Content of redox-active compounds (ie, antioxidants) in foods consumed in the United States

¹ From the Department of Nutrition, Institute of Basic Medical Sciences, University of Oslo, Oslo, Norway (BLH, MHC, SKB, KH, and RB); the Biochemistry Department, Virginia Polytechnic Institute and State University, Blacksburg, VA (KMP); and the Division of Epidemiology and Community Health, School of Public Health, University of Minnesota, Minneapolis, MN (DRJ)

² Supported by The Research Council of Norway, The Throne Holst Foundation, and the Norwegian Cancer Society. Food samples were obtained as part of specific agreement Y1-HV-8116-11 between the USDA Nutrient Data Laboratory and Virginia Polytechnic Institute and State University, with support from the National Heart, Lung, and Blood Institute and the National Cancer Institute through interagency agreement Y1-HV-8116 between the National Institutes of Health and the USDA.

³ Address reprint requests to R Blomhoff, Department of Nutrition, Institute of Basic Medical Sciences, University of Oslo, PO Box 1046 Blindern, 0316 Oslo, Norway. E-mail: rune.blomhoff{at}medisin.uio.no.

ABSTRACT  
Background: Supplements containing ascorbic acid, -tocopherol, or ß-carotene do not protect against oxidative stress–related diseases in most randomized intervention trials. We suggest that other redox-active phytochemicals may be more effective and that a combination of different redox-active compounds (ie, antioxidants or reductants) may be needed for proper protection against oxidative damage.

Objective: We aimed to generate a ranked food table with values for total content of redox-active compounds to test this alternative antioxidant hypothesis.

Design: An assay that measures the total concentration of redox-active compounds above a certain cutoff reduction potential was used to analyze 1113 food samples obtained from the US Department of Agriculture National Food and Nutrient Analysis Program.

Results: Large variations in the content of antioxidants were observed in different foods and food categories. The food groups spices and herbs, nuts and seeds, berries, and fruit and vegetables all contained foods with very high antioxidant contents. Most food categories also contained products almost devoid of antioxidants. Of the 50 food products highest in antioxidant concentrations, 13 were spices, 8 were in the fruit and vegetables category, 5 were berries, 5 were chocolate-based, 5 were breakfast cereals, and 4 were nuts or seeds. On the basis of typical serving sizes, blackberries, walnuts, strawberries, artichokes, cranberries, brewed coffee, raspberries, pecans, blueberries, ground cloves, grape juice, and unsweetened baking chocolate were at the top of the ranked list.

Conclusion: This ranked antioxidant food table provides a useful tool for investigations into the possible health benefit of dietary antioxidants.

Key Words: Redox active compounds • oxidative stress • antioxidants • chronic degenerative diseases • oxidative damage • ferric reducing ability of plasma

INTRODUCTION  
The source of energy for most forms of life is photosynthesis, which converts solar energy into redox energy in plants (1). Plants contain high concentrations of numerous redox-active secondary metabolites (ie, antioxidants), such as polyphenols, carotenoids, tocopherols, tocotrienols, glutathione, ascorbic acid, and enzymes with antioxidant activity, which help to protect them from hazardous oxidative damage to plant cell components (1, 2). In animal cells, de novo antioxidant production is much more limited, and oxidative damage is involved in the pathogenesis of most chronic degenerative diseases and aging (3-5). Furthermore, increased amounts of reactive oxygen and nitrogen species (ROS/RNS) are formed in animal cells as a consequence of disease processes (eg, inflammation) and from tobacco smoke, environmental pollutants, food constituents, drugs, ethanol, and radiation (3-6), and, if not eliminated by antioxidants, they may damage extracellular or cellular components (3-6). Oxidative stress reduction through the dietary intake of antioxidants from fruit and vegetables has been suggested to reduce such oxidative damage (7, 8). Many cell culture and experimental animal studies (6, 7, 9-11), as well as observational epidemiologic studies (7, 10, 11), support the hypothesis that intake of foods rich in -tocopherol, ß-carotene, and ascorbic acid were associated with reduced oxidative stress–related diseases. However, large randomized intervention trials using -tocopherol or ß-carotene have not been supportive (12-18). One possible explanation may be that the beneficial health effect is contributed by other antioxidants in fruit and vegetables. There are numerous antioxidants in plants consumed in the diet, including several hundred naturally occurring carotenoids and several thousand phenolic compounds, eg, benzoic acid derivatives, flavonoids, proanthocyanidins, stilbenes, coumarins, lignans, and lignins (19). We suggest that these redox-active compounds, which cooperate in an integrated manner in plants cells, also may cooperate in animal cells. Thus, a network of antioxidants with different chemical properties may be needed for proper protection against oxidative damage (3, 20-22). A ranked table with the total concentration of redox-active secondary plant metabolites may, therefore, be a useful tool for testing this alternative antioxidant hypothesis. Of the various antioxidant assays available (23-26), we decided to use the ferric reducing ability of plasma (FRAP) assay of Benzie and Strain (27). Results of the analysis of 200 fruits, vegetables, spices and herbs, cereals, supplements, juices, and drinks sampled mainly from European countries were previously reported (3, 28, 29). In this study we report the results of an analysis of 1113 food samples that were obtained from the US Department of Agriculture (USDA) National Food and Nutrient Analysis Program (NFNAP) (30), which is based on a nationally representative sampling of each food according to a statistical protocol based on US food consumption data.

METHODS  
Reagents
TPTZ (2,4,6-tri-pyridyl-s-triazine) was obtained from Fluka Chemie AG (Deisenhofen, Switzerland), sodium acetate trihydrate and FeSO₄ · 7H₂O from Riedel-deHaën AG (Seelze, Germany), acetic acid and hydrochloric acid from Merck (Darmstadt, Germany), and FeCl₃ · 6H₂O from BDH Laboratory Supplies (Dorset, United Kingdom). MilliQ water (Millipore, Bedford, MA) and methanol of HPLC-grade obtained from Merck was used for all extractions. HPLC-grade 2-propanol was obtained from Merck. Trolox, ascorbic acid, quercetin, myricetin, and -tocopherol were from Sigma-Aldrich Co (St Louis, MO).

Source of samples
Food samples were obtained from the USDA National Food and Nutrient Analysis Program (NFNAP) (30, 31). Products were collected according to a statistical sampling plan based on US food consumption data, designed to generate nationally representative composites (32, 33). Foods were procured primarily from retail outlets and shipped to a central facility (Virginia Polytechnic Institute and State University, Blacksburg, VA), where they were prepared if necessary (eg, cooked and trimmed of inedible portions), combined into composite samples in some cases, and homogenized. Samples were combined into composites by statistical sampling region or as a single nationwide composite, except for some foods that were shipped directly from the supplier (see footnotes to Tables). Many of the food composites from the NFNAP (172 raw and cooked fresh fruit and vegetables, nuts and seeds, and spices) have also been assayed by using the oxygen radical absorbance capacity method, and the data along with further details on sampling have been reported (33, 34).

Foods and composites were prepared according to standardized, thoroughly documented procedures. Each composite was typically 1000–3000 g in total weight. Representative subsamples of the original foods were taken as necessary. Fresh fruit and vegetables were trimmed of inedible portions (eg, cores, stems, and moldy or bruised areas) immediately before homogenization. Cooked foods were prepared by using conventional methods (eg, microwaving, oven baking, sautéing, boiling, and steaming) following label directions for packaged products. Most composites were homogenized with a 6-L capacity industrial food processor (model RSI6V or BS6V; Robot Coupe USA Inc, Jackson, MS). Fresh fruit and vegetables and other foods (eg, chocolate candy, potato chips, and prepared cakes) were frozen in liquid nitrogen before and during blending. Other homogenization techniques were used, depending on the type of food, and included simple stirring for homogeneous liquids and powders (eg, water, clear juices, oils, and drink mixes), mixing with a hand blender (salad dressings), and grinding with a mill (popcorn kernels and uncooked rice).

Each homogenate was dispensed among 30- or 60-mL glass jars with polytetrafluoroethylene-lined lids (Qorpak, Bridgeville, PA), sealed under nitrogen, and stored at –60 ± 5 °C in the dark. Homogeneity was validated by analysis of moisture, total lipid, ash, or minerals in aliquots drawn from across the typical dispensing sequence of selected composites as described elsewhere (35). The range of storage time at Virginia Tech was from 1 to 63 mo. Samples were shipped on dry ice via express air delivery from Blacksburg, VA, to Oslo, Norway; received in a frozen condition; and stored at –80 °C before analysis. The range of storage time in Oslo was from 0 to 25 wk.

Details of sample description and source, preparation methods, edible yield, compositing and homogenization procedures, and storage were maintained for every food sample and composite but are not included in this report.

Sample preparation
After being thawed, the composites were homogenized, and the analytic aliquots were weighed. Most of the samples were extracted in methanol:water (9:1, by vol). Because of difference in solubility, vegetable oils were extracted in 10 mL 2-propanol. Some fat-rich samples were extracted in 2-propanol:water (9:1, by vol). These alternative extraction procedures gave higher antioxidant values for vegetable oils and the fat-rich foods compared with methanol:water (9:1, by vol). For all other food samples tested, methanol:water (9:1) gave the highest antioxidant value (the solvent used for extraction is indicated in the footnotes to the data tables). The samples were mixed, sonicated in an ice water bath at 0 °C for 15 min, and mixed once more. Three 1.5-mL samples were centrifuged at 12.402 x g for 2 min at 4 °C. The concentration of antioxidants was measured in triplicate aliquots of the supernatant fluid (ie, 9 data points per sample).

Measurements of redox-active compounds, ie, antioxidants
The antioxidant assay of Benzie and Strain (26) was used with minor modifications that allowed quantitation of most water- and fat-soluble antioxidants (28, 29). A Technicon RA 1000 system (Technicon Instruments Corporation, Tarrytown, NY) was used for the measurements of absorption changes that appear when the TPTZ-Fe³⁺ complex reduces to the TPTZ-Fe²⁺ form in the presence of antioxidants. An intense blue color with absorption maximum at 593 nm develops. The measurements were performed after 4 min of incubation at 600 nm. An aqueous solution of 500 µmol/L FeSO₄ · 7H₂O was used to calibrate the instrument. The assay was fully validated as described in a previous report (28). The within-day repeatability measured as relative SD ranged from 0.4% to 6%. The variation in the values for replicate items obtained from the same source was typically between 3 and 10 relative SD (RSD) percentages. Occasionally, some values had a larger variation. In such cases, the antioxidant values were confirmed by reanalysis.

All antioxidant results are reported as absolute values in mmol of electrons/hydrogen atoms donated in the redox reaction per 100 g of sample. In some reports, antioxidant values are given in trolox equivalents. For conversion of absolute values to trolox equivalents, the following data can be used: Trolox has an activity of 831.00 mmol/100 g (n = 5; RSD = 5.5%, ie, 2.08 electrons/hydrogen atoms donated per molecule of trolox) in the assay used in the present study.

The linearity of the method was investigated with standard solutions of FeSO₄ · 7H₂O and ascorbic acid diluted in water and in methanol, trolox diluted in methanol, and -tocopherol diluted in methanol and in 2-propanol. The concentrations used were between 10 and 3000 µmol/L. All concentrations were used for determination of linearity for FeSO₄ · 7H₂O in water and methanol, the 6 lowest concentrations were used for -tocopherol in methanol and 2-propanol, and the 5 lowest concentrations were used for ascorbic acid in water and in methanol and for trolox in methanol. The concentrations were chosen to give an absorbance value of 1.7, which corresponded to an antioxidant value of 3000 µmol/L, which was the linear range according to the instrument manual. The correlation coefficients were in the range 1.000 to 0.998.

Different antioxidants in different solvents [ascorbic acid in water, methanol, and methanol:2-propanol (1:1, by vol); quercetin in methanol and 2-propanol; -tocopherol in methanol, ethanol, and 2-propanol; and myricetin in methanol] at equimolar concentrations gave the same antioxidant value. Thus, these solvents do not influence the examined antioxidants. It was also tested whether different antioxidants in a mixture were additive. The results from the sum of single analyses of each antioxidant corresponded very well with the antioxidant values found in a mixture of the same antioxidants (both in the same and in a mixture of solvents).

Serving sizes
The serving size of a typically consumed portion of each food was determined from the USDA National Nutrient Database for Standard Reference (36), from the US Food and Drug Administration Nutrition Labeling and Education Act (NLEA) guidelines (37), or from actual measurement of average portion weights taken during sample preparation. All serving sizes for fast foods were based on measurements of the samples.

Storage stability studies
Composites of vitamin E–fortified soybean oil, oranges, strawberry jam, raw broccoli, vitamin-enriched whole-grain ready-to-eat breakfast cereal, and a mixed food (vitamin E–fortified soybean oil, oranges, skim milk, raw broccoli, vitamin-enriched whole-grain ready-to-eat breakfast cereal, meatloaf frozen dinner, and teriyaki chicken frozen dinner) were prepared as described above and frozen immediately at –60 °C. Samples were shipped (3 d) from Blacksburg, VA, to Oslo, Norway, on dry ice. Antioxidant contents were determined immediately after arrival and at various times over 0–65 wk at –80 °C. The data show that negligible changes in antioxidant content occurred during storage of these samples at –80 °C for 65 wk (Table 1).

View this table:
TABLE 1. Stability of antioxidants in food composites during storage (for 0–65 wk) at –80°C¹

 
Statistics
The Pearson product-moment correlation coefficients were calculated by the Microsoft Excel software (Microsoft Corporation, Redmond, WA).

RESULTS  
Content of antioxidants in various food groups
The analysis showed large variations in the content of antioxidants in different foods and food categories. The food categories containing the highest antioxidant contents were spices and herbs, nuts and seeds, chocolate and sweets, vegetables and vegetable products, ready-to-eat cereals, desserts and cakes, and berries and berry products (Table 2). Notably, most of these food categories also contained products almost devoid of antioxidants. The food categories containing products with the lowest antioxidant contents were fats and oils; meat, meat products, and substitutes; poultry and poultry products, fish and seafood, and egg and egg dishes.

View this table:
TABLE 2. Antioxidant content in various food categories¹

 
The 50 food products with the highest antioxidant content
The 50 food products containing the highest contents of antioxidants are presented in Table 3. Values are presented as the mean of several brands or sources if several items were isolated of same or similar food products. Ground cloves, dried oregano, ground ginger, ground cinnamon, turmeric powder, walnuts, dried basil, and ground mustard seed contained >10 mmol antioxidants/100 g. Of the 50 food products highest in antioxidant content, 13 were spices, 8 were based on fruit and vegetables, 5 were berries, 5 were chocolate-based, 5 were breakfast cereals, and 4 were nuts or seeds. Notably, red wine and brewed coffee were also among the 50 items with the highest antioxidant content.

View this table:
TABLE 3. The 50 foods with the highest antioxidant content

 
The top 50 foods containing most antioxidants per serving are presented in Table 4. Based on typical serving sizes, blackberries, walnuts, strawberries, artichokes, cranberries, brewed coffee, raspberries, pecan nuts, blueberries, ground cloves, grape juice, and unsweetened baking chocolate were at the top of the ranked list. All of these foods contained more that 2.5 mmol antioxidants per serving. Of the top 50 food products based on serving size, 15 were fruits or fruit juices, 10 were vegetables, 6 were berry products, 4 were chocolate based, 2 were breakfast cereals and 2 were nuts. Red wine and coffee were also among the top 50 items based on antioxidant content per serving size.

View this table:
TABLE 4. The 50 foods with the highest antioxidant contents per serving size¹

 
Effect of food processing
The NFNAP set of foods we analyzed contains foods that were either raw (fresh) or that had been processed in various ways (eg, frozen, baked, microwaved, and boiled). In general, antioxidant contents were preserved fairly well during most types of processing, but there were some exceptions. Interestingly, the antioxidant content increased in products such as carrots, spinach, mushrooms, asparagus, broccoli, cabbage, red cabbage, green and red peppers, potatoes, and tomatoes during microwave cooking, steaming, or boiling (Table 5). Antioxidant values also increased after toasting or baking of bagels, French bread, wheat bread, whole-wheat bread, and pie crust. A decrease in antioxidant content was only observed after cooking by microwave, steaming, or boiling of corn grits, white rice, or spaghetti. Removing the peel from apples and cucumber decreased the antioxidant content to 33–66% and 50% of the amount in the unpeeled products, respectively.

View this table:
TABLE 5. Effects of processing on the antioxidant contents of foods

 
Complete antioxidant food list
The complete food list, which includes the antioxidant content for different brands and sources of products, is presented in Table 6. There was a strong asymmetric distribution of plant and animal food products in the ranked antioxidant content list. In general, plants and plant products in the diet have a much higher antioxidant content than do animal food products. Almost all of the 300 products containing the most antioxidants were pure plant-derived products, whereas almost all of the 300 products with the lowest antioxidant contents were animal-derived products.

View this table:
TABLE 6. The total antioxidant content of 1120 food samples from the US Department of Agriculture (USDA) National Food and Nutrient Analysis Program (NFNAP)¹

 
There are many interesting features of the antioxidant values presented in Table 6. For example, there are large variations in the antioxidant values for diluted fruit drinks. Most had a low antioxidant value, but some that were enriched with ascorbic acid were among the foods with highest antioxidant values. This was also the case for some other foods, such as fortified cereals or foods containing additives such as butylated hydroxyanisole or butylated hydroxytoluene. Interestingly, many plant foods containing high amounts of unsaturated fatty acids had high antioxidant values (eg, walnuts), whereas fish that also contain high amount of unsaturated fatty acids had low antioxidant values. Additionally, processed foods such as lasagna, pudding mix, and chocolate chip cookies were among the 50 foods with the highest antioxidant contents, and chocolate ice cream had a higher antioxidant value than did some fruits (eg, honeydew and green grapes).

Antioxidant values based on serving size are also shown in Table 6. There was also a strong asymmetric distribution of products in the ranked antioxidant content per serving size. Of the 1120 food products analyzed, 87 products contained >1 mmol/serving size. Most of the samples (609 food products) contained between 0.1 and 1.0 mmol/serving. Many samples contained <0.1 mmol/serving (424 food products).

Correlation between different methods for measuring antioxidants
Wu et al (34) previously reported antioxidant values for 172 foods (primarily fruit, vegetables, nuts, and spices) from the NFNAP, of which 93 overlapped with the samples analyzed in this study. They used both a lipophilic oxygen radical absorbance capacity (L-ORAC) assay and a hydrophilic ORAC (H-ORAC) assay. The antioxidant content was calculated as the sum of the values obtained with the L-ORAC and H-ORAC assays. Wu et al measured the total phenolic contents of the products by the Folin-Ciocalteu reagent. Pearson's correlation coefficients between the antioxidant values reported in the present study and the total antioxidant capacity (ie, sum of L-ORAC and H-ORAC assays), L-ORAC, H-ORAC, and total phenolics values of the 93 similar items as determined by Wu et al (34) were 0.788 (y = 0.0221x – 1.7626; R² = 0.6225), 0.823 (y = 0.0550x – 1.5841; R² = 0.6758), 0.579 (y = 0.0221x – 0.5704; R² = 0.3337), and 0.496 (y = 0.2735x – 1.4542; R² = 0.2435), respectively.

DISCUSSION  
The present results of the analysis of 1120 food samples that were obtained from the USDA National Food and Nutrient Analysis Program represent by far the largest published systematic screening of antioxidants in dietary items, including a wide range of both processed foods and fresh fruit and vegetables. Our results are in generally good agreement with those from earlier smaller-scale analyses that used various methods for estimating the total antioxidant contents of foods (28, 29, 34, 38–42). Collectively these data suggest that certain spices, berries, fruits, nuts, chocolate-containing products, vegetables, and cereals are good sources of dietary antioxidants. Additionally, the drinks coffee, green and black tea, red wine, and various berry and fruit juices are good sources of antioxidants.

It should be kept in mind that such antioxidant analyses estimate the content of many hundreds, probably thousands of different compounds belonging to several molecular families. These antioxidants may have very different absorption properties in humans and their transport to, and within, tissues is likely to vary. The antioxidant food table can therefore not be used for dietary recommendations at the present stage. It is necessary to test whether foods rich in antioxidants have the ability to reduce oxidative stress and to reduce the risk of diseases related to oxidative stress.

It is well known that different extraction methods and types of oxidants can produce different estimates of antioxidant content. Although we specifically selected work-up and analysis procedures aimed at including as many antioxidant species as possible in our antioxidant estimate, there may be specific antioxidants that were not detected in the analysis because of limited extractability from the food matrix or low reactivity against the oxidant (eg, reduction potential below that of the assay or slow reaction kinetics).

Plants produce a large diversity of >100 000 low-molecular-weight compounds known as secondary metabolites (43). Secondary metabolites are distinct from the components of intermediary (primary) metabolism in that they are nonessential for the basic metabolic processes of the plant. Most of these secondary metabolites are redox-active compounds (43, 44) that will be picked up by the FRAP assay used in the present study. This diversity of secondary metabolites results from an evolutionary process driven by selection for improved defense against microbial attack, insect or animal predation, ultraviolet light or drought related stress, or other stress factors (44). The phytoanticipins are compounds that are synthesized at a constant rate, whereas phytoalexins are compounds that are more actively synthesized during various types of stress (43, 44). Our food table, therefore, not only represents the amounts of antioxidants in foods, it also identifies those plant foods containing the highest concentrations of secondary plant metabolites. Some of the ambiguity in the interpretation of our table of antioxidant-rich foods relates to other potential stress-reducing effects of phytoalexins and phytoanticipins beyond those directly related to their ability to participate in redox reactions. Thus, in future studies, if any of the plant-based foods that are ranked highest in our food table are proven to be beneficial and reduce stress related–diseases in humans, care should be taken to distinguish such beneficial actions due to their role as redox-active antioxidants from their activity as modulators of specific molecular events in human cells, such as regulation of protein kinases, acetylases, deacetylases, and transcription factors.

Some initial experimental dietary studies support the beneficial effect of dietary plants rich in antioxidants. Pomegranate is a fruit that is extremely rich in antioxidants (28). Aviram et al (45, 46) showed that pomegranate juice administered orally to apolipoprotein E–deficient atherosclerotic mice decreases macrophage lipid peroxidation, LDL susceptibility to oxidation, aggregation and retention, cellular cholesterol accumulation, and development of atherosclerosis. In small-scale human studies they observed that pomegranate juice increased the activity of serum paraoxonase (an HDL-associated esterase that can protect against lipid peroxidation), inhibited serum angiotensin-converting enzyme activity, and reduced systolic blood pressure in hypertensive patients (45, 47). Finally, pomegranate juice consumption for 3 y by patients with carotid artery stenosis reduced common carotid intima-media thickness, blood pressure, and LDL oxidation (48).

Walnuts contain even more antioxidants than do pomegranates (Table 6). The high antioxidant content of walnuts may be related to the observation that walnuts are unique compared with most other nuts, which contain monounsaturated fatty acids, because walnuts are rich in n–6 (linoleate) and n–3 (linolenate) polyunsaturated fatty acids. Five short-term walnut-intervention trials in subjects at risk of coronary heart disease consistently show that walnuts, as part of a heart-healthy diet, lower blood cholesterol concentrations (reviewed in references 49 and 50). These results are supported by several large prospective observational studies in humans, all of which showed a dose response–related inverse association of the relative risk of coronary heart disease with frequent daily consumption of small amounts of nuts, including walnuts (51). In addition, Ros et al (52) recently showed that a walnut diet improves endothelial function in hypercholesterolemic subjects. In March 2004, the FDA accepted the following qualified health claim about walnuts: "Supportive but not conclusive research shows that eating 1.5 ounces per day of walnuts, as part of a low saturated fat and low cholesterol diet and not resulting in increased caloric intake may reduce the risk of coronary heart disease" (53). We suggest that a high antioxidant concentration as well as a favorable polyunsaturated fatty acid pattern may contribute to the beneficial health effects of walnuts.

Serafini et al (54) calculated the total dietary antioxidant intake based on ORAC analysis of 11 antioxidant-rich fruit and vegetables in a population-based case-control study, from which data were collected from 505 newly diagnosed gastric adenocarcinoma patients and 1116 control subjects. They observed that intake of antioxidant equivalents was inversely associated with the risk of both gastric cancers of the cardia and of the distal regions (odds ratio: 0.65; 95% CI: 0.48, 0.89 for the highest quartile of antioxidant intake). Never-smokers with the highest antioxidant intake had the lowest risk of cancer (odds ratio: 0.44; 95% CI: 0.27, 0.71). Among Helicobacter pylori–infected subjects, the odds ratio varied between 0.66 and 0.41 for quartiles 2–4 compared with the lowest quartile of antioxidant intake (54). The extensive antioxidant table we now present (Table 6) provides a useful tool for further epidemiologic investigations into the possible health benefit of dietary antioxidants.

Because an extensive antioxidant table was not previously available, Wright et al (55) constructed a dietary antioxidant index and evaluated its ability to predict lung cancer risk within the Alpha-Tocopherol Beta-Carotene Cancer Prevention Study cohort. The summation of principal component scores derived from individual food analyses of the carotenoid, flavonoid, and vitamin E nutrient groups, plus selenium and ascorbic acid, yielded the composite antioxidant index. At baseline (1985–1988), 27 111 Finnish male smokers aged 50–69 y completed a dietary questionnaire that assessed usual frequency of consumption and portion sizes for the previous 12 mo. A total of 1787 incident cases of lung cancer were identified during a follow-up period of up to 14.4 y (1985–1999). In multivariate proportional hazards models, the relative risks for lung cancer according to increasing quintiles of the antioxidant index were 1.00 (referent), 1.00 (95% CI: 0.87, 1.14), 0.91 (95% CI: 0.79, 1.05), 0.79 (95% CI: 0.68, 0.92), and 0.84 (95% CI: 0.72, 0.98) (P for trend = 0.002), which suggested that a high antioxidant intake is inversely related to lung cancer risk (55).

These preliminary studies suggest a beneficial health effect of consuming dietary plants rich in antioxidants. The overall evidence, however, is limited and much more research is needed. Our extensive total antioxidant food table should be useful for further testing of the antioxidant hypothesis.

ACKNOWLEDGMENTS  
We thank Amy Rasor and Nancy Conley for preparing, shipping, and maintaining the sample descriptive information.

MHC had full access to all of the data in the study and took responsibility for the integrity of the data and the accuracy of the data analysis. RB was responsible for the study concept and design, supervision, and drafting of the manuscript. BLH, MHC, KMP, SKB, and KH were responsible for data acquisition. All authors participated in the analysis and interpretation of the data and the critical revision of the manuscript for important intellectual content. None of the authors had a personal or financial conflict of interest. DRJ is an unpaid member of the Scientific Advisory Council of the California Walnut Commission.

REFERENCES