0.84e1.08
25 29 41 13
0.5 0.6 1.0 0.4
0.92e1.10
54
0.7
0.92e1.21 0.83e1.09 0.94e1.25
25 25 26 32
0.8 0.6 0.5 0.6
0.87e1.07
58
0.6
0.94e1.20 0.95e1.21 0.87e1.12
31 23 27 27
0.7 0.5 0.6 0.7
0.92e1.09
54
0.6
1.00 0.57 0.57 0.84 0.20 1.02 1.00 0.67 0.45 0.49 0.008 0.57 1.00 1.41 2.29 0.97 0.41 1.46 1.00 0.79 0.74 0.76 0.34 0.87 1.00 0.82 1.03 1.10 0.71 1.16
95% CI
0.28e0.88 0.41e1.18 0.47e1.40 0.73e1.59
0.28e1.15 0.29e1.13 0.45e1.59 0.65e1.59
0.41e1.10 0.26e0.80 0.28e0.87 0.37e0.86
0.82e2.44 1.37e3.84 0.49e1.95 1.00e2.17
0.44e1.42 0.41e1.36 0.42e1.36 0.58e1.31
0.47e1.42 0.61e1.74 0.65e1.88 0.79e1.72
AMD ¼ age-related macular degeneration; CI ¼ confidence interval; OR ¼ odds ratio. Statistically significant findings (P < 0.05) are shown in boldface. *The ORs and 95% CIs, adjusted for age, sex, country of origin, smoking status (never, current, and past), education level (below high school level vs. equal or more than high school), multivitamin supplement use, and total energy intake. y The ORs (95% CIs) for equal to or above median factor scores compared with below median scores.
Eating Index, which derives scores according to high/low intakes in each of 5 food groups (grains, vegetables, fruits, nuts and soy, and ratio of white to red meat) and fat intake, reported that people in the top quartile of diet quality had half the risk of advanced AMD (OR, 0.54; 95% CI, 0.30e0.99) than those in the bottom quartile.21 Although the dietary habits in this study were reported after AMD diagnosis, the results suggested that high grain intake along with white meat may be protective for advanced AMD. The second study was the Carotenoids in Age-Related Eye Disease Study of 1313 women aged 50 to 79 years,37 which used a 2005 modified Healthy Eating Index
4
(mHEI)13 based on FFQ data.38 The mHEI represented a healthy diet that included abundant intakes of fruits, vegetables, and whole grains, and a low intake of sugar, fat, saturated fat, and alcohol.13 They reported 46% lower odds for early AMD in the highest compared with the lowest quintile of mHEI (OR, 0.54; 95% CI, 0.33e0.88); because there were only 12 cases of advanced AMD, they were excluded from the main analyses.13 Previous studies evaluating single nutritional items have shown inverse associations between AMD and intakes of fish,2,10 fruits, vegetables,39e41 bread, cereals, and foods with a low glycemic index;42,43 whereas others, including
Amirul Islam et al
Associations of Dietary Patterns with AMD
the MCCS, have reported positive associations between red or processed meat intake and AMD.3,44,45 These findings are along similar lines to our food pattern analyses. When we performed subanalyses based on grouped country of birth, we found that the associations between lower prevalence of advanced AMD and Factor 3 (grains and fish) were similar in those born in Australia, New Zealand, the United Kingdom, and Greece. The associations between a higher prevalence of advanced AMD and Factor 4 (red meat) were in similar directions in both groups, but significant only in those born in Australia, New Zealand, and the United Kingdom. The latter may have been due to the smaller numbers available in this subgroup. We did not find any significant associations between dietary patterns and early AMD. This may be due to an issue of diet having less of a relationship in primary prevention (prevention of development of early AMD) compared with secondary prevention of AMD (prevention of advanced AMD from early AMD).7 It is possible that diet does not influence early AMD to the same extent as driving the disease to the advanced stages. This lack of association may also be due to other unadjusted variables, such as genetic data (effect modification due to genotype) or other unknown confounders.12 The early AMD phenotype is diverse; additional genotype and phenotype characterizations, with the use of other modalities, like autofluorescence, may help clarify associations with diet in the future. One advantage of our factor analysis is that foods were not grouped before analysis. Thus, we avoided making a priori assumptions about what should be grouped. For example, in the Carotenoids in Age-Related Eye Disease Study, meat, fish, and beans were grouped together, but from our data, steamed fish were loaded on a different factor than fried fish and red meat, where they had opposite associations with advanced AMD. Other strengths of our study include its large sample size, detailed AMD grading of high-resolution images with the prevalence of AMD closely matching other epidemiologic studies,46,47 and the broad range of dietary exposures as a result of targeting different ethnicities for recruitment and also the assessment of dietary intakes a decade before AMD detection, (before dietary modifications and nutrient supplements for AMD were commonly recommended).
Study Limitations The study limitations included a natural high attrition that is not unexpected over a long 10- to 17-year follow-up period in an aging cohort, which may lead to survival bias. Diagnosis of AMD based on color fundus photographs was the only feasible method of studying more than 22 000 people at the time when the study was carried out, as this was before the widespread use of other modalities. The lack of optical coherence tomography or fluorescein angiography data would most likely have led to an underdiagnosis of AMD. We did not obtain data on family history of AMD, and the lack of genotyping on all MCCS participants did not allow us to investigate the potential genetic and dietary interactions in AMD. We were also unable to adjust for potential Age-related Eye Disease Study (AREDS)
supplementation use because the dietary data were collected in 1990, 11 years before the AREDS report was published in 2001. Because the specific AREDS formulation was not commercially available in Australia during ophthalmic data collection and widespread use of supplementation for AMD was not common practice at that time, it is unlikely that this would have affected our results. There was also a lower rate of follow-up among Italian and Greek participants, which could introduce participation bias, but analyses for different subgroups did not change our results. Lastly, our measurement of diet was based on a single FFQ administered at baseline, and this may not have been representative of lifelong consumption. However, we did exclude people at baseline with previous diagnoses that appeared to be associated with dietary change. Random error in measuring dietary intake also is likely to have attenuated our associations. In conclusion, our results show that a diet characterized by frequent consumption of boiled rice, muesli, fish (not fried), chicken (not fried), and a variety of vegetables, and avoidance of white bread was associated with a lower prevalence of advanced AMD, whereas a diet characterized by a pattern of eating red and processed meats and fried foods was associated with a higher prevalence of advanced AMD. The results from the dietary pattern analysis cannot reveal the absolute amounts of certain foods to eat or avoid, but they do give direction as to the patterns of food intakes that are potentially important to consider when trying to reduce the risk of AMD. Thus, despite a well-characterized genetic risk profile for AMD, evidence is continuing to mount that the choices we make about foods we consume may play a role in contributing to the risk of developing AMD.
References 1. Fritsche LG, Chen W, Schu M, et al. Seven new loci associated with age-related macular degeneration. Nat Genet 2013;45:433–9, 9e1e2. 2. Seddon JM, George S, Rosner B. Cigarette smoking, fish consumption, omega-3 fatty acid intake, and associations with age-related macular degeneration: the US Twin Study of AgeRelated Macular Degeneration. Arch Ophthalmol 2006;124: 995–1001. 3. Chong EW, Simpson JA, Robman LD, et al. Red meat and chicken consumption and its association with age-related macular degeneration. Am J Epidemiol 2009;169:867–76. 4. Chong EW, Kreis AJ, Wong TY, et al. Alcohol consumption and the risk of age-related macular degeneration: a systematic review and meta-analysis. Am J Ophthalmol 2008;145: 707–15. 5. Adams MK, Chong EW, Williamson E, et al. 20/20eAlcohol and Age-related Macular Degeneration: The Melbourne Collaborative Cohort Study. Am J Epidemiol 2012;176: 289–98. 6. Chong EW, Robman LD, Simpson JA, et al. Fat consumption and its association with age-related macular degeneration. Arch Ophthalmol 2009;127:674–80. 7. Chong EW, Wong TY, Kreis AJ, et al. Dietary antioxidants and primary prevention of age related macular degeneration: systematic review and meta-analysis. BMJ 2007;335:755.
5
Ophthalmology Volume -, Number -, Month 2014 8. Lutein þ zeaxanthin and omega-3 fatty acids for age-related macular degeneration: the Age-Related Eye Disease Study 2 (AREDS2) randomized clinical trial. JAMA 2013;309:2005–15. 9. Seddon JM, Rosner B, Sperduto RD, et al. Dietary fat and risk for advanced age-related macular degeneration. Arch Ophthalmol 2001;119:1191–9. 10. Chong EW, Kreis AJ, Wong TY, et al. Dietary omega-3 fatty acid and fish intake in the primary prevention of age-related macular degeneration: a systematic review and metaanalysis. Arch Ophthalmol 2008;126:826–33. 11. Smith W, Mitchell P, Leeder SR. Dietary fat and fish intake and age-related maculopathy. Arch Ophthalmol 2000;118:401–4. 12. Wang JJ, Rochtchina E, Smith W, et al. Combined effects of complement factor H genotypes, fish consumption, and inflammatory markers on long-term risk for age-related macular degeneration in a cohort. Am J Epidemiol 2009;169:633–41. 13. Mares JA, Voland RP, Sondel SA, et al. Healthy lifestyles related to subsequent prevalence of age-related macular degeneration. Arch Ophthalmol 2011;129:470–80. 14. Mares-Perlman JA, Brady WE, Klein R, et al. Dietary fat and age-related maculopathy. Arch Ophthalmol 1995;113:743–8. 15. SanGiovanni JP, Chew EY, Clemons TE, et al. The relationship of dietary carotenoid and vitamin A, E, and C intake with age-related macular degeneration in a case-control study: AREDS Report No. 22. Arch Ophthalmol 2007;125:1225–32. 16. Tan JS, Wang JJ, Flood V, Mitchell P. Dietary fatty acids and the 10-year incidence of age-related macular degeneration: the Blue Mountains Eye Study. Arch Ophthalmol 2009;127:656–65. 17. Swenor BK, Bressler S, Caulfield L, West SK. The impact of fish and shellfish consumption on age-related macular degeneration. Ophthalmology 2010;117:2395–401. 18. Hu FB. Dietary pattern analysis: a new direction in nutritional epidemiology. Curr Opin Lipidol 2002;13:3–9. 19. Reedy J, Wirfalt E, Flood A, et al. Comparing 3 dietary pattern methodsecluster analysis, factor analysis, and index analysiseWith colorectal cancer risk: The NIH-AARP Diet and Health Study. Am J Epidemiol 2010;171:479–87. 20. Slattery ML, Edwards SL, Boucher KM, et al. Lifestyle and colon cancer: an assessment of factors associated with risk. Am J Epidemiol 1999;150:869–77. 21. Montgomery MP, Kamel F, Pericak-Vance MA, et al. Overall diet quality and age-related macular degeneration. Ophthalmic Epidemiol 2010;17:58–65. 22. Chiu CJ, Milton RC, Klein R, et al. Dietary compound score and risk of age-related macular degeneration in the age-related eye disease study. Ophthalmology 2009;116:939–46. 23. Jolliffe IT. Introduction: Definition and Derivation of Principal Components Analysis. In: Principal Component Analysis. Springer, NY: Springer Series in Statistics; 2002:1–6. 24. Newby PK, Tucker KL. Empirically derived eating patterns using factor or cluster analysis: a review. Nutr Rev 2004;62:177–203. 25. Giles GG. The Melbourne Study of Diet and Cancer. Proc Nutr Soc Aust 1990;15:61–8. 26. Giles GG, English DR. The Melbourne Collaborative Cohort Study. International Agency for Research on Cancer Science Publication. Lyon, France: International Agency for Research on Cancer; 2002;156:69–70. 27. Ireland P, Jolley D, Giles G, et al. Development of the Melbourne FFQ: a food frequency questionnaire for use in an Australian prospective study involving an ethnically diverse cohort. Asia Pac J Clin Nutr 1994;3:19–31. 28. Hodge AM. Diet Related Predictors of Type 2 Diabetes in the Melbourne Collaborative Cohort Study. PhD Dissertation. Melbourne, Australia: University of Melbourne, Department of Public Health; 2006:65–75.
6
29. Lewis J, Milligan G, Hunt A. NUTTAB95 Nutrient Data Table for Use in Australia. Canberra, Australia: Australian Government Publishing Service; 1995. 30. Willett W. Issues in analysis and presentation of dietary data. In: Willett W. Nutritional Epidemiology. 2nd ed. New York, NY: Oxford University Press; 1998:321–45. 31. Bird AC, Bressler NM, Bressler SB, et al. An international classification and grading system for age-related maculopathy and age-related macular degeneration. The International ARM Epidemiological Study Group. Surv Ophthalmol 1995;39: 367–74. 32. Aung KZ, Robman L, Chong EW, et al. Non-mydriatic digital macular photography: how good is the second eye photograph? Ophthalmic Epidemiol 2009;16:254–61. 33. McEvoy CT, Cardwell CR, Chakravarthy U, et al. A posterioriderived dietary patterns and retinal vessel caliber in an elderly population. Invest Ophthalmol Vis Sci 2013;54:1337–44. 34. Magalhaes B, Peleteiro B, Lunet N. Dietary patterns and colorectal cancer: systematic review and meta-analysis. Eur J Cancer Prev 2012;21:15–23. 35. Maruyama K, Iso H, Date C, et al. Dietary patterns and risk of cardiovascular deaths among middle-aged Japanese: JACC Study. Nutr Metab Cardiovasc Dis 2013;23:519–27. 36. Hodge AM, English DR, O’Dea K, Giles GG. Dietary patterns and diabetes incidence in the Melbourne Collaborative Cohort Study. Am J Epidemiol 2007;165:603–10. 37. Design of the Women’s Health Initiative clinical trial and observational study. The Women’s Health Initiative Study Group. Control Clin Trials 1998;19:61–109. 38. Guenther P, Reedy J, Krebs-Smith S, et al. Development and evaluation of the healthy eating index-2005. Technical report Center for Nutrition Policy and Promotion, United States Department of Agriculture 2007. Available at: http://www. cnpp.usda.gov/healthyeatingindex.htm. Accessed December 20, 2013. 39. Evans J. Antioxidant supplements to prevent or slow down the progression of AMD: a systematic review and meta-analysis. Eye (Lond) 2008;22:751–60. 40. Trevithick J, Massel D, Robertson JM, et al. Model study of AREDS antioxidant supplementation of AMD compared to Visudyne: a dominant strategy? Ophthalmic Epidemiol 2004;11:337–46. 41. A randomized, placebo-controlled, clinical trial of high-dose supplementation with vitamins C and E, beta carotene, and zinc for age-related macular degeneration and vision loss: AREDS report no. 8. Arch Ophthalmol 2001;119:1417–36. 42. Kaushik S, Wang JJ, Flood V, et al. Dietary glycemic index and the risk of age-related macular degeneration. Am J Clin Nutr 2008;88:1104–10. 43. Chiu CJ, Milton RC, Gensler G, Taylor A. Association between dietary glycemic index and age-related macular degeneration in nondiabetic participants in the Age-Related Eye Disease Study. Am J Clin Nutr 2007;86:180–8. 44. Arnarsson A, Sverrisson T, Stefansson E, et al. Risk factors for five-year incident age-related macular degeneration: the Reykjavik Eye Study. Am J Ophthalmol 2006;142:419–28. 45. Cho E, Hung S, Willett WC, et al. Prospective study of dietary fat and the risk of age-related macular degeneration. Am J Clin Nutr 2001;73:209–18. 46. Klein R, Chou CF, Klein BE, et al. Prevalence of age-related macular degeneration in the US population. Arch Ophthalmol 2011;129:75–80. 47. Mitchell P, Smith W, Attebo K, Wang JJ. Prevalence of agerelated maculopathy in Australia. The Blue Mountains Eye Study. Ophthalmology 1995;102:1450–60.
Amirul Islam et al
Associations of Dietary Patterns with AMD
Footnotes and Financial Disclosures Originally received: July 28, 2013. Final revision: December 20, 2013. Accepted: January 2, 2014. Available online: ---.
Manuscript no. 2013-1240.
1
Centre for Eye Research Australia, University of Melbourne, Royal Victorian Eye and Ear Hospital, Melbourne, Victoria, Australia. 2 Cancer Epidemiology Centre, Cancer Council Victoria, Melbourne, Victoria, Australia. 3
Centre for Molecular, Environmental, Genetic and Analytic Epidemiology, School of Public Health, University of Melbourne, Australia. *Drs. Islam and Chong contributed equally to this article as first authors. Presented at: The Association for Research in Vision and Ophthalmology, May 5e9, 2013, Seattle, Washington. Financial Disclosure(s): The author(s) have no proprietary or commercial interest in any materials discussed in this article. Funding: Victoria Health, The Cancer Council Victoria, and National Health & Medical Research Council of Australia (NHMRC) (*Program Grant 209057, Capacity Building Grant 251533, and Enabling Grant
396414) funded the MCCS study. Ophthalmic component was funded by the Ophthalmic Research Institute of Australia and American Health Assistance Foundation. People support was provided through the NHMRC Practitioner Fellowships (R.H.G.), NHMRC Senior Research Fellowships (P.N.B. and J.L.H.), and Macular Degeneration Foundation Blackmores Dr. Paul Beaumont Research Fellowship (L.D.R.). The Centre for Eye Research Australia is a recipient of the NHMRC Centre for Clinical Research Excellence Grant 529923 and Operational Infrastructure Support from the Victorian Government. Abbreviations and Acronyms: AMD ¼ age-related macular degeneration; AREDS ¼ Age-related Eye Disease Study; CI ¼ confidence interval; FFQ ¼ food frequency questionnaire; MCCS ¼ Melbourne Collaborative Cohort Study; mHEI ¼ modified Healthy Eating Index; OR ¼ odds ratio; PCA ¼ principal components analysis. Correspondence: Liubov D. Robman, MBBS, PhD, Centre for Eye Research Australia, 32 Gisborne St., East Melbourne, VIC, Australia, 3002. E-mail: lrobman@ unimelb.edu.au.
7
Ophthalmology Volume -, Number -, Month 2014 Table 1. Factor Loadings from Principle Component Analysis in the Melbourne Collaborative Cohort Study (Food Items with Loadings with Absolute Values of 0.2 or Greater are Shown) Factor 1: Fruit Apricots Peaches or nectarines Plums Grapes Watermelon Pears Honeydew melon Strawberries Oranges or mandarins Figs Apples Pineapple Bananas Olives Grapefruit Fruit salad Pumpkin Cauliflower Cabbage or brussels sprouts Green beans or peas Potatoes, cooked without fat Broccoli Carrots Pasta or noodles Silverbeet, spinach, or other leafy greens Hard grating cheeses Salami or continental sausages Pizza Other breakfast cereals Rice, boiled Mushrooms Zucchini, squash, or eggplant Mixed dishes with rice Yogurt Avocado Mixed dishes with chicken Cooked mixed vegetable dish Muesli Sweet corn Other dried fruits Fish, steamed, grilled, or baked Whole wheat or rye bread Herbal tea Canned fish Nuts Dried apricots or peaches Dip Wheat germ Cottage cheese Beef or veal roast Rissoles or meatloaf Veal or beef schnitzel Potatoes, fried or roasted Beef steak Mixed dishes with lamb Fish, fried Fried rice Chicken, roasted or fried Pies or savory pastries Mixed dishes with beef
0.721 0.712 0.683 0.626 0.586 0.585 0.584 0.520 0.476 0.464 0.432 0.356 0.300 0.291 0.237 0.228
Factor 2: Vegetables
Factor 3: Grains and Fish
Factor 4: Meat
Factor 5: Processed Foods
Factor 6: Salad
0.203 0.223 0.220 0.217 0.643 0.622 0.565 0.553 0.494 0.485 0.480 0.362 0.350
0.267 0.378 0.221 0.223 0.284
0.349 0.319 0.262 0.209
0.335
0.215 0.504 0.438 0.430 0.388 0.378 0.361 0.353 0.294 0.290 0.288 0.281 0.276 0.276 0.265 0.256 0.252 0.236 0.229 0.218 0.210
0.274
0.204
0.511 0.486 0.475 0.374 0.369 0.365 0.363 0.360 0.356 0.336 0.334 (Continued)
7.e1
Amirul Islam et al
Associations of Dietary Patterns with AMD Table 1. (Continued.)
Factor 1: Fruit Pork, chops or roast Eggs, fried or scrambled Feta cheese Seafood Cooked dried beans, chickpea or pea dishes Chicken, boiled or steamed Mixed dishes with egg Rabbit or other game Pickled vegetables Beans, pea, or lentil soup Other offal meats Fish, smoked White bread Cakes or sweet pastries Sweet biscuits Margarine Cheddar or similar cheeses Chocolate confectionery Puddings Sausages or frankfurters Cream or sour cream Ice cream Tea Jam, honey, or syrups Other confectionery Bacon Ham Vegemite Peanuts or peanut butter Lamb, chops or roast Custard Butter Dry biscuits Corn chips, potato chips Corned beef (silverside) Lettuce, endive, or other salad greens Cucumber Tomato Celery or fennel Capscium Beetroot Onion or leeks Coleslaw % of variance explained
Factor 2: Vegetables
Factor 3: Grains and Fish
0.240
Factor 4: Meat 0.324 0.303 0.274 0.268 0.265 0.264 0.261 0.244 0.222 0.221 0.213 0.201
Factor 5: Processed Foods
Factor 6: Salad
0.207 0.209 0.223
0.245
0.381
0.262
0.246
0.249
0.454 0.421 0.412 0.399 0.385 0.361 0.359 0.359 0.352 0.349 0.337 0.317 0.307 0.299 0.284 0.272 0.253 0.253 0.252 0.244 0.227 0.213
0.304 0.351 0.228 14.1
0.278 5.5
4.9
0.226 3.4
3.0
0.722 0.703 0.622 0.614 0.506 0.428 0.379 0.343 2.8
7.e2
Age-related macular degeneration (AMD) is a complex genetic disease, and the major genetic components are now well understood.1 However despite this, diet is emerging as a potentially important modifiable risk factor for AMD.2e9 To date, results from dietary studies have largely been based on analyses evaluating individual foods or food groups. The results suggest that some aspects of diet could influence the risk of AMD, but the associations found have not been consistent across studies.4,10e16 Diets high in trans fat, red meat, and alcohol have been associated with an increased risk of AMD,3e5 whereas higher intakes of fish have been associated with a lower risk of AMD.16,17 In general, foods are not consumed in isolation, and intakes of many nutrients are highly correlated to common food sources. Therefore, an integrated approach to the 2014 by the American Academy of Ophthalmology Published by Elsevier Inc.
investigation of dietary impact on chronic disease where dietary patterns are assessed, rather than each individual food or nutrient, may be more informative and predictive of disease risk.18e20 Several studies have evaluated dietary combinations and AMD using predefined food groups.13,21,22 However, to date, no study has evaluated the associations of dietary patterns derived from principal components analysis (PCA) and AMD. Principal components analysis is a data-driven statistical method used to reduce a large number of intercorrelated variables (e.g., dietary items) into a few distinct factors of intercorrelated variables within each factor that are dissimilar between factors.23 In this way, foods are grouped according to how they were actually consumed in a cohort, rather than by preconceived ideas of how foods should be consumed. ISSN 0161-6420/14/$ - see front matter http://dx.doi.org/10.1016/j.ophtha.2014.01.002
1
Ophthalmology Volume -, Number -, Month 2014 These groups of foods are aggregated together into patterns or factors, which are then evaluated for associations with AMD. Associations of dietary patterns with AMD may provide more insight into how diet is associated with the risk of developing AMD than concentrating on individual food items,24 providing information on which health messages can be based. We used PCA to identify dietary patterns among Melbourne Collaborative Cohort Study (MCCS) participants and assessed the associations of these patterns with AMD.
Methods Study Population The MCCS recruited 41 514 participants (24 469 women) aged 40 to 69 years at baseline (1990e1994) to examine links between diet and chronic diseases.25,26 To broaden the range of dietary intakes, participants born in Greece and Italy were deliberately oversampled to comprise one quarter of the cohort, whereas participants born in Australia, New Zealand, and United Kingdom made up the remainder of the cohort.25 Of the total sample, 27 883 (67%) attended a follow-up examination between 2003 and 2007; 3754 (9%) had died, and 819 (2%) had left Australia or Victoria. The remaining 9058 participants (22%) were lost to follow-up or not interested in participating. Of those who were alive and living in Victoria, the participation rate was 27 883/36 941 (73%), of whom 22 405 (80%) participated in fundus photography. The Human Research Ethics Committees of the Cancer Council Victoria and Royal Victorian Eye and Ear Hospital approved the AMD study protocol. The research adhered to the tenets of the Declaration of Helsinki.
Dietary Assessment Dietary data were collected at baseline using a self-administered 121-item food frequency questionnaire (FFQ) specifically developed for this study.27 Most items were analyzed as daily equivalent frequencies, whereas intakes of olive and vegetable oils were analyzed in milliliters/week. The repeatability of the FFQ after 12 months showed intraclass correlation coefficients >0.50 for most food items.28 Energy intake was calculated using sexspecific standard portions, together with Australian food composition data.29 Participants with estimated energy intakes in the upper and lower 1% of the sex-specific distributions, which suggested that the FFQ had been improperly completed, were excluded. Almost all dietary studies use criteria related to energy intake to exclude people with reported intakes that are unlikely to be real.30 Because we recognize the limitations of FFQs in measuring absolute amounts, people at the extreme 1% of high or low energy intakes were excluded. People with these energy extremes are unlikely to have reported a real usual intake, so including their data might distort the eating patterns identified. Participants with a history of heart attack, angina, or diabetes at baseline also were excluded from our analysis because these diagnoses had the potential to result in dietary change, and we have evidence from this cohort that this was the case (unpublished data, MCCS internal validation, December 2005).
Age-related Macular Degeneration Detection At follow-up from 2003 to 2007 (on average, 13 years after the baseline assessment), digital nonstereoscopic 45 macular photographs of both eyes were taken with a Canon CR6-45NM nonmydriatic retinal camera (Canon Inc, Tokyo, Japan). Retinal
2
photographs were graded for AMD by 2 experienced graders using “OptoLite/OptoMize Pro” software (Digital HealthCare Image Management Systems, Cambridge, UK), according to the International Classification System for grading of AMD.31 Qualitycontrol procedures for the photographs in the MCCS have been described.32
Definitions of Age-related Macular Degeneration Participants were allocated to a single AMD category according to the more advanced visible changes on the macular images of the worse affected eye. Early stages of AMD were defined as the presence of 1 or more drusen 125 mm (with or without pigmentary abnormalities) or 1 or more drusen 63 to 124 mm with pigmentary abnormalities in a 6000-mm diameter grading grid centered on the fovea, in the absence of advanced AMD in either eye (geographic atrophy or neovascular AMD).
Covariates Age, sex, country of origin, smoking status, education level, total energy intake, multivitamin supplement use, body mass index, waistehip ratio, physical activity, and alcohol consumption were included in logistic regression models and retained if they changed the beta-coefficient for any of the dietary factors by more than 5%.
Statistical Analysis The PCA23 of 121 food items plus olive and vegetable oils was performed to extract dietary factors using dietary data collected at baseline. This was followed by orthogonal (varimax) rotation to ensure that the factors were uncorrelated. Factor analysis aims to evaluate which foods are correlated, looking for underlying factors within diets of the population.19 It is used as a way of reducing many variables to few and has been recently used in dietary epidemiology.24 This means that a factor has a distinct cluster of interrelated items with similarities within, but minimal similarities between the factors. Factors with eigenvalues >2.7 were considered to reflect realistic eating patterns and were used in the analysis. Food items with an absolute value of at least 0.2 for the factor loading (the correlation between dietary items and factors) were retained to define the dietary Factors 1 to 6, which is a commonly used cutoff.20 The cutoff used does not affect the calculation of the factor score. For each participant, the factor score is computed as the sum of products of the observed intake frequency values multiplied by their factor loading. Factor scores were then analyzed by quartile and median groupings. We used binary logistic regression models to estimate the odds ratio (OR) and 95% confidence intervals (CIs) for both early-stage and advanced AMD, adjusting for covariates. In the analyses for advanced AMD, early-stage AMD cases were excluded and vice versa. Factor scores were divided into groups, cut at the median or at quartiles for analysis, with the first quartile as the reference group. Tests for trend across quartiles were then performed using the medians in each group. Interactions among sex, age group, grouped country of birth (participants born in Australia and the United Kingdom compared with Southern European participants), and factor scores were tested before performing the analysis. A subanalysis was performed investigating the associations with country of birth (participants born in Australia and the United Kingdom compared with Southern European participants) and food factors for early-stage and advanced AMD. All analyses were performed using the Statistical Package for the Social Sciences version 19.0.1 (SPSS Inc, Chicago, IL).
Amirul Islam et al
Associations of Dietary Patterns with AMD
Results Of the 22 405 participants who had retinal photographs assessed, 21 132 (94%) had retinal photographs gradable for AMD. After excluding 1364 participants at the extreme 1% of energy intake at baseline (n ¼ 311) and participants with a history of heart attack/ angina/stroke (n ¼ 688) or diabetes at baseline (n ¼ 365) (with the potential for dietary modification), we included the remaining 19 768 participants in the analysis. Of the 19 768 participants, 2508 (12.7%) had early stages of AMD and 108 (0.55%) had advanced AMD. Compared with the MCCS participants who were not included in this analysis, those included tended to be younger (mean age at baseline, 53 vs. 56 years), female (61% vs. 59%), and more educated (high school or higher level of education 91% vs. 72%), and were more likely to have never smoked (61% vs. 56%) and less likely to have been born in Greece (4.1% vs. 17.4%) or Italy (8.3% vs. 17.1%), all with P values <0.01. The prevalence of early AMD was 12.7% in participants included in the study and 13.8% in 1364 participants who were excluded (P ¼ 0.28). The prevalence of late AMD was 0.6% in included participants and 1.2% in excluded participants (P ¼ 0.02). We identified 6 factors with minimum eigenvalues of 2.7, which accounted for 33% of the total variance that could be explained by 120 factors using PCA. Factor loadings for food items with loadings having absolute values of 0.2 or greater for any factor, sorted in descending order, are shown in Table 1 (available at http:// aaojournal.org). The factor loadings are the correlation coefficients for each variable on that factor, so that variables with higher factor loadings are those that tend to define the factor. Factor 1, “fruit,” was characterized by intakes of a variety of fruits, including stone, citrus, and other fruits and olives. Factor 2, “vegetables,” was characterized by intakes of a variety of vegetables, including pumpkin, cauliflower, green beans, and potato and an avoidance of pasta, pizza, and salami. Factor 3, “grains and fish,” was characterized by more frequent intakes of boiled rice; muesli; steamed, grilled, and canned fish; steamed or boiled chicken; certain vegetables, including mushrooms, zucchini, spinach, and broccoli; and nuts; and avoidance of white bread. Factor 4, “red meat,” was characterized by intake of red meat (beef, lamb), fried and smoked fish, eggs, and avoidance of whole wheat or rye bread. Factor 5, termed “processed foods,” was characterized by intake of cakes, sweet biscuits, confectionery, ice cream, cheddar or similar cheeses, margarine, pudding, eggs, sausages, bacon, and ham. The last Factor, “salad,” was characterized mainly by intake of uncooked salad (lettuce, cucumber, tomato) and legumes. We have abbreviated the factors with a descriptive term based on the predominant items that load onto the factors. The MCCS participants included in these analyses had lower mean scores for “fruit” (0.042 vs. 0.047), “red meat” (0.116 vs. 0.131), and “salad” (0.029 vs. 0.033), but higher scores for “vegetables” (0.015 vs. 0.018), “grains and fish” (0.117 vs. 0.131), and “processed foods” (0.155 vs. 0.173) than those who were not included, all with P values <0.01. The associations of dietary factor scores with early stages and advanced AMD are shown in Table 2. The final adjusted model included age, sex, country of origin, smoking, education, multivitamin use, and total energy intake as covariates because they were established risk factors of AMD or a potential effect modifier such as sex, or they changed the beta-coefficient for any of the dietary factors by more than 5%. Body mass index, waiste hip ratio, physical activity, and alcohol consumption did not change the associations and were not retained in the analyses. We found Factor 3, “grains and fish,” associated with a 51% decreased odds of advanced AMD after adjustment for covariates, fourth versus first quartile (OR, 0.49; 95% CI, 0.28e0.87). We also found Factor 4,
“red meat,” associated with a higher prevalence of advanced AMD with increasing ORs across quartiles 1 to 3, third versus first quartile (OR, 2.29; 95% CI, 1.37e3.84). However, the odds were not increased for the fourth quartile of the “red meat” factor. A value above or equal to the median score for the “red meat” factor was associated with a higher prevalence of advanced AMD (OR, 1.46; 95% CI, 1.00e2.17). Factor 1, “fruit,” and Factor 2, “vegetables,” scores showed no significant associations with early or advanced AMD, although adjusted ORs trended downward across quartiles for advanced AMD. Factor 5, “processed food,” and Factor 6, “salad,” showed no associations with advanced AMD. None of the dietary patterns was associated with the early stages of AMD. Given the design of the study, we wanted to determine whether there was a difference in relationships between food patterns and AMD in people grouped by the country of birth: Australia/New Zealand/United Kingdom versus Italy or Greece. In this subanalysis, we found that an association of Factor 3 above the median with lower odds of advanced AMD was similar in both groups, with an OR of 0.63 (95% CI, 0.41e0.97) for those born in Australia/New Zealand/United Kingdom and an OR of 0.12 (95% CI, 0.02e0.95) for those born in Italy or Greece, adjusted for covariates. Factor 4 above the median was associated with a higher prevalence of advanced AMD, with an OR of 1.46 (95% CI, 1.0e2.21) for those born in Australia/New Zealand/United Kingdom and a similar trend for those born in Italy or Greece, although the association in this group was not significant (OR, 1.94; 95% CI, 0.55e6.84).
Discussion The MCCS presents a unique opportunity to study links between diets and those with AMD, because it is the largest single cohort study with comprehensive data on both diet and AMD status. With the use of PCA, we identified 6 dietary factors (patterns) at baseline. Of note, fruit and vegetables ended up in different factors in our data. We observed that participants born in Australia, New Zealand, and the United Kingdom scored higher on the “vegetable” factor, whereas those born in Italy and Greece scored higher on the “fruit” factor, suggesting that in our study people who ate fruit did not always eat vegetables at a similar frequency and vice versa. We observed that a high score for Factor 3, “grains and fish,” was associated with a lower prevalence of advanced AMD, whereas Factor 4, “red meat,” score was positively associated with the prevalence of advanced AMD, adjusted for covariates. To our knowledge, no other study assessing AMD risk has evaluated dietary patterns derived from datadriven PCA. The European Eye Study reported on an association of posteriori-derived dietary patterns with retinal vessel caliber in an elderly population, but to date it has not reported on any AMD association.33 However, dietary patterns and their associations with other chronic diseases, such as cancer and cardiovascular disease,24,34,35 have been reported. In the MCCS, a dietary pattern characterized by meats and fatty foods was associated with increased diabetes risk, which supports the notion that patterns of food intake are useful to study in relation to risks of chronic disease.36 Previous studies have evaluated predefined nutrient combinations and AMD risk. These studies used presumptive cutoff levels rather than evaluating the actual food intake in their cohorts. A retrospective case-control study evaluating overall diet quality on the basis of the Alternate Healthy
3
Ophthalmology Volume -, Number -, Month 2014 Table 2. Associations between Dietary Factors and Age-related Macular Degeneration* Reference Group Factor Scores Factor 1: Fruit First quartile, <0.59 Second quartile, 0.59 to 0.26 Third quartile, 0.26 to 0.28 Fourth quartile, >0.28 P for trend Mediany Factor 2: Vegetables First quartile, <0.63 Second quartile, 0.63 to 0.0.6 Third quartile, 0.06 to 0.57 Fourth quartile, >0.57 P for trend Mediany Factor 3: Grains and Fish First quartile, <0.69 Second quartile, 0.69 to 0.15 Third quartile, 0.15 to 0.55 Fourth quartile, >0.55 P for trend Mediany Factor 4: Red Meat First quartile, <0.63 Second quartile, 0.63 to 0.18 Third quartile, 0.18 to 0.40 Fourth quartile, >0.40 P for trend Mediany Factor 5: Processed Foods First quartile, <0.69 Second quartile, 0.69 to 0.08 Third quartile, 0.08 to 0.58 Fourth quartile, >0.58 P for trend Mediany Factor 6: Salad First quartile, <0.65 Second quartile, 0.65 to 0.14 Third quartile, 0.14 to 0.45 Fourth quartile, >0.45 P for trend Mediany
Early Stages of AMD
n
n
%
OR
4443 4454 4315 3937
630 640 653 585
12.4 12.6 13.1 12.9
8252
1238
13.0
1.00 0.93 0.96 0.93 0.40 0.98
3936 4443 4599 4171
573 583 660 692
12.7 11.6 12.5 14.2
8770
1352
13.4
3589 3961 4559 5040
547 603 671 687
13.2 13.2 12.8 12.0
9599
1358
12.4
4708 4553 4236 3652
709 658 631 510
13.1 12.6 13.0 12.3
7888
1141
12.6
2947 4394 4899 4909
470 647 637 754
13.8 12.8 11.5 13.3
9808
1391
12.4
4175 4465 4479 4030
606 662 667 573
12.7 12.9 13.0 12.4
8509
1240
12.7
1.00 0.90 0.92 0.97 0.77 0.99 1.00 1.04 1.07 1.04 0.60 1.03 1.00 0.98 1.02 0.95 0.54 1.0 1.00 1.05 0.95 1.08 0.41 0.97 1.00 1.06 1.07 0.99 0.84 1.0
Advanced AMD 95% CI
n
%
OR
0.82e1.05 0.85e1.08 0.82e1.05
29 21 29 29
0.6 0.5 0.7 0.7
0.90e1.07
58
0.7
1.00 0.49 0.69 0.81 0.97 1.08
0.80e1.03 0.81e1.05 0.85e1.11
20 17 25 46
0.5 0.4 0.5 1.1
0.91e1.09
71
0.8
0.92e1.18 0.94e1.21 0.92e1.18
43 28 18 19
1.2 0.7 0.4 0.4
0.94e1.13
37
0.4
0.87e1.10 0.91e1.15 0.84e1.08
25 29 41 13
0.5 0.6 1.0 0.4
0.92e1.10
54
0.7
0.92e1.21 0.83e1.09 0.94e1.25
25 25 26 32
0.8 0.6 0.5 0.6
0.87e1.07
58
0.6
0.94e1.20 0.95e1.21 0.87e1.12
31 23 27 27
0.7 0.5 0.6 0.7
0.92e1.09
54
0.6
1.00 0.57 0.57 0.84 0.20 1.02 1.00 0.67 0.45 0.49 0.008 0.57 1.00 1.41 2.29 0.97 0.41 1.46 1.00 0.79 0.74 0.76 0.34 0.87 1.00 0.82 1.03 1.10 0.71 1.16
95% CI
0.28e0.88 0.41e1.18 0.47e1.40 0.73e1.59
0.28e1.15 0.29e1.13 0.45e1.59 0.65e1.59
0.41e1.10 0.26e0.80 0.28e0.87 0.37e0.86
0.82e2.44 1.37e3.84 0.49e1.95 1.00e2.17
0.44e1.42 0.41e1.36 0.42e1.36 0.58e1.31
0.47e1.42 0.61e1.74 0.65e1.88 0.79e1.72
AMD ¼ age-related macular degeneration; CI ¼ confidence interval; OR ¼ odds ratio. Statistically significant findings (P < 0.05) are shown in boldface. *The ORs and 95% CIs, adjusted for age, sex, country of origin, smoking status (never, current, and past), education level (below high school level vs. equal or more than high school), multivitamin supplement use, and total energy intake. y The ORs (95% CIs) for equal to or above median factor scores compared with below median scores.
Eating Index, which derives scores according to high/low intakes in each of 5 food groups (grains, vegetables, fruits, nuts and soy, and ratio of white to red meat) and fat intake, reported that people in the top quartile of diet quality had half the risk of advanced AMD (OR, 0.54; 95% CI, 0.30e0.99) than those in the bottom quartile.21 Although the dietary habits in this study were reported after AMD diagnosis, the results suggested that high grain intake along with white meat may be protective for advanced AMD. The second study was the Carotenoids in Age-Related Eye Disease Study of 1313 women aged 50 to 79 years,37 which used a 2005 modified Healthy Eating Index
4
(mHEI)13 based on FFQ data.38 The mHEI represented a healthy diet that included abundant intakes of fruits, vegetables, and whole grains, and a low intake of sugar, fat, saturated fat, and alcohol.13 They reported 46% lower odds for early AMD in the highest compared with the lowest quintile of mHEI (OR, 0.54; 95% CI, 0.33e0.88); because there were only 12 cases of advanced AMD, they were excluded from the main analyses.13 Previous studies evaluating single nutritional items have shown inverse associations between AMD and intakes of fish,2,10 fruits, vegetables,39e41 bread, cereals, and foods with a low glycemic index;42,43 whereas others, including
Amirul Islam et al
Associations of Dietary Patterns with AMD
the MCCS, have reported positive associations between red or processed meat intake and AMD.3,44,45 These findings are along similar lines to our food pattern analyses. When we performed subanalyses based on grouped country of birth, we found that the associations between lower prevalence of advanced AMD and Factor 3 (grains and fish) were similar in those born in Australia, New Zealand, the United Kingdom, and Greece. The associations between a higher prevalence of advanced AMD and Factor 4 (red meat) were in similar directions in both groups, but significant only in those born in Australia, New Zealand, and the United Kingdom. The latter may have been due to the smaller numbers available in this subgroup. We did not find any significant associations between dietary patterns and early AMD. This may be due to an issue of diet having less of a relationship in primary prevention (prevention of development of early AMD) compared with secondary prevention of AMD (prevention of advanced AMD from early AMD).7 It is possible that diet does not influence early AMD to the same extent as driving the disease to the advanced stages. This lack of association may also be due to other unadjusted variables, such as genetic data (effect modification due to genotype) or other unknown confounders.12 The early AMD phenotype is diverse; additional genotype and phenotype characterizations, with the use of other modalities, like autofluorescence, may help clarify associations with diet in the future. One advantage of our factor analysis is that foods were not grouped before analysis. Thus, we avoided making a priori assumptions about what should be grouped. For example, in the Carotenoids in Age-Related Eye Disease Study, meat, fish, and beans were grouped together, but from our data, steamed fish were loaded on a different factor than fried fish and red meat, where they had opposite associations with advanced AMD. Other strengths of our study include its large sample size, detailed AMD grading of high-resolution images with the prevalence of AMD closely matching other epidemiologic studies,46,47 and the broad range of dietary exposures as a result of targeting different ethnicities for recruitment and also the assessment of dietary intakes a decade before AMD detection, (before dietary modifications and nutrient supplements for AMD were commonly recommended).
Study Limitations The study limitations included a natural high attrition that is not unexpected over a long 10- to 17-year follow-up period in an aging cohort, which may lead to survival bias. Diagnosis of AMD based on color fundus photographs was the only feasible method of studying more than 22 000 people at the time when the study was carried out, as this was before the widespread use of other modalities. The lack of optical coherence tomography or fluorescein angiography data would most likely have led to an underdiagnosis of AMD. We did not obtain data on family history of AMD, and the lack of genotyping on all MCCS participants did not allow us to investigate the potential genetic and dietary interactions in AMD. We were also unable to adjust for potential Age-related Eye Disease Study (AREDS)
supplementation use because the dietary data were collected in 1990, 11 years before the AREDS report was published in 2001. Because the specific AREDS formulation was not commercially available in Australia during ophthalmic data collection and widespread use of supplementation for AMD was not common practice at that time, it is unlikely that this would have affected our results. There was also a lower rate of follow-up among Italian and Greek participants, which could introduce participation bias, but analyses for different subgroups did not change our results. Lastly, our measurement of diet was based on a single FFQ administered at baseline, and this may not have been representative of lifelong consumption. However, we did exclude people at baseline with previous diagnoses that appeared to be associated with dietary change. Random error in measuring dietary intake also is likely to have attenuated our associations. In conclusion, our results show that a diet characterized by frequent consumption of boiled rice, muesli, fish (not fried), chicken (not fried), and a variety of vegetables, and avoidance of white bread was associated with a lower prevalence of advanced AMD, whereas a diet characterized by a pattern of eating red and processed meats and fried foods was associated with a higher prevalence of advanced AMD. The results from the dietary pattern analysis cannot reveal the absolute amounts of certain foods to eat or avoid, but they do give direction as to the patterns of food intakes that are potentially important to consider when trying to reduce the risk of AMD. Thus, despite a well-characterized genetic risk profile for AMD, evidence is continuing to mount that the choices we make about foods we consume may play a role in contributing to the risk of developing AMD.
References 1. Fritsche LG, Chen W, Schu M, et al. Seven new loci associated with age-related macular degeneration. Nat Genet 2013;45:433–9, 9e1e2. 2. Seddon JM, George S, Rosner B. Cigarette smoking, fish consumption, omega-3 fatty acid intake, and associations with age-related macular degeneration: the US Twin Study of AgeRelated Macular Degeneration. Arch Ophthalmol 2006;124: 995–1001. 3. Chong EW, Simpson JA, Robman LD, et al. Red meat and chicken consumption and its association with age-related macular degeneration. Am J Epidemiol 2009;169:867–76. 4. Chong EW, Kreis AJ, Wong TY, et al. Alcohol consumption and the risk of age-related macular degeneration: a systematic review and meta-analysis. Am J Ophthalmol 2008;145: 707–15. 5. Adams MK, Chong EW, Williamson E, et al. 20/20eAlcohol and Age-related Macular Degeneration: The Melbourne Collaborative Cohort Study. Am J Epidemiol 2012;176: 289–98. 6. Chong EW, Robman LD, Simpson JA, et al. Fat consumption and its association with age-related macular degeneration. Arch Ophthalmol 2009;127:674–80. 7. Chong EW, Wong TY, Kreis AJ, et al. Dietary antioxidants and primary prevention of age related macular degeneration: systematic review and meta-analysis. BMJ 2007;335:755.
5
Ophthalmology Volume -, Number -, Month 2014 8. Lutein þ zeaxanthin and omega-3 fatty acids for age-related macular degeneration: the Age-Related Eye Disease Study 2 (AREDS2) randomized clinical trial. JAMA 2013;309:2005–15. 9. Seddon JM, Rosner B, Sperduto RD, et al. Dietary fat and risk for advanced age-related macular degeneration. Arch Ophthalmol 2001;119:1191–9. 10. Chong EW, Kreis AJ, Wong TY, et al. Dietary omega-3 fatty acid and fish intake in the primary prevention of age-related macular degeneration: a systematic review and metaanalysis. Arch Ophthalmol 2008;126:826–33. 11. Smith W, Mitchell P, Leeder SR. Dietary fat and fish intake and age-related maculopathy. Arch Ophthalmol 2000;118:401–4. 12. Wang JJ, Rochtchina E, Smith W, et al. Combined effects of complement factor H genotypes, fish consumption, and inflammatory markers on long-term risk for age-related macular degeneration in a cohort. Am J Epidemiol 2009;169:633–41. 13. Mares JA, Voland RP, Sondel SA, et al. Healthy lifestyles related to subsequent prevalence of age-related macular degeneration. Arch Ophthalmol 2011;129:470–80. 14. Mares-Perlman JA, Brady WE, Klein R, et al. Dietary fat and age-related maculopathy. Arch Ophthalmol 1995;113:743–8. 15. SanGiovanni JP, Chew EY, Clemons TE, et al. The relationship of dietary carotenoid and vitamin A, E, and C intake with age-related macular degeneration in a case-control study: AREDS Report No. 22. Arch Ophthalmol 2007;125:1225–32. 16. Tan JS, Wang JJ, Flood V, Mitchell P. Dietary fatty acids and the 10-year incidence of age-related macular degeneration: the Blue Mountains Eye Study. Arch Ophthalmol 2009;127:656–65. 17. Swenor BK, Bressler S, Caulfield L, West SK. The impact of fish and shellfish consumption on age-related macular degeneration. Ophthalmology 2010;117:2395–401. 18. Hu FB. Dietary pattern analysis: a new direction in nutritional epidemiology. Curr Opin Lipidol 2002;13:3–9. 19. Reedy J, Wirfalt E, Flood A, et al. Comparing 3 dietary pattern methodsecluster analysis, factor analysis, and index analysiseWith colorectal cancer risk: The NIH-AARP Diet and Health Study. Am J Epidemiol 2010;171:479–87. 20. Slattery ML, Edwards SL, Boucher KM, et al. Lifestyle and colon cancer: an assessment of factors associated with risk. Am J Epidemiol 1999;150:869–77. 21. Montgomery MP, Kamel F, Pericak-Vance MA, et al. Overall diet quality and age-related macular degeneration. Ophthalmic Epidemiol 2010;17:58–65. 22. Chiu CJ, Milton RC, Klein R, et al. Dietary compound score and risk of age-related macular degeneration in the age-related eye disease study. Ophthalmology 2009;116:939–46. 23. Jolliffe IT. Introduction: Definition and Derivation of Principal Components Analysis. In: Principal Component Analysis. Springer, NY: Springer Series in Statistics; 2002:1–6. 24. Newby PK, Tucker KL. Empirically derived eating patterns using factor or cluster analysis: a review. Nutr Rev 2004;62:177–203. 25. Giles GG. The Melbourne Study of Diet and Cancer. Proc Nutr Soc Aust 1990;15:61–8. 26. Giles GG, English DR. The Melbourne Collaborative Cohort Study. International Agency for Research on Cancer Science Publication. Lyon, France: International Agency for Research on Cancer; 2002;156:69–70. 27. Ireland P, Jolley D, Giles G, et al. Development of the Melbourne FFQ: a food frequency questionnaire for use in an Australian prospective study involving an ethnically diverse cohort. Asia Pac J Clin Nutr 1994;3:19–31. 28. Hodge AM. Diet Related Predictors of Type 2 Diabetes in the Melbourne Collaborative Cohort Study. PhD Dissertation. Melbourne, Australia: University of Melbourne, Department of Public Health; 2006:65–75.
6
29. Lewis J, Milligan G, Hunt A. NUTTAB95 Nutrient Data Table for Use in Australia. Canberra, Australia: Australian Government Publishing Service; 1995. 30. Willett W. Issues in analysis and presentation of dietary data. In: Willett W. Nutritional Epidemiology. 2nd ed. New York, NY: Oxford University Press; 1998:321–45. 31. Bird AC, Bressler NM, Bressler SB, et al. An international classification and grading system for age-related maculopathy and age-related macular degeneration. The International ARM Epidemiological Study Group. Surv Ophthalmol 1995;39: 367–74. 32. Aung KZ, Robman L, Chong EW, et al. Non-mydriatic digital macular photography: how good is the second eye photograph? Ophthalmic Epidemiol 2009;16:254–61. 33. McEvoy CT, Cardwell CR, Chakravarthy U, et al. A posterioriderived dietary patterns and retinal vessel caliber in an elderly population. Invest Ophthalmol Vis Sci 2013;54:1337–44. 34. Magalhaes B, Peleteiro B, Lunet N. Dietary patterns and colorectal cancer: systematic review and meta-analysis. Eur J Cancer Prev 2012;21:15–23. 35. Maruyama K, Iso H, Date C, et al. Dietary patterns and risk of cardiovascular deaths among middle-aged Japanese: JACC Study. Nutr Metab Cardiovasc Dis 2013;23:519–27. 36. Hodge AM, English DR, O’Dea K, Giles GG. Dietary patterns and diabetes incidence in the Melbourne Collaborative Cohort Study. Am J Epidemiol 2007;165:603–10. 37. Design of the Women’s Health Initiative clinical trial and observational study. The Women’s Health Initiative Study Group. Control Clin Trials 1998;19:61–109. 38. Guenther P, Reedy J, Krebs-Smith S, et al. Development and evaluation of the healthy eating index-2005. Technical report Center for Nutrition Policy and Promotion, United States Department of Agriculture 2007. Available at: http://www. cnpp.usda.gov/healthyeatingindex.htm. Accessed December 20, 2013. 39. Evans J. Antioxidant supplements to prevent or slow down the progression of AMD: a systematic review and meta-analysis. Eye (Lond) 2008;22:751–60. 40. Trevithick J, Massel D, Robertson JM, et al. Model study of AREDS antioxidant supplementation of AMD compared to Visudyne: a dominant strategy? Ophthalmic Epidemiol 2004;11:337–46. 41. A randomized, placebo-controlled, clinical trial of high-dose supplementation with vitamins C and E, beta carotene, and zinc for age-related macular degeneration and vision loss: AREDS report no. 8. Arch Ophthalmol 2001;119:1417–36. 42. Kaushik S, Wang JJ, Flood V, et al. Dietary glycemic index and the risk of age-related macular degeneration. Am J Clin Nutr 2008;88:1104–10. 43. Chiu CJ, Milton RC, Gensler G, Taylor A. Association between dietary glycemic index and age-related macular degeneration in nondiabetic participants in the Age-Related Eye Disease Study. Am J Clin Nutr 2007;86:180–8. 44. Arnarsson A, Sverrisson T, Stefansson E, et al. Risk factors for five-year incident age-related macular degeneration: the Reykjavik Eye Study. Am J Ophthalmol 2006;142:419–28. 45. Cho E, Hung S, Willett WC, et al. Prospective study of dietary fat and the risk of age-related macular degeneration. Am J Clin Nutr 2001;73:209–18. 46. Klein R, Chou CF, Klein BE, et al. Prevalence of age-related macular degeneration in the US population. Arch Ophthalmol 2011;129:75–80. 47. Mitchell P, Smith W, Attebo K, Wang JJ. Prevalence of agerelated maculopathy in Australia. The Blue Mountains Eye Study. Ophthalmology 1995;102:1450–60.
Amirul Islam et al
Associations of Dietary Patterns with AMD
Footnotes and Financial Disclosures Originally received: July 28, 2013. Final revision: December 20, 2013. Accepted: January 2, 2014. Available online: ---.
Manuscript no. 2013-1240.
1
Centre for Eye Research Australia, University of Melbourne, Royal Victorian Eye and Ear Hospital, Melbourne, Victoria, Australia. 2 Cancer Epidemiology Centre, Cancer Council Victoria, Melbourne, Victoria, Australia. 3
Centre for Molecular, Environmental, Genetic and Analytic Epidemiology, School of Public Health, University of Melbourne, Australia. *Drs. Islam and Chong contributed equally to this article as first authors. Presented at: The Association for Research in Vision and Ophthalmology, May 5e9, 2013, Seattle, Washington. Financial Disclosure(s): The author(s) have no proprietary or commercial interest in any materials discussed in this article. Funding: Victoria Health, The Cancer Council Victoria, and National Health & Medical Research Council of Australia (NHMRC) (*Program Grant 209057, Capacity Building Grant 251533, and Enabling Grant
396414) funded the MCCS study. Ophthalmic component was funded by the Ophthalmic Research Institute of Australia and American Health Assistance Foundation. People support was provided through the NHMRC Practitioner Fellowships (R.H.G.), NHMRC Senior Research Fellowships (P.N.B. and J.L.H.), and Macular Degeneration Foundation Blackmores Dr. Paul Beaumont Research Fellowship (L.D.R.). The Centre for Eye Research Australia is a recipient of the NHMRC Centre for Clinical Research Excellence Grant 529923 and Operational Infrastructure Support from the Victorian Government. Abbreviations and Acronyms: AMD ¼ age-related macular degeneration; AREDS ¼ Age-related Eye Disease Study; CI ¼ confidence interval; FFQ ¼ food frequency questionnaire; MCCS ¼ Melbourne Collaborative Cohort Study; mHEI ¼ modified Healthy Eating Index; OR ¼ odds ratio; PCA ¼ principal components analysis. Correspondence: Liubov D. Robman, MBBS, PhD, Centre for Eye Research Australia, 32 Gisborne St., East Melbourne, VIC, Australia, 3002. E-mail: lrobman@ unimelb.edu.au.
7
Ophthalmology Volume -, Number -, Month 2014 Table 1. Factor Loadings from Principle Component Analysis in the Melbourne Collaborative Cohort Study (Food Items with Loadings with Absolute Values of 0.2 or Greater are Shown) Factor 1: Fruit Apricots Peaches or nectarines Plums Grapes Watermelon Pears Honeydew melon Strawberries Oranges or mandarins Figs Apples Pineapple Bananas Olives Grapefruit Fruit salad Pumpkin Cauliflower Cabbage or brussels sprouts Green beans or peas Potatoes, cooked without fat Broccoli Carrots Pasta or noodles Silverbeet, spinach, or other leafy greens Hard grating cheeses Salami or continental sausages Pizza Other breakfast cereals Rice, boiled Mushrooms Zucchini, squash, or eggplant Mixed dishes with rice Yogurt Avocado Mixed dishes with chicken Cooked mixed vegetable dish Muesli Sweet corn Other dried fruits Fish, steamed, grilled, or baked Whole wheat or rye bread Herbal tea Canned fish Nuts Dried apricots or peaches Dip Wheat germ Cottage cheese Beef or veal roast Rissoles or meatloaf Veal or beef schnitzel Potatoes, fried or roasted Beef steak Mixed dishes with lamb Fish, fried Fried rice Chicken, roasted or fried Pies or savory pastries Mixed dishes with beef
0.721 0.712 0.683 0.626 0.586 0.585 0.584 0.520 0.476 0.464 0.432 0.356 0.300 0.291 0.237 0.228
Factor 2: Vegetables
Factor 3: Grains and Fish
Factor 4: Meat
Factor 5: Processed Foods
Factor 6: Salad
0.203 0.223 0.220 0.217 0.643 0.622 0.565 0.553 0.494 0.485 0.480 0.362 0.350
0.267 0.378 0.221 0.223 0.284
0.349 0.319 0.262 0.209
0.335
0.215 0.504 0.438 0.430 0.388 0.378 0.361 0.353 0.294 0.290 0.288 0.281 0.276 0.276 0.265 0.256 0.252 0.236 0.229 0.218 0.210
0.274
0.204
0.511 0.486 0.475 0.374 0.369 0.365 0.363 0.360 0.356 0.336 0.334 (Continued)
7.e1
Amirul Islam et al
Associations of Dietary Patterns with AMD Table 1. (Continued.)
Factor 1: Fruit Pork, chops or roast Eggs, fried or scrambled Feta cheese Seafood Cooked dried beans, chickpea or pea dishes Chicken, boiled or steamed Mixed dishes with egg Rabbit or other game Pickled vegetables Beans, pea, or lentil soup Other offal meats Fish, smoked White bread Cakes or sweet pastries Sweet biscuits Margarine Cheddar or similar cheeses Chocolate confectionery Puddings Sausages or frankfurters Cream or sour cream Ice cream Tea Jam, honey, or syrups Other confectionery Bacon Ham Vegemite Peanuts or peanut butter Lamb, chops or roast Custard Butter Dry biscuits Corn chips, potato chips Corned beef (silverside) Lettuce, endive, or other salad greens Cucumber Tomato Celery or fennel Capscium Beetroot Onion or leeks Coleslaw % of variance explained
Factor 2: Vegetables
Factor 3: Grains and Fish
0.240
Factor 4: Meat 0.324 0.303 0.274 0.268 0.265 0.264 0.261 0.244 0.222 0.221 0.213 0.201
Factor 5: Processed Foods
Factor 6: Salad
0.207 0.209 0.223
0.245
0.381
0.262
0.246
0.249
0.454 0.421 0.412 0.399 0.385 0.361 0.359 0.359 0.352 0.349 0.337 0.317 0.307 0.299 0.284 0.272 0.253 0.253 0.252 0.244 0.227 0.213
0.304 0.351 0.228 14.1
0.278 5.5
4.9
0.226 3.4
3.0
0.722 0.703 0.622 0.614 0.506 0.428 0.379 0.343 2.8
7.e2