October 2007: 439 – 450

Special Article

Food, Not Nutrients, Is the Fundamental Unit in Nutrition

David R. Jacobs, Jr., PhD, and Linda C. Tapsell, PhD, FDAA

The identification of nutrients and the study of their of interesting turns, including the attention given to

Downloaded from https://academic.oup.com/nutritionreviews/article/65/10/439/1879933 by guest on 11 September 2020

bioactivity were significant developments in the evo- nutrients over foods. Following observations of relation-
lution of contemporary nutrition science. This review ships between food consumption patterns and health
argues for shifting the focus towards food in order to outcomes, modern science has attempted to identify
better understand the nutrition-health interface. It bioactive components. One classic example concerns the
begins by introducing the concept of food synergy (a consumption of citrus fruit, scurvy in 18th century sea-
perspective that more information can be obtained by farers, and vitamin C.1,2 Sir Walter Haworth and Albert
looking at foods than at single food components) to von Szent-Gyorgyi were awarded the 1937 Nobel Prize
denote the action of the food matrix (the composite of for their work in identifying vitamin C.3,4 Through food
naturally occurring food components) on human bio- guidelines responsive to recommended daily allowances
logical systems. A proposal is then made for the
of nutrients, food has appropriately remained the primary
means by which food-focused research might build the
agent in the practice of scurvy prevention; however, this
knowledge base for etiologic discovery and appropri-
science was eventually commercialized with the produc-
ate dietary advice. The diet-heart disease dilemma is
put forward as an example of where a nutrient-based tion of purified vitamin C, and the use of vitamin C as a
approach has limitations, and a summary of studies nonfood supplement is currently popular.4 Behind those
targeting food composition strengthens the case for a early events rests a multitude of pathways for the devel-
food-based approach. Finally, the argument is made opment of nutrition science and practice.5 This review
that evidence from interventions points back to the argues for the need to build the science base on foods
central position of food in the relationship between because a focus on nutrients tends to oversimplify a
nutrition and health, a position that begs for more complex system and has even done harm.6
whole food-based research.
Key words: foods, nutrients, heart disease, food syn- RESEARCH ON FOOD AND FOOD
ergy, apples, fruit, pomegranates, broccoli, peanuts, COMPONENTS
tomatoes
© 2007 International Life Sciences Institute While identifying bioactive components in food may
doi: 10.1301/nr.2007.oct.439 – 450 be helpful in explaining the health effects of food, such
information is likely to be incomplete. Much remains to
be learned from the context in which those bioactive
INTRODUCTION agents operate in the physiology of the food source (plant
or animal). All foods come from living organisms where
Food is fundamental to health, yet the history of the components interacted to produce biological systems.
scientific endeavor in food and health has had a number These interactions are relevant when the plant or animal
is consumed as food. We call the naturally occurring
Dr. Jacobs is Professor of Epidemiology, School constituents and their interactions the food matrix. Food
of Public Health, University of Minnesota, Minneapo- is consumed to maintain human biological processes, and
lis, Minn., USA and affiliated with the Department of the food matrix influences these processes. We use the
Nutrition, University of Oslo, Oslo, Norway. Dr. Tapsell term food synergy to denote this health action of the food
is Professor and Director of the National Centre of matrix. While the study of these multiple constituents in
Excellence in Functional Foods, University of Wollon- the lives of the plants, animals, and human consumers in
gong, Wollongong, NSW 2522, Australia.
itself represents a whole research enterprise that is
Please direct all correspondence to Dr. Jacobs,
School of Public Health, University of Minnesota, 1300 emerging through the ‘omic’ disciplines (e.g., nutrigeno-
S. 2nd Street, Suite 300, Minneapolis, MN 55454, mics, proteomics, metabolomics), comprehensive, cor-
USA. Phone: ⫹1-612-624-4196; Fax: ⫹1-612-624- rect, and utilitarian models of nutritional biochemistry
0315; E-mail: [email protected] are not likely to be at hand in the near future. At the point

Nutrition Reviews姞, Vol. 65, No. 10 439

of nutrition practice, this suggests a need to treat plant clinical trials seems to be irresistible. The food compo-
and animal foods as primary objects of interest. nent perspective appears to have great weakness in ex-
Epidemiologic studies point to the need to revisit the posing the role of diet in multifactorial chronic disease.
food matrix as the variable of interest in disease risk. The Further, although bias that is not unaccounted for is a
dietary lipid-based understanding of cholesterol and major issue in observational studies, it is unlikely that the
heart disease is a case in point. The nutrient-based form consistent epidemiologic findings of lower chronic dis-
of pursuit was encouraged by profound scientific ad- ease risk in those eating a “healthy” diet can be fully
vances associated with the discovery of vitamins and explained by residual confounding. More research is
other biochemicals and their associated biochemical warranted where food rather than food components are
pathways. In this mode, diet and heart disease was seen the variables of interest.
through an emphasis on total dietary fat, protein, and
carbohydrate. This sort of focus on nutrients may distract

Downloaded from https://academic.oup.com/nutritionreviews/article/65/10/439/1879933 by guest on 11 September 2020

METHODOLOGICAL CHALLENGES
from finding healthful foods not readily identified as
such by a nutrient approach, or the reputation of such Disentangling the specific role of each element in a
food may even be blemished by its high fat content. For synergy becomes both a study design and a mathematical
example, a number of prospective observational studies challenge when more than two or three elements are
suggest that foods and food patterns reduce chronic considered. Given food synergy, the complexity of in-
disease risk. Examples of such foods include whole teractions of food components and even of foods with
grains,7 nuts,8,9 olives,9 and high-fat fish,10,11 and rele- each other presents a challenge for synthesis,28-31
vant food patterns include those that prominently feature whether with a view towards advising individuals or
high-fat foods such as olives and nuts.12-14 The findings towards understanding etiologic pathways in nutrition.
have substantial biological coherence and consistency The study of foods as synergistic wholes has much to
across many study settings. offer for nutrition, particularly in understanding the role
Despite the fact that clinical trials of food pat- of diet and health. The DASH diet study32 is a good
terns9,14 are rare, it is ironic that recent support for example of a food-based approach to studying the role of
building the food-based model of nutrition comes from diet on health over weeks, while the Women’s Health
clinical trials of dietary supplements or are focused on Initiative24 and Lyon Diet Heart Study14 are good exam-
single macronutrients. For example, trials testing the ples of longer term studies; the latter simultaneously
effect of beta-carotene supplementation,6,15-23 total fat evaluated a diet supplement.
reduction,24 and B vitamin supplementation25-27 have An example of how the research might be separated
failed to show reduced chronic disease risk, and in some out is provided by Jacobs and Steffen29 who proposed a
cases have shown increased risk.6,15,25,26 The logic lead- hierarchical structure of dietary patterns, foods, and nu-
ing to these clinical trials is worth examining. For ex- trients for the study of food synergy. They called this
ample, the variable representing ␤-carotene in observa- “top down” or “bottom up” research, as illustrated in
tional studies, which was the precursor to clinical trials Table 1, referring to the spectrum from diets containing
of ␤-carotene supplementation, was computed by aver- whole grains, to whole grains and then parts of the whole
aging ␤-carotene intakes of carotenoid-rich foods. The grain and whole grain components as an example. The
failure of the supplement trials suggests that the original studies at the top of the hierarchy in Table 1 would tend
food variables might have contained more information to have an observational epidemiologic design, those in
about health than did the estimated food component. the middle are more amenable to study in clinical trials,
Thus, estimating the food component lost information. and those at the bottom are more likely to be done in
Whether the studies of ␤-carotene would have been more animal or in vitro models. The intent is that studies at any
successful with doses within the range achievable by diet one of the five levels of experimental design be inter-
has not been studied; rather, the temptation to study preted jointly with studies at other levels. The hypothesis
larger doses of apparently valuable food components in is that some patterns would emerge from this kind of

Table 1. Hierarchical structure of dietary patterns, foods, and nutrients for study of food synergy
Food synergy level Examples of dietary component at each synergy level
Level 5: Dietary pattern ‘Prudent diet’, ‘Western diet’, other combinations of food groups
Level 4: Food groups Whole grain, dairy, fruit, vegetables, meat
Level 3: Whole grain Whole wheat, brown rice, rolled oats
Level 2: Whole wheat Bran, germ, endosperm; extract of fat-soluble portion
Level 1: Bran or a single phytochemical Specific nutrients or phytochemicals
Reprinted with permission from Jacobs and Steffen29 Am J Clin Nutr. 2003;78(Suppl 3):S508 –S513, American Society for Nutrition.

440 Nutrition Reviews姞, Vol. 65, No. 10

thinking that would suggest an important influence on randomized clinical trial settings. Food studies always
health, whether of whole diets, single foods, or food substitute one food for another, because daily energy
components. This form of information then readily trans- requirements must be met. Foods are bulky and easily
lates to food guidance systems. identified by taste and texture, so blinding of study
To use the food synergy concept, it is helpful to treatments is difficult. In trials, particularly those that
distinguish between a food and a drug. All foods derive take years to complete, participant compliance with diet
from living organisms, whether animal or plant, and may is an extremely difficult issue. So, even if the hypothesis
include certain commonly consumed extracts such as were to evaluate the effect of, say, carbohydrate levels in
coffee, tea, and orange juice. Living tissue is extremely the diet, there are real problems with research design and
complex. It seems a good assumption that the vast etiologic questions are difficult to resolve definitively. It
majority of components of plant and animal-based food is possible that the emerging study of metabolomics will

Downloaded from https://academic.oup.com/nutritionreviews/article/65/10/439/1879933 by guest on 11 September 2020

is functional, that it has some kind of biological activity. help to unravel the impact of foods on human systems
Given that some food patterns seem to support the biology, but the methodology for reliably extracting and
human biological system better than others, the health- interpreting information from the large arrays of sub-
fulness or adversity of biological activity in food may be stances encountered in a single metabolomic study needs
quite subtle. It is reasonable that biological activity (and further development. Beyond etiologic questions is the
thereby the bioactive substance) can be healthful or not. translation of that research to practice. In this case, while
The more obvious instances of adverse biological activ- the variable studied may have been nutrient-based, foods
ity, such as with poisonous mushrooms, have been elim- are central to diet advice in nutritional practice.
inated from the diet by the collective human experience. In contrast to foods, drugs are isolated substances,
Food allergies are an example of more nuanced adverse notwithstanding the fact that many are found in plants
biological activity resulting from food consumption. The (for example, HMG-coA reductase is a metabolite in
most subtle adverse effects are with foods or diet pat- fungus34 and salicylate is found in wintergreen35). A
terns, such as the Western diet pattern33 that apparently drug is designed or noted to block or enhance some
slowly increases risk over long periods of ingestion. particular pathway, which is hoped to influence patho-
All of this biological activity must be kept in balance logic progression. Drugs are fairly straightforward to
or else the situation becomes toxic to the living organ- study. Visually, they appear compact and not easily
ism. Under this assumption, living tissue must contain identifiable, so high compliance over long periods is
substances that keep other substances in check. This possible and randomization and blinding can be accom-
represents a real challenge in researching the health plished, even if the associated cost and effort is consid-
effects of food, particularly compared to therapeutics, in erable. Many supplements derived from food are isolated
which there is not a counterbalance of constituents. The substances, such as ␤-carotene, and could therefore be
study designs most suited to therapeutics may not be classified as drugs. In this categorization, and where the
optimal for foods. intended health outcomes are defined in therapeutic
The most obvious and well studied food constituents terms or in terms of prevention of disease, foods enriched
are macronutrients, i.e., those that supply energy: fat with an isolated substance, such as margarine containing
9 kcal/g, protein and carbohydrate 4 kcal/g, alcohol ␤-sitostanol to lower cholesterol levels, can be seen as
7 kcal/g. Other substances are acaloric vitamins and drugs delivered via a food. Here, ␤-sitostanol is a natural
minerals. These substances are often called micronutri- plant substance provided in excess of what would be
ents. However, there are thousands of other substances in obtained from foods, and the observed health benefit can
the food matrix that must be considered as possibly be attributed to the substance.36 Calcium-enriched or-
leading to biological activity, possibly synergistically ange juice is an example of the added substance not
with each other, that in some sense could equally be being a natural part of the food vehicle, and therefore not
deemed essential for life because they are not produced naturally part of the given food matrix. Not being in the
by nor do they augment human biological systems. These food matrix, the added substance, whether ␤-carotene,
include signal transducers, hormones, sterols, enzymes, ␤-sitostanol, calcium, or something else, may not have
enzyme inhibitors, polyphenols, and fungicides, among the same biologic action as if it were naturally occurring
others. These substances are whatever the plant or animal and integrated with the rest of the compounds in the
needs to stay alive, reproduce, and protect itself from food. A question in the practice of nutrition is whether
predators, enemies, and disease. In this light, it seems obtaining a similar and adequate amount of a food
rather simplistic to evaluate a diet solely in terms of a component in a naturally occurring food matrix is
single food component, yet this is a very popular strategy achievable and what the health effect of that strategy
for evaluating a given dietary pattern. Further, because would be. Either way, the question would need to be
everyone must eat, food is rather difficult to study in addressed through scientific study.

Nutrition Reviews姞, Vol. 65, No. 10 441

 Whole food extracts such as echinacea, soy protein, most coherent approach was nutrient-based, holding en-
or cod liver oil, are also used as supplements, as is ergy intake fixed, rather than being food-based, and it
psyllium, a plant that is not generally eaten as food. contributed to the practice of categorizing foods based on
These substances fall between foods and drugs and nutrient consistencies. Keys et al.50,51 did note in passing
thereby carry some issue with respect to variation in that there were some other dietary factors that influenced
composition. They are more complex than drugs, so serum cholesterol, for example, ␤-sitosterol, which is
could well have the complex nutritional properties of prominent in corn, and pectin, found in apple skins.
foods. They tend to be consumed in relatively large However, these were considered unimportant because
quantities by the few people who consume them at all, the quantity of corn and apples needed to achieve an
and in this way may be more like drugs than foods, important amount of pectin intake was not likely to be
losing some of the multi-substance balance characteristic achievable in a free-living person.
of food. They can often be delivered in pill form, like

Downloaded from https://academic.oup.com/nutritionreviews/article/65/10/439/1879933 by guest on 11 September 2020

These studies and this formula have played a large
drugs, so it may be relatively easy to study them in role in diet and public policy in the United States and
double blind, randomized clinical trials. A design caveat elsewhere over the last 40 or more years. They estab-
with respect to quality control is that supplement com- lished the basis for diets low in total fat, designed to
position varies across products; this aspect would have to reduce saturated fat and lower serum cholesterol, as a
be addressed in a research design. Even then, the re- means to reduce the coronary heart disease epidemic.
search answer might not have uniform applicability to The positive qualities of plant foods that may have been
the commercial product. Processes for dealing with the overlooked in this process, however, were picked up in
evidence base for food are developing across the globe more recent studies that have emphasized nonlipid di-
and represent an attempt to deal with such issues.37-40 etary influences on serum cholesterol. For example, the
The ideas presented above are aimed at learning “Portfolio diet” of Jenkins et al.,52,53 which has a vege-
about the health effects of foods, in order to further tarian pattern high in plant sterols (including ␤-sitostanol
understanding of the nutrition-health interface, enhance enriched margarine), soy protein, viscous fibers includ-
knowledge of etiologic pathways, and provide better ing psyllium, and almonds, lowered blood cholesterol
nutritional advice to consumers. Use of food as the comparably to statins. A related study excluding
fundamental unit of nutrition, attention to the food ma- ␤-sitostanol margarine and psyllium still reported achieving
trix, a hierarchical and integrative view of the way very substantial reductions in serum cholesterol.54
nutrition research is conducted, and a clearer distinction
Concordant with this research showing that it is,
between foods and drugs may all be helpful as the
possibly, not just the macronutrients that explain dietary
science progresses.
effects, biomarkers of heart disease other than lipid
levels have emerged in the scientific literature. Other
FOOD AND HEART DISEASE investigators have shown that diets based on unrefined
plant foods, which, by nature, tend to be low in saturated
One way of viewing the significance of food in the fat and have substantial effects on LDL cholesterol, also
nutrition-health schema is through an analysis of re- influenced emerging important biochemical markers
search underpinning the diet-heart disease observations. such as oxidative stress.55 In this study the unrefined
Keys and Keys41-43 made the classic early observations foods diet contained whole grains, dark green and yel-
that diet pattern plays an important role in this chronic low- or orange-fleshed vegetables and fruits, legumes,
disease. These observations asserted that the many-col- nuts and seeds that contain high concentrations of anti-
ored Mediterranean diet, rich in varied plant foods, was oxidant phenolics, fibers, and numerous other phyto-
healthful, without further feeding study or nutrient re- chemicals that could explain the difference from a re-
search. However profound these observations were, they fined-foods diet.
did not fulfill the criteria for best evidence; hence, From this analysis, our current interpretation is that
feeding studies ensued to try to unravel the complexities the foods approach to inference about diet and serum
and details of the diet and serum cholesterol associations. cholesterol yielded more information about serum cho-
Based on these studies, Keys et al.44-49 developed a lesterol lowering than did the earlier macronutrient ap-
formula that has stood the test of time for expected proach. While the macronutrient approach is not intrin-
change in serum total cholesterol (mg/dl) according to sically incorrect and does answer the specific questions it
changes in the intake of saturated fat as a percent of was intended to answer, it seems to have had a serious
energy intake (S), polyunsaturated fat as a percent of unintended effect. Namely, it appears to have contributed
energy intake (P), and cholesterol (C, in mg/1000 kcal): to a likely false conclusion, that reduced total fat is
1.35 (2 ⌬ S- ⌬ P) ⫹ 1.5 ⌬ √(C), where ⌬indicates healthful independent of the rest of the diet. This has had
change. This neat formula tantalizingly implied that the significant implications for dietary guidelines and the

442 Nutrition Reviews姞, Vol. 65, No. 10

relative positioning of foods in healthy food choice from two general sources, namely analogy with the
guidance systems, such as diet pyramids. Whole foods Mediterranean diet, which is interestingly consistent with
with full nutritional profiles, such as nuts and oils, have the ideas of Ancel Keys, and a statistical method, factor
been repositioned in various ways in these schemes, and analysis. It may be surprising that the two sources give a
‘fat free’ products have emerged with little else to offer similar answer given that the factor analysis method does
in a nutritional sense. not use disease connections; instead, it relies solely on
Dietary intervention studies that show an awareness correlations among foods eaten. Factor analysis is a
of this problem have begun to emerge. In one study of method that arrays people on a scale, and the scale, in
the effect of walnuts on lipids in adults with type 2 turn, depends on the foods, in this case intake of vege-
diabetes mellitus, Tapsell et al.56 positioned walnuts in a tables, fruit, legumes, whole grains, nuts, herbs, dairy
dietary model to assure a favorable P:S ratio, and products, fish, and poultry. We think it likely that the

Downloaded from https://academic.oup.com/nutritionreviews/article/65/10/439/1879933 by guest on 11 September 2020

achieved positive results, but they also reported on the factor analysis can identify diet scores that predict future
protected total antioxidant status of subjects. This anti- disease outcomes because the science is correct and
oxidant protection might have been compromised with some of the public got the healthy diet message. If so,
high intakes of polyunsaturated fat alone. They attributed some people would eat a sufficiently homogeneous diet
this outcome to concomitant high intakes of antioxidants that the prudent and Western patterns can be identified
from walnuts themselves and from the diet model gen- by the factor score.
erous in fruits, vegetables, and whole grains.56 Two The existence of food patterns that relate to future
issues of food synergy emerge here: 1) the synergy chronic disease is supportive of the food synergy per-
provided by the single food itself (in this case polyun- spective. Somehow, the foods together provide risk in-
saturated fat partnered with antioxidants and possibly formation that is harder to see in studies of single foods
other bioactive components), and 2) the synergy of the or nutrients. Whether there is synergy in the statistical
combination of foods in the diet model. Nutrition prac- sense of the whole risk being greater than the sum of its
tice aspects of these points were expanded in a subse- parts, or there is simply an additive effect (i.e., the whole
quent article by this team, which outlined how macro- risk is the sum of its parts) is somewhat irrelevant. Thus,
nutrient targets might be achieved through an integration additional information about risk can be obtained by
of knowledge of food composition and overall nutritional looking at the highest level in the food synergy approach.
balance in a food-based diet model.57 At the same time, however, the food pattern information
Further evidence for the food synergy position can is invariably incomplete. People eat individual foods and
be found in thoughtful analyses of data from observa- combined food dishes. There is some uncertainty as to
tional studies. Hu et al.33 identified two major dietary whether any given food is in the Mediterranean diet or
patterns using a principal components analysis of food the prudent diet pattern. For example, Hu et al.33 did not
intake data obtained with a 131-item food frequency identify nuts as part of the prudent diet pattern, but
questionnaire. The prospective cohort they studied com- others, deriving similar patterns from intake correlations,
prised 44,875 men aged 40 –75 years who were free of do find nuts to be part of the prudent diet.12,13 Alcohol
cardiovascular disease and cancer in 1986 and followed was not considered at all, but it has been shown in many
for 8 years; during the study period, 1089 coronary heart investigations that moderate alcohol consumption is as-
disease cases developed. The prudent pattern was based sociated with reduced risk of coronary heart disease.
on a higher intake of vegetables, fruit, legumes, whole Dairy products were incompletely considered. Almonds
grains, fish, and poultry. The Western pattern was based are part of the portfolio diet and walnuts are both very
on higher intakes of red meat, processed meat, refined high in a long chain fatty acid (␣-linolenic, C18:3, ␻3)
grains, sweets, and dessert, French fries, and high-fat and among the foods with the highest natural antioxidant
dairy products. After adjusting for age and coronary content; the diet pattern scores do not have this level of
heart disease risk factors, the relative risk (RR) for the detail. Therefore, from the food synergy perspective, it is
highest versus the lowest quintile of the prudent pattern helpful to know that some diet patterns are associated
score was 0.70 [95% confidence interval (95%CI), 0.56– with better health than others, but additional separate
0.86; P for trend ⫽ 0.0009]. In contrast, the RR for the investigation of individual foods is also needed.
highest versus the lowest quintile of the Western pattern
score was 1.64 (95%CI, 1.24–2.17; P for trend FOOD COMPOSITION
⬍0.0001). These findings are consistent with the argu-
ment that a food-based approach yields information Another way of viewing the significance of food in
about the relative value of food intake patterns that can the nutrition-health schema is to focus on individual
be readily translated to practice. foods themselves. The food synergy perspective suggests
The idea of a prudent diet may be seen as arising that more information can be obtained by looking at

Nutrition Reviews姞, Vol. 65, No. 10 443

foods than at single food components one at a time. idant activity was attributable to the conventional anti-
There are several examples in the literature that provide oxidant nutrient, vitamin C.
more direct information about interaction among specific More relevant to public health than levels of phe-
food constituents. nolic constituents is the bioactivity seen from the deliv-
The hypothesis that a nutrient, such as cereal fiber, ery of the food components. Following their observa-
could be less informative about mortality risk than whole tions, Eberhardt et al.59 exposed in vitro cancer cell lines
grain foods can be tested with careful analysis of epide- to 50 mg/ml apple extract. Here, the percent inhibition of
miological data. In this example, Jacobs proposed that proliferation in the colon cancer cell line was 43% using
the phytochemicals that distinguish whole grain from apple extract with skin compared to 29% for that without
refined grain food are apparently more healthful than the skin; correspondingly, 57% and 40% inhibition was
cereal fiber.58 This particular analysis saw two subgroups found in the liver cancer cell line (Table 4).
of participants in the Iowa Women’s Health Study se- While these results suggest benefit from the whole

Downloaded from https://academic.oup.com/nutritionreviews/article/65/10/439/1879933 by guest on 11 September 2020

lected, both eating about 6 g/day of cereal fiber, but one food at the cellular level, further evidence could be
getting about 75% of that fiber from whole grain food obtained from animal models to see whether a dose
sources, while the other consumed it from refined grain response could be observed. A subsequent study of 150
food sources. The women who ate cereal fiber from Sprague-Dawley rats studied the effect of apple extract
whole grain sources had significantly reduced total and on drug-induced mammary tumors. The equivalent of
non-cancer, non-cardiovascular mortality, compared to one and six apples per day reduced tumor incidence by
the women with equal intakes of cereal fiber, but mostly 17% and 46%, respectively; the number of tumors and
derived from refined grain sources (Tables 2 and 3). tumor weight were also lower, with the degree increasing
Assuming that cereal fiber is one and the same as the dose of apple extract increased. Thus, the whole
substance, whether derived from nutrient-poor en- food had both in vitro and in vivo dose response effects
dosperm (the only grain part in refined grain) or nutrient- on cancer beyond the vitamin C content, with the skin
rich bran, one concludes that one or more phytochemi- contributing beyond the rest of the apple.60
cals may be more important than fiber in determining Research on the synergistic effects of food compo-
risk. Yet the fiber itself is known to have health effects, nents in whole foods has begun to emerge for a range of
for example on laxation; thus, the picture is complex and foods in relation to diverse biological processes and
consistent with food synergy. conditions. For example, Seeram et al.61 studied synergy
The food synergy argument can be supported from in pomegranate and some of its constituents. In vitro
analyses of parts of food. For example, Eberhardt et al.59 antiproliferative, apoptotic and antioxidant activities of
found that 100 g of apple with skin contained 290 mg of pomegranate juice were compared to similar effects from
total phenolics and 143 mg of total flavonoids, compared equivalent amounts of several of its polyphenol constit-
to 220 mg and 98 mg, respectively, in 100 g of apple uents, including punicalagin, the major fruit ellagitannin,
without skin. This research exposed the diversity and and ellagic acid. Pomegranate juice showed the highest
differential distribution of biochemical activity through- antioxidant activity and the greatest antiproliferative ac-
out the apple. Further, the characterization of the apple tivity against oral, colon, and prostate tumor cell lines by
referred to far more than a nutrient popularly associated inhibiting proliferation from 30% to 100%. Lansky62
with apples, namely vitamin C, only 5.7 mg of which reiterated these general ideas in relation to pomegranates.
was found in 100 g of apples with skin, and 4.0 mg in As with pomegranates, synergy has been observed
100 g of apples without skin. Correspondingly, the total in crucifers. Keck et al.63 fed male F344 rats for 5 days
antioxidant activity was nearly double in the apples with to show that colon and liver quinone reductase activity
skin than those without skin; almost none of this antiox- was greatest with the diet containing broccoli with intact

Table 2. Grain fiber intake (g/2000 kcal) in women aged 55– 69 years in Iowa in 1986 in two matched
groups eating approximately 6 g/2000 kcal per day of total grain fiber intake
Whole grain fiber 71% of Refined grain fiber 77%
total grain fiber intake* of total grain fiber intake†
Sample size 7481 3559
Whole grain fiber intake 4.7⫾0.7 1.3⫾1.0
Refined grain fiber intake 1.9⫾0.8 4.5⫾0.9
Total grain fiber intake (group matching criterion) 6.6⫾1.0 5.8⫾1.2
*3.6 – 6 g/2000 kcal of whole grain fiber and ⬍3.6 g/2000 kcal of refined grain fiber
†⬍3.6 – 6 g/2000 kcal of whole grain fiber and ⱖ3.6 g/2000 kcal of refined grain fiber
Reprinted with permission from Jacobs et al.58 J Am Coll Nutr. 2000;19(Suppl 3):S326 –S330

444 Nutrition Reviews姞, Vol. 65, No. 10

Table 3. Proportional hazards regression analysis of mortality in women in Iowa matched for total grain
intake in 1986 and followed from 1986 to 1997
Hazard rate ratio adjusted Hazard rate ratio
Mortality for age and energy intake multivariate adjusted
Total mortality 0.75 (0.67, 0.84) 0.83 (0.73, 0.94)
Coronary heart disease 0.82 (0.64, 1.05) 0.89 (0.66, 1.20)
Other cardiovascular disease 0.72 (0.54, 0.96) 0.75 (0.54, 1.06)
All cancer mortality 0.83 (0.70, 0.99) 0.92 (0.75, 1.12)
All other mortality 0.74 (0.64, 0.84) 0.83 (0.71, 0.97)
Reprinted with permission from Jacobs et al.58 J Am Coll Nutr. 2000;19(Suppl 3):S326 –S330

Downloaded from https://academic.oup.com/nutritionreviews/article/65/10/439/1879933 by guest on 11 September 2020

glucosinolates and purified sulforaphane, even though tripled in the men consuming tomato sauce. Oxidative
the broccoli with intact glucosinolates contained far less DNA damage in leukocytes and prostate tissues was
sulphoraphane than was given in the purified sulfora- significantly diminished, with the latter observed mainly
phane form. A second experiment had similar results. in the tumor cell nuclei. There was a decrease in blood
The authors concluded that broccoli with intact glucosi- prostate-specific antigen, secondary to an increase in
nolates was more potent in enhancing antioxidant activ- apoptotic death of prostate cells, especially in the carci-
ity than sulforaphane or broccoli with sulforaphane noma regions. The research did not examine lycopene
formed in situ during laboratory hydrolysis. In a similar alone, but prostate cancer cell cultures were sensitive to
study with F344 rats, Wallig et al.64 confirmed synergis- lycopene in growth medium (increased apoptosis and
tic effects, but with some qualifications. They fed rats arrested cell cycle). The authors considered a direct
two supplemental glucosinolate derivatives found in effect of lycopene or synergy with other constituents as
broccoli, namely indole-3-carbinol (I3C) and crambene explanatory for the apparently beneficial effects of the
(1-cyano-2-hydroxy-3-butene). The high-dose supple- tomato sauce consumption. In the argument for synergy,
ments protected against precarcinogenic liver changes they considered important an animal study by Boileau et
induced by aflatoxin B1 and the high-dose combination al.,66 in which drug-induced prostate cancer mortality
group had liver scores close to the negative controls. The was 62% for those consuming a tomato powder with
low-dose group, which represented a realistic human 13 mg lycopene/kg versus 72% for those consuming
intake from diet, had no effect. Thus, synergy was lycopene beadlets with 161 mg/kg versus 80% for con-
demonstrated, but the authors were of the opinion that trol animals. Both of these studies expose the way in
the combination treatment did not provide the effect
which both the active ingredient and the food source as
expected from cruciferous vegetable intake. Clearly there
a whole need to be considered in exposing health
is more to be known and feeding trials in human subjects
benefits.
are required.
Another way of looking at the food synergy issue is
One relevant area has concerned the potential pro-
to consider negative effects from food, for example,
tective effects of tomatoes and lycopene on prostate
whether peanut allergy might be attributed to whole
health. In one study, pasta with tomato sauce was con-
peanuts or a toxic component. In this case, van Wijk et
sumed daily by 32 men for 3 weeks before scheduled
al.67 focused on several measures of immunogenicity of
prostatectomy (adenocarcinoma), with a convenience
four presumed allergens and the influence of the food
sample of seven non-consumers acting as controls.65
Blood lycopene levels doubled and prostate lycopene matrix. The purified peanut allergens were Ara h 1, Ara
h 2, Ara h 3, and Ara h 6; the other two treatments were
a whole peanut extract with and without peanut oil. None
Table 4. Percent inhibition of cancer cell lines of the allergens induced significant footpad immune
using apple extract (50 mg/ml), with and without activation. In contrast, both whole-peanut extract treat-
skin
ments induced an increase in cell number, cytokine
Inhibition of proliferation production, and activation of antigen-presenting cells. In
of human cancer cell lines*
an oral model of peanut hypersensitivity using C3H/
Cancer type With skin Without skin HeOuJ mice, oral exposure to the purified allergens in
Colon 43% 29% the presence of cholera toxin induced specific IgE re-
Liver 57% 40% sponses, regardless of the presence of a food matrix (that
*5.7 mg vitamin C did not affect proliferation is, the purified proteins were recognized as allergens),
Reprinted with permission from Eberhardt et al.59 Nature. but they did not elicit immune stimulation without the
2000;405:903–904 food matrix. Here, the relationship between the food

Nutrition Reviews姞, Vol. 65, No. 10 445

component and the whole food itself in manifesting a diet and health in 1968 recommended a low-fat diet for
negative health outcome remains consistent with the Americans, thus beginning an era without regard to
general theory of food synergy and the need to consider either differential fat type or phytochemical content or
whole foods to understand the biologic response. type of carbohydrate in the low-fat foods. It was a period
Moving from a single food to food combinations, of extreme weight gain in the general population. Even
Liu et al.68 studied the in vitro synergy of several fruits though it was known that different fat types raise or
in antioxidant activity. The combination of orange, ap- lower cholesterol, the total fat message was felt to be
ple, grape, and blueberry displayed a synergistic effect in satisfactory because it was easy to understand and im-
antioxidant activity. The median effective dose (EC50, plement. Large-scale intervention studies, however, do
the dose exhibiting 50% of total antioxidant activity) of not support the premise. For example, in the Women’s
the combination of fruits was one-fifth of the EC50 of Health Initiative trial,24 a low-fat diet was advised to
each fruit alone, suggesting synergistic effects (Fig-

Downloaded from https://academic.oup.com/nutritionreviews/article/65/10/439/1879933 by guest on 11 September 2020

19,541 free-living women and no such advice was given
ure 1). to 29,294 women; both groups were followed for 8.1
The food synergy concept gains credence when years. By year 6, mean fat intake had decreased by 8.2%
effects observed under laboratory conditions can also be of energy (saturated 2.9%, monounsaturated 3.3%, and
seen in the everyday use of food, such as in salad polyunsaturated 1.5% fat). This was about half the in-
preparation. Ninfali et al.69 reported that when marjoram tended fat reduction. There were modest increases in
is added to salad it can increase the antioxidant capacity consumption of fruit and vegetables (1.1 serving/d) and
of the dish by 200%, taking our appreciation of the whole grains (0.2 serving/d). Low-density lipoprotein
interactions of biochemicals in food to another, and cholesterol levels, diastolic blood pressure, and factor
perhaps more fundamental, level. This might explain, in VIIc levels were significantly, but not substantially, re-
part, the apparent superior effects of certain cuisines and duced by 3.55 mg/dL, 0.31 mm Hg, and 4.29%. There
cultural dishes that involve combinations of foods, in was no change in the level of HDL cholesterol, triglyc-
particular herbs and spices. As we explore the effects of erides, glucose, or insulin. No difference was seen be-
individual components and build the evidence base,70 the tween randomized fat intake groups in the incidence of
next step will be to examine the combinations of foods coronary heart disease, stroke, and cardiovascular dis-
under experimental conditions in the context of health ease. There were suggestive trends toward greater reduc-
and disease, particularly in view of current epidemiolog- tions in coronary heart disease risk in those whose
ical evidence.
low-fat diet was achieved by lower intakes of saturated
fat or trans fat or higher intakes of fruit and vegetables.
OUTCOMES FROM INTERVENTIONS This and other studies point to the danger of focusing on
a nutrient without considering the foods from which that
Just as the case of peanut allergy supported the food nutrient is obtained in the diet.
synergy argument by negative example in the food com- Similar concepts are now emerging from interven-
positional studies, so too can an examination of nutrition tions targeting vitamin supplements. The ␤-carotene in-
interventions based on single food components. Accord- terventions drew on inferences of lung cancer risk, most
ing to Taubes,71,72 the Senate McGovern Commission on notably from food intake summarized by its ␤-carotene
content, then extrapolated to doses higher than those
typically obtained from foods.73 This observation was
confirmed by others and trials were undertaken to show
the disease benefits of ␤-carotene and other isolated
compounds taken as supplements. Eight large random-
ized clinical trials have been conducted,6,15-23 two of
which comprised participants who were mostly smok-
ers6,15 (because circulating ␤-carotene is known to be
lower in smokers). We pooled the trials using meta-
analytic techniques. The RR for the eight trials pooled
(treatment group that included ␤-carotene vs. treatment
group that did not include ␤-carotene), was 1.03 (95%CI,
0.97–1.08). In the pooled trials of smokers, the corre-
sponding RR was 1.09 (95%CI, 1.02–1.16), while in the
Figure 1. The synergistic effect of the combination of orange,
apple, grape, and blueberry on total antioxidant activity. Re- pool of six trials16-23 in which smokers were infrequent,
printed with permission from Liu68 J Nutr. 2004;134(Suppl the RR was 1.00 (95%CI, 0.94–1.06). One of the six
12):S3479 –S3485, American Society for Nutrition trials in which there were few smokers was conducted in

446 Nutrition Reviews姞, Vol. 65, No. 10

Linxian, China16-18 and did show significant heterogene- CONCLUSION
ity from the other studies with benefit for the ␤-carotene
supplement group and a RR of 0.91 (95%CI, 0.84–0.99). A nutrition perspective that focuses on food compo-
Thus, supplementation with purified ␤-carotene in clin- nents has been successful in improving public health in a
ical trials showed little evidence of benefit and some number of cases, from identifying the cause of deficiency
possible evidence of harm, conceivably enhanced in diseases such as scurvy or pellagra to finding that folate
smokers, an effect that seriously challenged thinking in has a special etiologic role in congenital birth defects.
terms of nutrition intervention. Other instances of single-nutrient solutions to health
This challenge has been fueled further with other problems exist, such as nutrient supplementation during
vitamin supplement trials. Two recent randomized clin- pregnancy, B12 supplementation in vegetarians and the
ical trials of B vitamins failed to reduce cardiovascular elderly, and iron supplementation in iron deficiency ane-

Downloaded from https://academic.oup.com/nutritionreviews/article/65/10/439/1879933 by guest on 11 September 2020

disease occurrence or recurrence in participants with mia. However, these conditions represent only one do-
recent acute myocardial infarction, preexisting vascular main where nutrition intervention is required, and this
disease, or diabetes. The NORVIT trial25 examined 3749 domain may be characterized as conditions of relative
men and women with acute myocardial infarction within deficiency (whether through dietary restriction, the phys-
7 days before randomization in a two-by-two factorial iologic needs of pregnancy, or pathophysiological con-
design (0.8 mg folic acid and 0.4 mg B12, 40 mg B6, ditions). As the nutrition discipline develops and broad-
both treatments, and placebo). Median follow-up was 40 ens its horizons, however, we argue for a need to return
months and the outcome variable was a composite of to food as the source of nutrition knowledge and thereby
recurrent myocardial infarction, stroke, and sudden car- its applications.
diac death. The mean total homocysteine level was This review has illustrated that it is the combination
lowered by 27% among participants given folic acid plus of foods within diet patterns and of food components in
vitamin B12. However, outcome risk ratios were not single foods (food synergy) that is often etiologic in diet
significantly different from 1 compared to placebo treat- and health relationships, including relations with many
ment [folic acid and B12: 1.08 (95%CI, 0.93–1.25), P ⫽ chronic diseases and metabolic pathways. While being
0.31; B6: 1.14 (95%CI, 0.98 –1.32), P ⫽ 0.09] or signif- sympathetic to attempts to map out detailed causes, we
icantly adverse in the combined treatment group [folic believe that the complexity of metabolism and pathology
acid, B12, B6: 1.22 (95%CI, 1.00 –1.50), P ⫽ 0.05]. The is such that nutrition research is a long way from being
Heart Outcomes Prevention Evaluation (HOPE) 2 trial26 able to do such mapping accurately and completely. The
examined 5522 patients who were ⱖ55 years old with current amount of food-based research in humans is far
vascular disease or diabetes. Treatment arms were a lower than that building up in the areas of basic science
combination of folic acid (2.5 mg), B12 (1 mg), and B6 of cellular metabolism or of pharmaceuticals, for exam-
(50 mg) versus placebo and the average follow-up was 5 ple. At minimum, to achieve a clear understanding of
years. Mean plasma homocysteine levels decreased by diet and health, whole foods should be studied in parallel
about 20% in the treated participants. Relative risk for with food components suspected of having important
the primary outcome (i.e., death from cardiovascular biological activity. The context of individual foods in
causes, myocardial infarction, and stroke) was 0.95 whole diets must also be considered. A comprehensive
(95%CI, 0.84 –1.07), P ⫽ 0.41. For stroke, a statistically knowledge of food and the relationships of food patterns
significant benefit was indicated [RR 0.75 (95%CI, to health, given existing knowledge of food components,
0.59 – 0.97), P ⫽ 0.03], but there was significant detri- would be extremely helpful. If the enterprise is limited to
ment for hospitalized unstable angina [RR 1.24 (95%CI, research on food components alone, there is a risk of
1.04 –1.49), P ⫽ 0.02]. Although a meta-analysis of producing false answers because of the failure to account
homocysteine levels and MTHFR found increased risk of for synergy in foods and whole diets. In fact, it is
venous thrombosis for higher homocysteine,74 a third arguable that experiences with population-wide total fat
randomized clinical trial, the Dutch Vitamins and limitation, fad diets based on total carbohydrates, and
Thrombosis study (VITRO) failed to show an effect of widespread supplementation with ␤-carotene, B vita-
supplemental B vitamins in the secondary prevention of mins, and other substances should be called misadven-
venous thrombosis.27 In contrast, in support of the food tures that have not helped the population at large.
synergy theory, a recent prospective study found that Public policy based on simple nutrient formulations
increased intake of fruit, vegetables, and fish (which had can be successful if targeted toward specific pathways,
a higher B-complex content in combination) was associ- such as exposing non-pregnant women of child-bearing
ated with reduced incidence of idiopathic venous throm- age to increased levels of folate in order to reduce
bosis.75 congenital birth defects, B12 supplementation in vege-

Nutrition Reviews姞, Vol. 65, No. 10 447

tarians, and iron supplements for iron deficiency anemia. ACKNOWLEDGMENTS
In these cases, the delivery agent may be fortified foods
or pharmaceutical preparations, but the desired outcomes David Jacobs gratefully acknowledges the assistance
are the same (as would be dose requirements and con- of Atsushi Hozawa and Sue Duval in the meta-analysis
sumption patterns); the approach can also be seen as reported here of clinical trials of ␤-carotene supplemen-
more pharmacological and therapeutic than dietary and tation. This work benefited greatly from conversation
health promoting. and collaboration with Lyn Steffen, Myron Gross, and
Whether this ongoing strategy is equally appropriate Rune Blomhoff. Both David Jacobs and Linda Tapsell
to the rest of the population, and to more complex targets work in unpaid positions with a number of other scien-
that relate to multiple nutritional pathways, such as the tists through the science program of the California Wal-
nut Commission, where they regularly engage in discus-
prevention of diabetes, heart disease, and cancer, needs
sion on issues related to this manuscript.

Downloaded from https://academic.oup.com/nutritionreviews/article/65/10/439/1879933 by guest on 11 September 2020

to be questioned. This is where a full appreciation of the
role of food in nutrition and health is required. The
REFERENCES
␤-carotene and B-vitamin clinical trials lead one to
question a strategy of pursuing pharmacological ap- 1. Lind J. A treatise of the scurvy. In three parts.
Containing an inquiry into the nature, causes and
proaches to nutrition interventions in the area of chronic
cure, of that disease. Together with a critical and
diseases related to lifestyle. Nutritional composition chronological view of what has been published on
should be seen as a base in food-oriented research to the subject. Edinburgh: Printed by Sands, Murray
study the effects that naturally occurring food synergy and Cochran for A Kincaid and A Donaldson; 1753.
provides. The development of new foods could then 2. Tröhler U. James Lind and scurvy: 1747 to
1795. Available at: http://www.jameslindlibrary.
draw on this integrated model rather than on a single- org/trial_records/17th_18th_Century/lind/lind_1753_
nutrient model of nutritional benefits. commentary.html. Accessed 14 August 2007.
In the past and in the foreseeable future, at least, 3. Carpenter KJ. The History of Scurvy and Vitamin C.
food will remain fundamental to health. The history of Cambridge: Cambridge University Press; 1986:423.
4. Manela-Azulay M, Mandarim-de-Lacerda CA, de
Western nutrition science has shown how knowledge of Andrade Perez M, Filgueira AL, Cuzzi T. Vitamin C.
food components has an effective utility in supporting Anais Brasileiros Dermatologia 2003;78:265–274.
health where deficiencies and excesses occur. The iden- 5. Hegsted DM. A look back at lessons learned and
tification and purification of food components to effec- not learned. J Nutr. 1994;124(9 Suppl 9):S1867–
S1870.
tively produce pharmacological agents has led to thera-
6. The Alpha-Tocopherol, Beta Carotene Cancer Pre-
pies for the successful management of these deficiencies vention Study Group. The effect of vitamin E and
(or the consequences of potential deficiencies). Changing beta carotene on the incidence of lung cancer and
the delivery agent from pills to fortified foods does not other cancers in male smokers. N Engl J Med.
necessarily change the treatment strategy and the com- 1994;330:1029 –1035.
7. Jacobs DR, Gallaher DD. Whole grain intake and
pliance behaviors required for consumption. However, cardiovascular disease: a review. Curr Atheroscler
this approach represents just one of a multitude of path- Rep. 2004;6:415– 423.
ways for the development of nutrition science and prac- 8. Kelly JH Jr, Sabate J. Nuts and coronary heart
tice. In addition, focusing on single food components, disease: an epidemiological perspective. Br J Nutr.
2006;96(Suppl 2):S61–S67.
such as total fat because it is related to saturated fat, to 9. Estruch R, Martinez-Gonzalez MA, Corella D, et al.
address the negative effects of whole diets has proven PREDIMED Study Investigators. Effects of a Medi-
problematic. An understanding of the interactions be- terranean-style diet on cardiovascular risk factors: a
tween food components in both single foods and whole randomized trial. Ann Intern Med. 2006;145:1–11.
10. He K, Song Y, Daviglus ML, Liu K, Van Horn L, Dyer
diets opens up new areas of thinking that appear to have
AR, Greenland P. Accumulated evidence on fish
greater application to contemporary population health consumption and coronary heart disease mortality:
issues, particularly those related to chronic lifestyle dis- a meta-analysis of cohort studies. Circulation. 2004;
ease. In this context, it is necessary to appreciate the 109:2705–2711.
systems biology of the food sources (plant and animal) in 11. He K, Song Y, Daviglus ML, Liu K, Van Horn L, Dyer
AR, Goldbourt U, Greenland P. Fish consumption
relation to human biology, as well as the effects of food and incidence of stroke: a meta-analysis of cohort
processing and the impact of cuisine. It is this new studies. Stroke. 2004;35:1538 –1542.
understanding that reminds us emphatically of the central 12. Schulze MB, Hoffmann K. Methodological ap-
position of food in the nutrition-health interface, which proaches to study dietary patterns in relation to risk
of coronary heart disease and stroke. Br J Nutr.
begs for much more whole food-based research, and 2006;95:860 – 869.
which encourages us in both etiologic research and 13. Kant AK. Dietary patterns and health outcomes.
dietary advice to “think food first”. J Am Diet Assoc. 2004;104:615– 635.

448 Nutrition Reviews姞, Vol. 65, No. 10

14. de Lorgeril M, Salen P, Martin JL, Monjaud I, Delaye E, Berry B, Jacobs DR. Reductionism and the nar-
J, Mamelle N. Mediterranean diet, traditional risk rowing nutrition perspective: time for reevaluation
factors, and the rate of cardiovascular complica- and emphasis on food synergy. J Am Diet Assoc.
tions after myocardial infarction: final report of the 2001;101:1416 –1419.
Lyon Diet Heart Study. Circulation. 1999;99:779 – 29. Jacobs DR, Steffen LM. Nutrients, foods, and di-
785. etary patterns as exposures in research: a frame-
15. Omenn GS, Goodman GE, Thornquist MD, et al. work for food synergy. Am J Clin Nutr. 2003;
Effects of a combination of beta carotene and vita- 78(Suppl 3):S508 –S513.
min A on lung cancer and cardiovascular disease. 30. Liu RH. Health benefits of fruit and vegetables are
N Engl J Med. 1996;334:1150 –1155. from additive and synergistic combinations of phy-
16. Blot WJ, Li J-Y, Taylor PR, et al. Nutrition interven- tochemicals. Am J Clin Nutr. 2003;78(Suppl 3):
tion trials in Linxian, China. J Natl Cancer Inst. S517–S520.
1993;85:1483–1492. 31. Liu RH. Potential synergy of phytochemicals in can-
17. Blot WJ, Li JY, Taylor PR, Guo W, Dawsey SM, Li B.

Downloaded from https://academic.oup.com/nutritionreviews/article/65/10/439/1879933 by guest on 11 September 2020

cer prevention: mechanism of action. J Nutr. 2004;
The Linxian trials: mortality rates by vitamin-mineral 134(Suppl 12):S3479 –S3485.
intervention group. Am J Clin Nutr. 1995;62(Suppl 32. Appel LJ, Moore TJ, Obarzanek E, Vollmer WM,
6):1424S–1426S. Svetkey LP, Sacks FM, Bray GA, Vogt TM, Cutler
18. Blot WJ, Li JY, Taylor PR, Li B. Lung cancer and JA, Windhauser MM, Lin PH, Karanja N. A clinical
vitamin supplementation. N Engl J Med. 1994;331: trial of the effects of dietary patterns on blood
614. pressure. DASH Collaborative Research Group.
19. Hennekens CH, Buring JE, Manson JE, et al. Lack N Engl J Med. 1997;336:1117–1124.
of effect of long-term supplementation with beta 33. Hu FB, Rimm EB, Stampfer MJ, Ascherio A,
carotene on the incidence of malignant neoplasms Spiegelman D, Willett WC. Prospective study of
and cardiovascular disease. N Engl J Med. 1996; major dietary patterns and risk of coronary heart
334:1145–1149. disease in men. Am J Clin Nutr. 2000;72:912–921.
20. Lee IM, Cook NR, Manson JE, Buring JE, Henne- 34. Chakravarti R, Sahai V. Compactin-a review. Appl
kens CH. Beta-carotene supplementation and inci- Microbiol Biotechnol. 2004;64:618 – 624.
dence of cancer and cardiovascular disease: the 35. Wintergreen. Available at: http://www.drugs.com/
Women’s Health Study. J Natl Cancer Inst. 1999;
npp/wintergreen.html. Accessed 14 August 2007.
91:2102–2106.
36. Weststrate JA, Meijer GW. Plant sterol-enriched
21. Heart Protection Study Collaborative Group. MRC/
margarines and reduction of plasma total- and LDL-
BHF Heart Protection Study of antioxidant vitamin
cholesterol concentrations in normocholesterol-
supplementation in 20,536 high-risk individuals: a
aemic and mildly hypercholesterolaemic subjects.
randomised placebo-controlled trial. Lancet. 2002;
Eur J Clin Nutr. 1998;52:334 –343.
360:23–33.
37. Aggett PJ, Antoine J-M, Asp N-G, et al. Process for
22. Greenberg ER, Baron JA, Karagas MR, et al. Mor-
the assessment of scientific support for claims on
tality associated with low plasma concentration of
foods (PASSCLAIM), Consensus on criteria. Eur
beta carotene and the effect of oral supplementa-
tion. JAMA. 1996;275:699 –703. J Clin Nutr. 2005;44(Suppl 1):15–30.
23. Age-Related Eye Disease Study Research Group. A 38. United States Food and Drug Administration. Guid-
randomized, placebo-controlled, clinical trial of ance for industry. Significant scientific agreement in
high-dose supplementation with vitamins C and E the review of health claims for conventional foods
and beta carotene for age-related cataract and vi- and dietary supplements. http://vm.cfsan.fda.gov/
sion loss: AREDS report no. 9. Arch Ophthalmol. ⬃dms/ssaguide.html. Accessed 14 August 2007.
2001;119:1439 –1452. 39. Richardson DP. The scientific substantiation of
24. Howard BV, Van Horn L, Hsia J, et al. Low-fat health claims with particular reference to the grad-
dietary pattern and risk of cardiovascular disease: ing of evidence and consumer understanding. Food
the Women’s Health Initiative Randomized Con- Science and Technology Bulletin: Functional Foods.
trolled Dietary Modification Trial. JAMA. 2006;295: 2005;2:39 – 48.
655– 666. 40. Wilson T, Jacobs DR. Functional foods: a critical
25. Bonaa KH, Njolstad I, Ueland PM, et al. NORVIT appraisal. In: Temple NJ, Wilson T, and Jacobs DR,
Trial Investigators. Homocysteine lowering and car- eds, Nutritional Health: Strategies for Disease Pre-
diovascular events after acute myocardial infarc- vention, 2nd ed. Totowa, NJ, Humana Press; 2006:
tion. N Engl J Med. 2006;354:1578 –1588. 452.
26. Lonn E, Yusuf S, Arnold MJ, et al. Heart Outcomes 41. Keys AB, Keys M. Eat Well and Stay Well. Garden
Prevention Evaluation (HOPE) 2 Investigators. Ho- City, NY, Doubleday; 1959:394.
mocysteine lowering with folic acid and B vitamins 42. Keys M, Keys AB. The Benevolent Bean. Garden
in vascular disease. N Engl J Med. 2006;354:1567– City, NY, Doubleday; 1967:192.
1577. 43. Keys AB, Keys M. How to Eat Well and Stay Well the
27. den Heijer M, Willems HP, Blom HJ, et al. Homo- Mediterranean Way. Garden City, NY, Doubleday;
cysteine lowering by B vitamins and the secondary 1975:488.
prevention of deep-vein thrombosis and pulmonary 44. Keys A, Anderson JT, Grande F. Serum cholesterol
embolism. A randomized, placebo-controlled, dou- response in the diet. I. Iodine value of dietary fat
ble blind trial. Blood. 2006;109:139 –144. versus 2S-P. Metabolism. 1965;14:747–758.
28. Messina M, Lampe JW, Birt DF, Appel LJ, Pivonka 45. Keys A, Anderson JT, Grande F. Serum cholesterol

Nutrition Reviews姞, Vol. 65, No. 10 449

 response in the diet. II. The effect of cholesterol in tumors in rats. J Agric Food Chem. 2005;53:2341–
the diet. Metabolism. 1965;14,:759 –765. 2343.
46. Keys A, Anderson JT, Grande F. Serum cholesterol 61. Seeram NP, Adams LS, Henning SM, Niu Y, Zhang
response to changes in the diet. III. Differences Y, Nair MG, Heber D. In vitro antiproliferative, apo-
among individuals. Metabolism. 1965;14:766 –775. ptotic and antioxidant activities of punicalagin, el-
47. Keys A, Anderson JT, Grande F. Serum cholesterol lagic acid and a total pomegranate tannin extract
response to changes in the diet. IV. Particular sat- are enhanced in combination with other polyphe-
urated fatty acids in the diet. Metabolism. 1965;14: nols as found in pomegranate juice. J Nutr Bio-
776 –787. chem. 2005;16:360 –367.
48. Fetcher ES, Foster N, Anderson JT, Grande F, Keys 62. Lansky EP. Beware of pomegranates bearing 40%
A. Quantitative estimation of diets to control serum ellagic acid. J Med Food. 2006;9:119 –122.
cholesterol. Am J Clin Nutr. 1967;20:475– 492. 63. Keck AS, Qiao Q, Jeffery EH. Food matrix effects on
49. Jacobs DR Jr, Anderson JT, Hannan P, Keys A, bioactivity of broccoli-derived sulforaphane in liver
Blackburn H. Variability in individual serum choles- and colon of F344 rats. J Agric Food Chem. 2003;

Downloaded from https://academic.oup.com/nutritionreviews/article/65/10/439/1879933 by guest on 11 September 2020

terol response to change in diet. Arteriosclerosis. 51:3320 –3327.
1983;3:349 –356. 64. Wallig MA, Heinz-Taheny KM, Epps DL, Gossman
50. Grande F, Anderson JT, Keys A. Serum cholesterol T. Synergy among phytochemicals within crucifers:
in man and the unsaponifiable fraction of corn oil in does it translate into chemoprotection? J Nutr.
the diet. Proc Soc Exp Biol Med. 1958;98:436 – 440. 2005;135(Suppl 12):S2972–S2977.
51. Keys A, Grande F, Anderson JT. Fiber and pectin in 65. Stacewicz-Sapuntzakis M, Bowen PE. Role of lyco-
the diet and serum cholesterol concentration in pene and tomato products in prostate health. Bio-
man. Proc Soc Exp Biol Med. 1961;106:555–558. chim Biophys Acta. 2005;1740:202–205.
52. Jenkins DJ, Kendall CW, Faulkner D, et al. A dietary 66. Boileau TW, Liao Z, Kim S, Lemeshow S, Erdman
portfolio approach to cholesterol reduction: com- JW Jr, Clinton SK. Prostate carcinogenesis in N-
bined effects of plant sterols, vegetable proteins, methyl-N-nitrosourea (NMU)-testosterone-treated
and viscous fibers in hypercholesterolemia. Metab- rats fed tomato powder, lycopene, or energy-re-
olism. 2002;51:1596 –1604. stricted diets. J Natl Cancer Inst. 2003;95:1578 –
53. Jenkins DJ, Kendall CW, Marchie A, et al. Effects of 1586.
a dietary portfolio of cholesterol-lowering foods vs 67. van Wijk F, Nierkens S, Hassing I, et al. The effect of
lovastatin on serum lipids and C-reactive protein. the food matrix on in vivo immune responses to
JAMA. 2003;290:502–510. purified peanut allergens. Toxicol Sci. 2005;86:333–
54. Gardner CD, Coulston A, Chatterjee L, Rigby A, 341.
Spiller G, Farquhar JW. The effect of a plant-based 68. Liu RH. Potential synergy of phytochemicals in can-
diet on plasma lipids in hypercholesterolemic cer prevention: mechanism of action. J Nutr. 2004;
adults: a randomized trial. Ann Intern Med. 2005; 134(Suppl 12):S3479 –S3485.
142:725–733. 69. Ninfali P, Mea G, Giorgini S, Rocchi M, Bacchiocca
55. Bruce B, Spiller GA, Klevay LM, Gallagher SK. A diet M. Antioxidant capacity of vegetables, spices and
high in whole and unrefined foods favorably alters dressings relevant to nutrition. Br J Nutr. 2005;93:
lipids, antioxidant defenses, and colon function. 257–266.
J Am Coll Nutr. 2000;19:61– 67. 70. Tapsell LC, Hemphill I, Cobiac L, et al. Health ben-
56. Tapsell LC, Gillen LJ, Patch CS, Batterham M, efits of herbs and spices: the past, the present, the
Owen A, Bare M, Kennedy M. Including walnuts in a future. Med J Aust. 2006;185(Suppl 4):S1–S24.
low-fat/modified-fat diet improves HDL cholesterol- 71. Taubes G. Nutrition. What if Americans ate less
to-total cholesterol ratios in patients with type 2 saturated fat? Science. 2001;291:2538.
diabetes. Diabetes Care. 2004;27:2777–2783. 72. Taubes G. Nutrition. The soft science of dietary fat.
57. Gillen LJ, Tapsell LC, Patch CS, Owen A, Batterham Science. 2001;291:2536 –2545.
M. Structured dietary advice incorporating walnuts 73. Shekelle RB, Lepper M, Liu S, et al. Dietary vitamin
achieves optimal fat and energy balance in patients A and risk of cancer in the Western Electric study.
with type 2 diabetes mellitus. J Am Diet Assoc. Lancet. 1981;2:1185–1190.
2005;105:1087–1096. 74. Den Heijer M, Lewington S, Clarke R. Homocys-
58. Jacobs DR, Pereira MA, Meyer KA, Kushi LH. Fiber teine, MTHFR and risk of venous thrombosis: a
from whole grains, but not refined grains, is in- meta-analysis of published epidemiological studies.
versely associated with all-cause mortality in older J Thromb Haemost. 2005;3:292–299.
women: the Iowa women’s health study. J Am Coll 75. Steffen LM, Folsom AR, Cushman M, Jacobs DR,
Nutr. 2000;19(Suppl 3):S326 –S330. Rosamond WD. Greater fish, fruit, and vegetable
59. Eberhardt MV, Lee CY, Liu RH. Antioxidant activity intakes are related to lower incidence of venous
of fresh apples. Nature. 2000;405:903–904. thromboembolism: The LITE Study. Circulation.
60. Liu RH, Liu J, Chen B. Apples prevent mammary 2007;115:188 –195.

450 Nutrition Reviews姞, Vol. 65, No. 10