Caballero B. (ed.) Encyclopaedia of Food Science, Food Technology and Nutrition. Ten-Volume Set

Подождите немного. Документ загружается.

food item or nutrient as promoter, the relative risk

associated with the nutritional factor increases

with the prevalence or the abundance of the

unknown initiator.

0010 When many nutritional factors are examined in the

context of a particular study, multiple comparisons

can generate both false-positive and false-negative

results.

0011 Food patterns, for instance the Mediterranean diet,

may have effects that are considerably different to

those of the constituent foods or nutrients because

of complex interactions.

0012 Interindividual variability of certain food items

or nutrients may be too small in comparison with

the corresponding intraindividual variability, thus

considerably reducing the power of analytical epi-

demiological studies. Indeed, in many populations,

the diet may be too uniform and etiologically

important food items may be too common or too

rare to allow statistical substantiation of risk dif-

ferentials with standard analytic epidemiological

studies.

0013 Energy intake may be an important cause, an

equally important confounder or an indicator of

bias, particularly in case-control studies. Control

of energy intake is thus necessary whenever par-

ticular foods or nutrients are evaluated.

Epidemiological Evidence on the Nutritional

Etiology of Cancer

0014 The effect of diet on cancer occurrence is poorly

understood in molecular or mechanistic terms,

although it is generally assumed that dietary factors

act at a late stage in the long process of carcino-

genesis. Several patterns have emerged, however,

from the many epidemiological studies that have

been undertaken during the last 30 years on the rela-

tion between diet and cancer, and these are summar-

ized in Tables 1–4. The associations are distinguished

into convincing, probable, and possible, and can be

either positive (increasing intake of the dietary factor

increases cancer risk) or inverse (increasing intake of

the dietary factor reduces cancer risk).

0015Table 1 refers to the cancer risk implications in

relation to the intake of major food groups. Vege-

tables and, to a slightly lesser extent, fruits are in-

versely associated with risk for several forms of

cancer. There is also evidence that intake of red

meat increases the risk for some forms of cancer,

particularly cancer of the large bowel.

0016The collective evidence for macronutrients, as sum-

marized in Table 2, is considerably weaker for obvi-

ous reasons. In the absence of long-term randomized

intervention studies, the evidence concerning macro-

nutrients is derived from the evidence concerning

food groups with additional uncertainties introduced

by variability in food composition tables and analyt-

ical methods, as well as by uncontrollable confound-

ing due to the possible operation of unidentifiable or

immeasurable factors that tend to coexist with some

of the macronutrients.

0017Table 3 summarizes the risk implications for

major forms of cancer by intake of selected micro-

nutrients. Although the study of micronutrients

shares the same constrains as that of macronutrients,

tbl0001 Table 1 Risk implications for major forms of cancer by consumption of major food groups

Cancer site Cereals Pulses Vegetables Fruits Meat (red) Fish Dairies Eggs Sugars

Mouth/pharynx ### ###

Nasopharynx

Esophagus ### ###

Stomach ### ###

Large bowel ## # "" " "

Liver #

Gall-bladder

Pancreas ##"

Larynx ## ##

Lung ### ##

Breast ## ##

Endometrium ##

Cervix uteri ##

Ovary ##

Prostate #""

Urinary bladder ## ##

Kidney #"

###, convincing inverse association; """, convincing positive association; ##, probable inverse association; "", probable positive association; #, possible

inverse association; ", possible positive association.

796 CANCER/Epidemiology

this table appears more informative than Table 2 for

three reasons, two of which are purely artificial. First,

of the many micronutrients, only those for which

some evidence exists are included. Second, several

carotenoids are included in the single column dedi-

cated to these compounds. Thus, the evidence for

lung cancer mostly refers to b-carotene, whereas

the evidence for prostate cancer refers almost exclu-

sively to lycopene. The third reason, which is more

substantive, is that the physiologic and pathophysio-

logic effects of certain micronutrients have been

extensively studied, so that the criteria of biologic

plausibility can be more successfully relied upon for

micronutrients rather than for macronutrients.

For instance, several carotenoids, vitamins C and E,

and selenium have been intensively studied on

account of their experimentally documented antioxi-

dant potential.

0018Table 4 refers to nonnutrients and nutritional cor-

relates like height, obesity, and physical activity.

Entries in the same column do not imply identical

products or processes. For instance, alcohol causes

liver cancer through cirrhosis and cancer of the

esophagus and larynx, mostly through interaction

with tobacco smoking, and cancer of the breast

possibly by increasing levels of estrogens. Moreover,

tbl0002 Table 2 Risk implications for major forms of cancer by consumption of macronutrients

Cancer site Protein Starch Fiber Saturated lipids Monounsaturated lipids Polyunsaturated lipids

Mouth/pharynx

Nasopharynx

Esophagus

Stomach "

Large bowel "#"

Liver

Gall-bladder

Pancreas ""#

Larynx

Lung "

Breast #

Endometrium #"

Cervix uteri

Ovary

Prostate "

Urinary bladder

Kidney

###, convincing inverse association; """, convincing positive association; ##, probable inverse association; "", probable positive association; #, possible

inverse association; ", possible positive association.

tbl0003 Table 3 Risk implications for major forms of cancer by consumption of selected micronutrients

Cancer site Vitamin A and carotenoids Folate Vitamin C Vitamin D Vitamin E Calcium Selenium

Mouth/pharynx #

Nasopharynx

Esophagus ##

Stomach ###

Large bowel #####

Liver

Gall-bladder

Pancreas #

Larynx

Lung ####

Breast #

Endometrium

Cervix uteri ###

Ovary

Prostate ## # #

Urinary bladder

Kidney

###, convincing inverse association; """, convincing positive association; ##, probable inverse association; "", probable positive association; #, possible

inverse association; ", possible positive association.

CANCER/Epidemiology 797

coffee and mate are mixtures that have both

similarities and differences. Lastly, the effect of

obesity may refer to stages of carcinogenesis that are

not necessarily identical in the various forms of

cancer. At the extreme, obesity among premenopau-

sal women reduces the risk of breast cancer, whereas

it considerably increases this risk among postmeno-

pausal women.

Additives and Contaminants

0019 Some readers may be surprised by the absence of a

table referring to additives and contaminants that are

frequently reported as carcinogens in the lay press.

There are several reasons for this. First, the intake of

additives and contaminants cannot be documented

in most epidemiological research. Second, additives

and contaminants are extensively scrutinized in

laboratory assays, and whenever evidence incrimin-

ates them as carcinogens, these are removed from

the diet. Third, the alleged breast carcinogenicity of

organochlorine compounds, which can be measured

in the blood or adipose tissue, has not been docu-

mented in large and sophisticated epidemiological

investigations; nor has there been any direct epi-

demiological support for a role of nitrosamines in

human carcinogenesis. At this stage, the only additive

that is likely to play a major role in human carcino-

genesis is salt. Among dietary contaminants, only

aflatoxin has been conclusively linked to hepatocel-

lular carcinoma and may be causing a nonnegligible

fraction of liver cancer in developing countries.

The suspicion that chlorination byproducts in

drinking water may be involved in urinary bladder

carcinogenesis has not been documented but is cur-

rently under investigation.

Prevention of Diet-related Cancers

0020On the basis of studies in migrants and ecologic

investigations, it has been estimated that diet may

be responsible for up to 30% of cancer mortality in

developed countries. Although we know little about

the mechanisms of diet-related carcinogenesis, we

have a very good idea of what a prudent diet that

reduces cancer in humans should be. This diet should

be high in vegetables, fruits, pulses, and cereals, low

in red meat and salt, and low in saturated fat of

animal origin. Added lipids should be of plant origin,

and, among them, olive oil has a safety record of

several thousand years. Obesity should be avoided,

preferably by increasing physical activity, which, in

itself, can reduce the incidence of colorectal cancer

and, perhaps, other forms of cancer as well. Regular

physical activity during childhood and adolescence

may also slow down excessive growth, as reflected

in attained height, and may have beneficial conse-

quences on several cancer types. Alcohol, which is

known to protect against cardiovascular diseases,

should be consumed in moderation, particularly by

women, because it contributes to the causation of

cancer at several sites, including the breast.

See also: Aflatoxins; Alcohol: Metabolism, Beneficial

Effects, and Toxicology; Antioxidants: Natural

Antioxidants; Cancer: Carcinogens in the Food Chain;

Carcinogens: Carcinogenic Substances in Food:

Mechanisms; Colon: Cancer of the Colon;

Contamination of Food

tbl0004 Table 4 Risk implications for major forms of cancer by exposure to selected nonnutrients and nutrition-related indicators

Cancer site Alcohol Salt Coffee Hot drinks Height Obesity Physicalactivity

Mouth/pharynx """ " (mate)

Nasopharynx ""

Esophagus """ "" (mate) ""

Stomach " """

Large bowel " " " ###

Liver """

Gall-bladder ""

Pancreas "

Larynx """

Lung

Breast "" """ ##/""" #

Endometrium """

Cervix uteri

Ovary

Prostate "" #

Urinary bladder "

Kidney ""

###, convincing inverse association; """, convincing positive association; ##, probable inverse association; "", probable positive association; #, possible

inverse association; ", possible positive association.

798 CANCER/Epidemiology

Further Reading

Lagiou P, Trichopoulou A and Trichopoulos D (2002) Nu-

tritional epidemiology of cancer: accomplishments and

prospects. Proceedings of the Nutrition Society. 61:

217–222.

MacMahon B and Trichopoulos D (1996) Epidemiology:

Principles and Methods, Second Edition. Boston: Little

Brown.

Willett W (1998) Nutritional Epidemiology, 2nd edn. New

York: Oxford University Press.

Willett WC and Trichopoulos D (eds) (1996) Nutrition and

cancer. Cancer Causes and Control 7: 3–180.

World Cancer Research Fund and American Institute for

Cancer Prevention (1997) Food, Nutrition and the Pre-

vention of Cancer: A Global Perspective. Washington,

DC: World Cancer Research Fund and American Insti-

tute for Cancer Prevention.

Carcinogens in the Food Chain

M J Prival, United States Food and Drug

Administration, Washington, DC, USA

This article is reproduced from Encyclopaedia of Food Science,

Background

0001 It has become increasingly clear in recent decades that

food can have a substantial effect on cancer risk. This

conclusion is based on the results of both human popu-

lation studies and animal studies. Presently available

evidence indicates that dietary fat may be a risk factor

for cancer and that fresh fruits and vegetables may have

a protective effect. In addition, the human diet contains

specific carcinogenic substances, most, but not all of

which are present at such low levels that their effects on

human cancer rates, if any, are likely to be small. This

article summarizes current information on carcinogens

in the diet that are (1) naturally occurring components

of food, (2) produced by the processing of food, or (3)

added intentionally to food. Carcinogenic contamin-

ants of food, including mycotoxins and pesticides, are

discussed elsewhere. (See Carcinogens:Carcinogenic

Substances in Food: Mechanisms; Carcinogenicity

Tests; Fatty Acids: Dietary Importance; Contamination

of Food; Mycotoxins: Occurrence and Determination.)

Naturally Occurring Components of Foods

Bracken Fern

0002 The toxic effects of bracken fern on livestock have

been observed for a century or more. Bracken fern is

not only toxic, but also carcinogenic when fed to rats

and other experimental animals at high levels.

Both the toxicity and the carcinogenicity are attrib-

uted to the same chemical constituent, ptaquiloside

(Figure 1).

0003Young bracken fronds in the fiddlehead or crosier

stage are used as a salad green in some countries,

including Japan. In Japan, bracken fern is boiled or

pickled before eating to remove the astringent taste;

this reduces, but does not eliminate, its carcinogen-

icity. Milk from cattle that were fed bracken fern has

been found to be carcinogenic to rats. This raises the

possibility that there may be a risk associated with

the drinking of milk in areas where dried bracken fern

is used for bedding in winter or where range-fed cattle

may consume bracken fern; this occurs, for example,

in Turkey and Bulgaria.

Pyrrolizidine Alkaloids

0004Pyrrolizidine alkaloids occur in a variety of plants

used as food and herbal medicines in many parts of

the world, particularly in Africa and Asia. The first

natural substances shown to cause liver tumors were

pyrrolizidine alkaloids from plants. Among the pyr-

rolizidine-alkaloid-containing plants used by humans

are Petasites japonicus Maxim (a kind of coltsfoot),

which is used as a food and herbal remedy in Japan,

and Symphytum officinale L. (comfrey), which is

widely used as a medicinal herb, salad ingredient, or

‘tea.’ Both of these plants have been shown to be

carcinogenic when fed to rats. Senkirkine (Figure 1),

which is found in P. japonicus M., induced liver

tumors when injected intraperitoneally into male

rats, as did symphytine (Figure 1), extracted from

roots of comfrey (S. officinale). Both alkaloids can

also induce mutations in the bacterium Salmonella

typhimurium; this is an indication of their ability to

react with deoxyribonucleic acid (DNA) and may be

relevant to the mechanism of their carcinogenicity.

(See Alkaloids: Toxicology.)

Mate

0005Mate (pronounced ‘mah-tay

’) is a beverage that is

widely consumed in southern Brazil, Paraguay,

Uruguay, and northern Argentina. It is made by

brewing the toasted, dried, and aged leaves of a par-

ticular type of tree, Ilex paraguariensis, which is indi-

genous to this area. Several studies have compared

mate consumption among patients with cancers of

the esophagus, larynx, oral cavity, or pharynx with

that of comparable individuals who did not have

these cancers. Taken as a whole, these studies indicate

that consumption of hot mate is probably carcino-

genic to the upper gastrointestinal tract. There is

currently no way of knowing to what extent the

CANCER/Carcinogens in the Food Chain 799

chemicals present in the mate contribute to its

apparent carcinogenicity and to what extent the

high temperature of the consumed beverage is the

problem. Experiments on the carcinogenicity of

mate on animals might help to resolve this question,

but unfortunately, no such experiments have been

reported.

Coffee

0006 The many studies on the possible association

between coffee consumption and cancer in people

have given conflicting results. Overall, the evidence

suggests that coffee consumption may increase the

risk of bladder cancer in humans, but this conclusion

is by no means definitive because of the possible

failure to account properly for other factors that can

affect cancer rates. Some studies, but not others, have

linked high levels of coffee consumption to cancer of

the pancreas, but again, the positive findings may

result from methodological problems. Unfortunately,

currently available studies on the carcinogenicity of

coffee to experimental animals do little to shed light

on the issue owing to the inadequacy of these studies.

Caffeine, which is found in tea, cola drinks, and

cocoa as well as coffee, has not been found to

be carcinogenic when tested in animals. (See Caffeine;

Coffee: Physiological Effects.)

Alkenylbenzenes and Benzene

0007Several alkenylbenzene compounds have been shown

to induce liver tumors in experimental animals. One

of these compounds, safrole (Figure 1), constitutes

80–90% of the volatile oil of sassafras and is also

present at much lower concentrations in certain

spices and flavoring ingredients such as sweet basil,

cananga oil, nutmeg, pepper, tamarind, and ylang

ylang oil. As a result of the carcinogenicity of safrole,

the US Food and Drug Administration (FDA) pro-

hibited the use of safrole or sassafras oil in food,

which resulted in the cessation of its use as the princi-

pal flavoring ingredient in root beer. Furthermore,

sassafras extracts or leaves can only be sold in the

USA if the safrole has been removed from them.

0008Another carcinogenic alkenylbenzene related in

chemical structure to safrole is estragole (Figure 1;

also called methylchavicol), which is found in

OCH

(a) (b) (c)

(d) (e) (f)

OOC

CO CO

fig0001 Figure 1 Chemical structures of some naturally occurring dietary carcinogens: (a) ptaquiloside, the carcinogenic constituent of

bracken fern; (b) symphytine and (c) senkirkine, two pyrrolizidine alkaloids; (d) safrole and (e) estragole, two alkenylbenzene

carcinogens found in foods; and (f) gyromitrin, a toxic and carcinogenic compound found in the false morel mushroom, Gyromitra

esculenta. Reproduced from Cancer: Carcinogens in the Diet, Encyclopaedia of Food Science, Food Technology and Nutrition, Macrae R,

Robinson RK and Sadler MJ (eds), 1993, Academic Press.

800 CANCER/Carcinogens in the Food Chain

tarragon, sweet basil, anise, West Indian bay, fennel,

chervil, and marjoram. The level of estragole from

these sources in a typical diet is very low, and the

resulting risk is, at most, vanishingly small.

0009 Benzene is a well-established human and animal

carcinogen, inducing leukemia in occupationally

exposed workers and several types of cancer in ben-

zene-treated rodents. Traces of benzene have been

detected among the volatile components of a variety

of foods, including oat groats, processed pork and

ham, cooked meats and baked potatoes, and in the

aromas of coffee and cocoa.

Mushrooms and Hydrazine Derivatives

0010 Gyromitra esculenta is one of the mushrooms known

as false morels. It is harvested and eaten by many

people in northern Europe and also in the USA,

although it is poisonous, even fatal, unless properly

dried and boiled. The principal poisonous ingredient

is gyromitrin (Figure 1), which is carcinogenic to

animals and can give rise to the carcinogen methylhy-

drazine under acidic conditions, such as exist in the

stomach. Both methyl-hydrazine and another car-

cinogen, N-methyl-N-formyl-hydrazine, appear to

form during the preparation of false morel for con-

sumption. In fact, cooking must be performed in an

open vessel to allow methylhydrazine to escape;

otherwise, poisoning can occur.

0011 The widely available edible mushroom cultivated

in countries with a temperate climate is Agaricus

bisporus. The principal hydrazine derivative in

A. bisporus, agaritine, has not been shown to be

carcinogenic, but there are reports of carcinogenicity

for some hydrazine derivatives related to compounds

that have been identified in this mushroom. When

uncooked A. bisporus was fed to mice as their only

food for 3 days per week throughout their lifetimes,

tumors developed in several organs. No adequate

study has yet been performed on the carcinogenicity

of cooked mushrooms, and cooking may inactivate

carcinogens that may be present. Additional stud-

ies are needed to determine the significance of the

results with uncooked A. bisporus and the hydrazine

derivatives.

Carcinogens Produced by the Processing

of Food

Alcoholic Beverages

0012 Alcoholic beverages probably constitute the most

widely consumed class of substances for which

human carcinogenicity is definitively established.

The most clearly affected sites are the mouth, phar-

ynx, larynx, esophagus, and liver. Smoking of

tobacco and consumption of alcohol, each of which

is carcinogenic independently, appear to have a far

greater additive effect in inducing human cancer at

these sites, excluding the liver. There are also studies

implying that alcoholic beverages increase the risk

of breast cancer, but the data in this regard are

not nearly as definitive as those for the other sites

mentioned.

0013It is not clear which components of the alcoholic

beverages are responsible for the human cancers.

Animal experiments have failed to show that ethanol

(alcohol) itself is carcinogenic, although the major

metabolite of ethanol, i.e., acetaldehyde, induces

respiratory-tract tumors when inhaled by experimen-

tal animals. Alcoholic beverages contain numerous

chemicals that may contribute to the carcinogenicity

of the beverages, including traces of acetaldehyde,

N-nitroso compounds (including N-nitrosodimethy-

lamine), and urethan (ethyl carbamate). Urethan has

been found in the highest concentrations in ‘stone

fruit’ (i.e., cherry, plum, apricot, etc.), brandies,

sake, and rice wine; at lower levels in various other

types of distilled spirits and wines; and at even lower

levels in beer. Urethan is also found in other fer-

mented food products such as soya sauce, bread,

yogurt, and some cheeses. The carcinogenic risks

from urethan appear to be extremely low and are

probably negligible, but may be somewhat increased

among those who frequently consume alcoholic

beverages with the highest levels of this carcinogen.

(See Alcohol: Metabolism, Beneficial Effects, and

Toxicology; Alcohol Consumption; Smoking, Diet,

and Health.)

Carcinogens Produced by Cooking

0014Under most circumstances, the cooking of meat, fish,

or eggs results in the formation, at very low concen-

trations, of a group of heterocyclic amines that are

very potent mutagens when tested on bacteria. A few

of these mutagens, all of which are carcinogenic when

fed to experimental animals, are shown in Figure 2

with their commonly used abbreviated names.

0015The formation of mutagenic activity has been ob-

served in beef, pork, ham, bacon, lamb, chicken, fish,

and eggs after broiling (grilling), frying, and barbecu-

ing. Other high-protein foods, such as tofu, beans,

and cheese, gave little or no mutagenic activity

when cooked under similar conditions. Beef extract

is also mutagenic. (See Browning: Nonenzymatic.)

0016While the cooking-induced mutagens that have

been isolated are extraordinarily potent mutagens

when tested on bacteria, these chemicals tend to be

only moderately potent as carcinogens. The fact that

the mutagens are found in foods in the parts per

billion range, and the fact that their carcinogenic

CANCER/Carcinogens in the Food Chain 801

potency, in animals at least, does not seem to be

particularly great, are factors that minimize the an-

ticipated risk from these substances. However, expos-

ure to them occurs each time a person eats cooked

meat or fish, and such exposure often occurs at a high

frequency over the lifetime of the individual. While it

is possible to reduce the level of carcinogenic hetero-

cyclic amines by reducing the temperature or time of

cooking, the reduction in carcinogenic risk thus

effected must be weighed against the very real possi-

bility of increased risk from bacterial or parasitic

infection from undercooked meat or fish. Recent

reports concerning the isolation of an anticarcinogen

from fried ground beef also imply that caution is

necessary before recommending any reduction in the

degree of cooking. Microwave cooking without sur-

face browning apparently does not result in the

formation of heterocyclic amine mutagens or carcino-

gens. (See Mutagens.)

0017 Another class of chemicals containing a number of

carcinogens, the polynuclear aromatic hydrocarbons,

are also introduced into food when meat or fish are

grilled, but not when cooking is performed by the

lower-temperature processes of frying, roasting, or

microwave cooking. The pyrolysis of fat dripping on

to the heat source appears to be a major source of this

class of carcinogens. Benzo[a]pyrene and other poly-

nuclear aromatic hydrocarbon carcinogens appear to

be relatively weak carcinogens when administered

orally to experimental animals, but much more

potent when applied to the skin. Whether this is also

true for humans is unknown, and the risks associated

with these carcinogens are difficult to evaluate. (See

Polycyclic Aromatic Hydrocarbons.)

Salting of fish

0018The salting of fish in southeastern China is carried out

for several days using sea salt, followed by drying in

direct sunlight. Sometimes, the fish is permitted to

soften by decomposition. Several studies have found

that the consumption of such salted fish contributes

to the high incidence of cancer of the nasal passages

and pharynx (nasopharyngeal carcinoma) among

Chinese people in southeastern China, Hong Kong,

(a) (b)

(d) (e)

(f) (g)

(c)

fig0002 Figure 2 Chemical structures of some cooking-induced heterocyclic amine carcinogens. Their commonly used short names are

(a) PhIP, (b) IQ, (c) 8-MeIQx, (d) Trp-P-1, (e) Glu-P-2, (f) A-a-C, and (g) MeA-a-C. Reproduced from Cancer: Carcinogens in the Diet,

Encyclopaedia of Food Science, Food Technology and Nutrition, Macrae R, Robinson RK, and Sadler MJ (eds), 1993, Academic Press.

802 CANCER/Carcinogens in the Food Chain

and Malaysia. Nasopharyngeal carcinoma is rare in

most populations but accounts for about 15% of all

cancer deaths among males in Guangdong Province,

China. Feeding of salted fish to children and to babies

during weaning seems to be a particularly strong risk

factor.

0019 The conclusions of these population studies have

been confirmed by animal studies in which feeding of

Cantonese-style salted fish to rats resulted in the in-

duction of nasopharyngeal carcinomas. N-Nitroso

compounds formed as a result of the salting process

may be responsible for the carcinogenicity of the

salted fish. Nitrate, which contaminates the salt

used to prepare the fish, can be reduced to nitrite

during the processing, and the nitrite can react with

amines in the fish to form N-nitroso compounds.

Since salted fish consumption appears to account for

most of the nasopharyngeal carcinoma cases among

Chinese populations, it must be considered to be a

very high risk factor when consumed regularly, par-

ticularly by children and infants.

Chemicals Added to Foods

Saccharin

0020 Saccharin has been used as a nonnutritive sweetening

agent since 1907, and its safety has been a subject of

debate by scientists and public health officials from

that time. Experiments conducted in the 1970s and

1980s have demonstrated that high levels of sodium

saccharin in the diets of rats resulted in the appear-

ance of tumors of the urinary bladder. In spite of this,

the controversy concerning the safety of saccharin

continues. (See Saccharin.)

0021 Studies on human populations have failed to show

any association between bladder cancer incidence and

the consumption of saccharin or other artificial

sweeteners, but such studies have a limited ability to

detect such associations, even if they exist. The fact

that saccharin is generally not mutagenic, or is very

marginally mutagenic, in a variety of test systems may

imply that saccharin does not interact with DNA and,

therefore, that it may not pose a significant carcino-

genic risk at the usually consumed human doses,

which are far below the doses fed to the rats in the

positive carcinogenicity studies. Saccharin continues

to be widely used in food products around the world.

Cyclamates

0022 Other widely used nonnutritive sweeteners include

sodium cyclamate and calcium cyclamate. In a study

performed in the 1970s, rats that were fed high doses

of a mixture of sodium cyclamate and saccharin

had an increased incidence of bladder tumors. Some

additional studies reported that rats receiving sodium

cyclamate alone developed bladder tumors, but the

results were not statistically significant. Based on

these findings, cyclamates were banned in some coun-

tries, including the UK and the USA. However, many

consider the experimental reports of carcinogeni-

city to be of questionable value, and cyclamates con-

tinue to be approved for use in many countries. As

mentioned above, there are no data from human

populations indicating that consumption of nonnutri-

tive sweeteners are associated with increased cancer

risk, although such studies are, by their very nature,

insensitive. (See Cyclamates.)

Nitrate and Nitrite

0023Sodium nitrate and sodium nitrite are used as preser-

vative agents in cured meats, such as bacon, sausage,

and ham, and in some cheeses. While neither salt has

been shown to be carcinogenic, there is some concern

because nitrite can react with (nitrosate) amines pre-

sent in food to form N-nitroso compounds, many of

which are carcinogens; this reaction can occur under

the acid conditions in the stomach. Since nitrate can

be converted to nitrite either in a food product or in

the mouth or gastrointestinal tract, both nitrate and

nitrite must be considered together in assessing any

possible hazard. (See Curing; Nitrosamines.)

0024There are several dietary sources of nitrate and

nitrite other than cured meats, including vegetables

and some drinking waters. Over the past few decades,

the levels of nitrate and nitrite in cured meats have

been reduced considerably, and the potential for for-

mation of N-nitroso compounds has been further

mitigated by the addition to cured meats of antioxi-

dants such as ascorbic acid, which inhibit the nitrosa-

tion of amines.

0025The addition of nitrate and nitrite to food is only

one of many sources of N-nitroso compounds to

which humans are exposed; salt-preserved fish and

smoked foods are among the other sources of N-

nitroso compounds in the human diet. While there

are some data indicating that nasopharyngeal,

esophageal, and stomach cancer are more common

in populations that consume high levels of nitrate or

certain types of preserved meat or fish, the specific role

of N-nitroso compounds or of added nitrate or nitrite

in human cancer remains to be clarified. It is possible

that the presence of inhibitors of nitrosation in fresh

fruits and vegetables is a factor in explaining why

consumption of these foods appears to have a protect-

ive effect against cancer in human populations.

Butylated Hydroxyanisole

0026The antioxidant butylated hydroxyanisole (BHA) has

been used since 1947 to prevent rancidity in edible

CANCER/Carcinogens in the Food Chain 803

fats and oils and in fat-containing foods. Experiments

performed since 1982 have shown that BHA in the

diet can cause tumors of the forestomach in rats and

hamsters. However, the relevance of this finding has

been questioned because humans do not possess a

forestomach, and their stomachs do not contain the

type of tissue (squamous epithelium) found in

the rodent forestomach. Although humans have squa-

mous epithelium at other sites along the digestive

tract, no tumors have been noted in such sites in

experimental animals. The possibility that BHA may

be a carcinogenic hazard at normal human dietary

levels, which are much lower than the experimental

tumor-inducing levels, has also been questioned on

the grounds that it is not mutagenic in a variety of test

systems. Because of the questions that have been

raised concerning the relevance of the experimental

forestomach tumors induced by BHA to human

exposure at low dietary levels, and the effectiveness

of BHA as an antioxidant in fatty foods, it continues

to be approved for use in many countries around the

world. (See Dietary Fiber: Properties and Sources;

Food Additives: Safety.)

See also: Alcohol: Metabolism, Beneficial Effects, and

Toxicology; Alcohol Consumption; Alkaloids: Toxicology;

Carcinogens: Carcinogenic Substances in Food:

Mechanisms; Carcinogenicity Tests; Contamination of

Food; Curing; Cyclamates; Dietary Fiber: Properties

and Sources; Food Additives: Safety; Mutagens;

Mycotoxins: Occurrence and Determination; Polycyclic

Aromatic Hydrocarbons; Saccharin; Smoking, Diet,

and Health

Further Reading

Aeschbacher H-U (ed.) (1991) Potential carcinogens in the

diet. Mutation Research special issue, vol. 259, nos 3

and 4.

Hayatsu H (ed.) (1991) Mutagens in Food: Detection and

Prevention. Boca Raton, FL: CRC Press.

Hirono I (ed.) (1987) Bioactive Molecules, vol. 2. Naturally

Occurring Carcinogens of Plant Origin. Amsterdam:

Elsevier Science.

IARC Monographs on the Evaluation of Carcinogenic

Risks to Humans: vol. 22, Some Non-Nutritive

Sweetening Agents (1980); Vol. 31, Some Food Addi-

tives, Feed Additives, and Naturally Occurring Sub-

stances (1983); vol. 40, Some Naturally Occurring and

Synthetic Food Components, Furocoumarins, and Ultra-

violet Radiation (1986); vol. 44, Alcohol Drinking

(1988); vol. 51, Coffee, Tea, Mate, Methylxanthines,

and Methylglyoxal (1991) Lyon: International Agency

for Research on Cancer.

National Research Council (1989) Diet and Health: Impli-

cations for Reducing Chronic Disease Risk. Washington,

DC: National Academy Press.

Pariza MW, Aeschbacher H-U, Felton JS and Sato S

(eds)(1990) Progress in Clinical and Biological Re-

search, Vol. 347. Mutagens and Carcinogens in the

Diet. New York: Wiley–Liss.

Weisburger EK (1987) Carcinogenic natural products in the

environment. In: Mehlman M (ed.) Advances in Modern

Environmental Toxicology, vol. X, pp. 243–266.

Diet in Cancer Prevention

I T Johnson, Institute of Food Research, Norwich,

Introduction

0001Cancer is primarily a disease of old age. In societies

where infant mortality and infections have ceased

to be major causes of death, a large proportion of

the population will survive to die eventually of cancer.

Nevertheless, the age-specific incidence of many

common types of cancer varies by a factor of 10-fold

or more between populations, the age-standardized

incidences of particular cancers change within popu-

lations over time, and the risks alter consistently in

populations migrating between countries. Since only

a small proportion of human cancers are directly

attributable to inherited susceptibility, most must be

caused by environmental factors that are, in principle,

avoidable. In their 1981 review of the causes of

cancer, Sir Richard Doll and Richard Peto showed

that, apart from tobacco, which at that time was the

direct cause of 30% of deaths from cancer in the USA,

the largest proportion of avoidable cancer (approxi-

mately 35%) could be attributed to diet. However the

uncertainty attached to this estimate was very high

(10–70%). Surprisingly little has changed in the inter-

vening years. The American epidemiologist WC

Willet has recently suggested that the true proportion

of cancers avoidable by dietary change is probably

between 20 and 42%, but the means by which such

reductions might be achieved remain stubbornly ob-

scure. There are two major sources of difficulty. First,

there are many different types of cancer, and the

various mechanisms by which they arise have only

recently begun to be understood. Second, human diets

are immensely complex and difficult to measure. The

biology of cancer, and the epidemiological evidence

for effects of diet, will be briefly reviewed here, and

the current state of knowledge regarding the role of

dietary factors in the prevention of cancer will be

discussed.

804 CANCER/Diet in Cancer Prevention

The Nature of Cancer

0002 Cell division is fundamental to all life. The human

body begins life with the fertilization of a single cell,

yet by the time maturity is reached the total number

has risen to about 10

. Throughout the earliest

stages of development all the embryonic cells divide

constantly and undergo differentiation to form the

specialized tissues and organs. Cell division is

minimal in many adult tissues but those of the

blood-forming organs, the reproductive tissues, and

the epithelia lining the gut continue to divide at a high

rate throughout life. A tumor can be defined as a

localized increase in the number of cells in an organ,

beyond what is necessary for tissue growth and repair,

or for the reproductive needs of the host. Benign

tumors grow relatively slowly, and retain an approxi-

mately normal anatomical location, whereas malig-

nant tumors acquire the ability to invade surrounding

tissues and form secondary tumors in remote organs

such as the liver or brain. Until relatively recently it

was assumed that the crucial abnormality of cancer-

ous cells was their loss of proliferative control, but

their ability to evade apoptosis, the internal pathway

which obliges damaged cells to undergo suicide, is at

least as important.

000 3 Carcinogenesis occurs in stages that can be broadly

characterized as initiation, promotion, and progres-

sion. The earliest event is thought to be the acquisi-

tion of an unrepaired mutation by a single cell. DNA

damage and repair occur constantly, but a mutation

which remains unrepaired, fails to trigger apoptosis,

confers some growth advantage on the cell, and can

be passed on through successive cellular divisions, is

potentially carcinogenic. During the promotion stage,

the normal constraints on proliferation and spatial

location are disrupted by the acquisition of further

mutations within genes essential for the normal

growth and differentiation. At the progression stage

the lesion has made the transition to malignancy and

can give rise to secondary tumors at remote sites. This

multistage model of carcinogenesis is exemplified by

colorectal cancer, which in most cases develops

through the adenoma–carcinoma sequence. Around

40% of all males over 50 in the UK and other western

industrialized countries have one or more colorectal

polyps. Only a few percent of these will ever progress

to carcinomas, but regular screening and removal of

such polyps are an effective, though costly means of

preventing cancer.

Cancer and Populations

0004 Some examples of the geographical variations that

occur in the incidence of important cancers of the

alimentary tract are given in Table 1. These data are

of comparable reliability, and have been selected to

illustrate some important features of cancer incidence

across the globe. For example, colorectal carcinoma

is about 12 times more common in New Zealand than

in India. This cancer is consistently more common in

richer, westernized countries than in the developing

nations of Asia and Africa that have primarily agrar-

ian economies. Cancer of the stomach is much less

consistently related to economic development. The

highest incidence of gastric cancer in the world occurs

in Japan, but rates are also high in other much poorer

countries of eastern Asia, and in Central and South

America. Within countries, gastric cancer tends to

decline with increasing socioeconomic status. The

third example, esophageal cancer, shows an extra-

ordinary degree of geographical variation. In general

it tends to be more common in poorer societies, but

there is a recognized ‘‘esophageal cancer belt’’ which

runs from China through central Asia to Iran with

incidence rates hundreds of times higher than the least

affected countries of the world. Such extremes cannot

be attributed to general economic conditions. They

suggest that the esophagus is exquisitely sensitive to

some combination of environmental conditions

amongst which diet must play a major role.

Cancer and Diet

0005Three types of interaction between diet and the car-

cinogenic process can be envisaged.

0006The diet may contain specific carcinogens or pro-

moters that initiate or accelerate carcinogenesis.

0007The provision of known nutrients and /or energy

may be a key determinant of the body’s suscepti-

bility to cancer.

tbl0001Table 1 Geographical variations in the incidence of major

cancers of the alimentary tract

Country Incidence of carcinoma for both sexes (cases/100 000)

Esophageal Gastric Colorectal

Brazil 13.6 31.3 27.3

Denmark 8.3 12.6 69.3

Germany 7.0 25.3 77.0

India 12.7 8.5 8.0

Israel 3.1 22.1 78.1

Japan 11.6 97.8 68.5

New Zealand 7.9 16.2 98.7

South Africa 22.6 12.6 23.0

Uganda 24.5 12.5 14.4

UK 13.2 17.9 60.7

USA 6.3 11.2 71.2

Data are predicted age-specific incidence rates for the year 2000. Source:

Globocan2000databaseoftheWorldHealthOrganisation,Geneva(http://

www.who.int).

CANCER/Diet in Cancer Prevention 805