2705 Am J C/in Nutr l99l;53:270S-82S. Printed in USA. © 1991 American Society for Clinical Nutrition

Vitamin C and cancer prevention:the epidemiologic evidence”2

Gladi�s Block

ABSTRACT Epidemiologic evidence of a protective effect

of vitamin C for non-hormone-dependent cancers is strong. Of

the 46 such studies in which a dietary vitamin C index was

calculated, 33 found statistically significant protection, with high

intake conferring approximately a twofold protective effect

compared with low intake. 0f29 additional studies that assessed

fruit intake, 2 1 found significant protection. For cancers of the

esophagus, larynx, oral cavity. and pancreas, evidence for a pro-

tective effect of vitamin C or some component in fruit is strong

and consistent. For cancers ofthe stomach, rectum, breast, and

cervix there is also strong evidence. Several recent lung cancer

studies found significant protective effects of vitamin C or of

foods that are better sources of vitamin C than of /3-carotene. It

is likely that ascorbic acid, carotenoids, and other factors in fruits

and vegetables act jointly. Increased consumption of fruits and

vegetables in general should be encouraged. Am J C/in Nuir

199 1;53:270S-82S.

KEY WORDS Ascorbic acid, vitamin C, epidemiologic

studies, epithelial cancers, cancer, review

Introduction

Ascorbic acid is increasingly recognized as an agent with broad

biologic function and importance. Well-established functions

include synthesis of hormones and neurotransmitters, cyto-

chrome P-450 activity and detoxification of exogenous com-

pounds. carnitine synthesis and cholesterol metabolism ( 1 , 2),

as well as its well-known antioxidant functions with protective

results that may extend to cancer, coronary artery disease, ar-

thntis, and aging. Several possible mechanisms of action of

ascorbic acid in cancer prevention have been described exten-

sively elsewhere (3-5). It plays a major and perhaps even pre-

dominant role in free-radical scavenging and protection against

lipid peroxidation (6, 7). It appears to have a role in sparing or

reconstituting the active forms of vitamin E (8- 14), and spares

other important antioxidants (7). Several functions of vitamin

C in the immune system have been described (1 5, 16), including

enhancement of leukocyte chemotaxis ( 1 7), stimulation of in-

terferon production ( 1 8), and complement Clq activity (19). Its

role in collagen synthesis and basement membrane integrity and

in hyaluronidase inhibition (20-22) may be important in inhib-

iting tumor spread and micrometastases.

Before an examination ofthe epidemiologic evidence that cx-

ists regarding ascorbic acid and cancer prevention, it is important

to consider several sources ofpotential misinterpretation. These

have been discussed in some detail elsewhere (4, 23). Briefly,

studies of the role of nutrients in disease are hampered by at

least three sources of error: 1) error in the classification of in-

dividuals with respect to their nutrient intake, 2) errors of in-

terpretation arising from the fact that nutrients are correlated

both negatively and positively with other nutrients, and 3) studies

within populations that are quite homogeneous with respect to

intake ofa nutrient may be unable to detect an effect ofhigh or

low intake.

Misclassification errors cause a bias ofobserved risk estimates

toward the null, making the observed risk estimate appear weaker

than it really is and making it more difficult to achieve statistical

significance. Studies that nevertheless show an effect in the face

of the misclassification inherent in dietary or biochemical as-

sessment probably indicate an even stronger effect if the true

long-term dietary or biochemical status could be known.

Correlated variables mislead investigators into identifying the

wrong nutrient as the effective one because, although an index

for a certain nutrient may be calculated, other nutrients mcvi-

tably accompany it in the food and are correlated with it in the

plasma. The nutrient that is calculated reflects the interests and

presuppositions of the investigator and does not necessarily re-

fleet the true causal agent. Furthermore, in biological systems it

is likely that nutrients do not act singly but jointly and very

rarely have investigators studied the joint effects ofhaving a low

intake of two or more nutrients simultaneously.

Finally, studies in homogeneous populations may produce

negative results that are difficult to interpret. For example, in

studies in very high risk populations everyone in the population

may have a very low intake of a protective nutrient, control

subjects as well as cases. Some of these are highlighted under

esophageal cancer, below, but the problem may be more wide-

spread. The same situation may exist in the opposite direction;

if all in a study group are very well nourished with respect to a

nutrient, the effect of low levels cannot be assessed well.

Epidemiologic studies by cancer site

The studies reported below are limited to analytic epidemi-

ologic studies in humans-that is, case-control and prospective

I From the Division of Cancer Prevention and Control, National

Cancer Institute, Bethesda, MD.2 Address reprint requests to G Block, National Cancer Institute, FPN

Room 3 13, 9000 Rockville Pike, Bethesda, MD 20892.

at TE

L AV

IV U

NIV

ER

SIT

Y on N

ovember 21, 2011

ww

w.ajcn.org

Dow

nloaded from

ASCORBIC ACID IN CANCER PREVENTION 2715

studies. No international correlation studies are included nor

any studies that simply attempt to relate nutrient intake in a

region with population mortality or morbidity rates in a region.

The studies reported below are all at the individual level: mdi-

viduals with and without cancer are identified and dietary intake

in those same individuals is assessed. Such studies generally result

in estimates of relative risk (occasionally termed risk ratio or

odds ratio and abbreviated RR or OR). The risk in the group

exposed to a risk factor (here, low intake of vitamin C or foods

rich in it) is expressed as a ratio ofthose not exposed (here, those

with high intake). For simplicity, all risks have been expressed

in this direction so that a relative risk > 1 .0 means an elevated

risk of the cancer among those with low intake. For example, a

relative risk (RR) = 2.0 indicates that those with low intake of

vitamin C are estimated to have twice the risk of that cancer

compared with those with high intake.

Studies are described below if they reported either an actual

vitamin C score or estimate or if they reported on a food that

is rich in vitamin C, usually fruit but occasionally tomatoes or

raw vegetables. Studies that simply reported on green vegetables

are not included because, although those foods may contain

vitamin C. their carotenoid content is usually such that it is

impossible to attribute a protective effect to vitamin C with any

confidence.

Non-hormone-dependent cancers

C’ancers oJ�t/ie oral cavity, larynx, or esophagus

Oral cancer has been the subject ofseven case-control studies,

most of them large and well controlled, investigating the role of

dietary factors (24-30). All but one have found significant pro-

tective effects for increasing vitamin C intake or fruit intake.

Most ofthese studies included control for smoking and alcohol.

McLaughlin et al (25) found that those in the lowest quartile of

vitamin C intake had a risk ratio of 1.7 (men) or 2.0 (women).

Winn et al (28) found that those who reported consuming fruit

once per week or less often had I .7 times the risk of those who

consumed it seven times a week or more. Two of the studies

also found protection from carotene-containing vegetables,

whereas in the others these foods were weaker or nonsignificant.

Two studies of cancer of the larynx have examined the role

of vitamin C (3 1) or fruit (32). Graham et al (3 1 ) studied 374

patients with cancer of the larynx and hospital control subjects

using the Roswell Park questionnaire. After control for possible

confounding by alcohol intake and smoking, a low intake of

vitamin C was associated with a risk ratio of 2.4 (P < 0.005).

Notani and Jayant (32) observed a statistically significant in-

creased risk of2.0 associated with consumption offruit less than

once a week compared with at least once a week. Vegetable

consumption was also a significant factor.

Cancer of the esophagus has been studied by several investi-

gators (33-44) and again, most have found strong and statistically

significant relationships with vitamin C or fruit intake. All four

that examined a vitamin C index (33, 34, 36, 37) found statis-

tically significant relative risks of ‘�2.0 after adjustment for

smoking and alcohol consumption. Brown et al (33) found that

compared with those in the upper third of the distribution for

vitamin C intake, those in the lower third had a statistically

significant twofold risk ratio. Fiber was also significantly pro-

tective but carotenoids were not. Fruits, especially citrus fruits

and juices, were the only significantly protective foods or food

groups. Tuyns et al (34, 35) studied esophageal cancer in a large

case-control study conducted in a high-risk region of France. A

statistically significant protective effect was observed for citrus

fruit (34) and for vitamin C (35). Compared with heavy con-

sumers of vitamin C, light consumers had a risk ratio of 2.6

after adjustment for age, smoking. and alcohol consumption.

Carotene also had a significant but somewhat weaker effect.

Ziegler et al (36) used a 3 1-item next-of-kin questionnaire to

investigate the role of diet among 120 black males who died of

esophageal cancer and 250 others who died of other causes. Vi-

tamin C was the only nutrient with a statistically significant

elevation in risk associated with being in the low one third of

the distribution, RR = 1 .8 after adjustment for ethanol intake.

Finally, Mettlin et al (37) found a risk ratio of 2.4 (P = 0.004)

for low vs high quartile ofvitamin C intake even after adjustment

for smoking and alcohol.

Four studies have found fruit or citrus fruit to be lower in

cases with esophageal cancer (38-41 , 44). Decarli et al (44) in-

vestigated the role ofdiet among 105 cases ofesophageal cancer

and hospital controls. Fruit and vegetable questions consisted

ofonly three items, carrots, green vegetables, and fresh fruit. All

were in the protective direction, but after adjustment for alcohol,

smoking, and other factors only fruit intake remained strong

(RR = 3.3) and statistically significant (trend P < 0.001). Cook-

Mozaffari et al (38) studied 354 patients with esophageal cancer

and population controls in Iran, an area ofhigh esophageal can-

cer incidence. Low intake of fresh fruit and vegetables was

strongly associated with risk ofesophageal cancer. with frequent

consumption of oranges the most significantly protective food.

In Japan, Hirayama (39) found a risk ratio of -��2.7 for less than

daily consumption of fruit in a case-control study of esophageal

cancer. Bjelke (40, 41), in a study of 52 patients with esophageal

cancer and hospital control subjects in Minnesota, observed

nonsignificantly lower intake of apples and green salad in case

subjects, but did not report on a vitamin C index.

Some studies represent classic examples of population ho-

mogeneity ofintake that may make the existence ofa protective

effect impossible to detect. In a region of China with very high

esophageal cancer rates, Li et al (42) conducted a large case-

control study ofesophageal or gastric cardia cancer. No protective

effect was seen for fruit consumption. However, the authors note

that fruit is very infrequently eaten in this high-risk area, and

the highest intake category consisted ofthose who ate fresh fruit

> 35 times per year. Martinez (43) studied risk factors for

esophageal and oral cancer in Puerto Rico, an area in which the

incidence of these cancers is among the highest in the world.

Risk ratios were not specifically reported. but the author reported

that both patients and control subjects were chronically deficient:

75% of both patients and control subjects never had fresh fruit

in their diets and 87% of both patients and control subjects had

only two portions or less ofvegetables per week. Similarly, Notani

and Jayant (32) studied esophageal and oral cancer in regions

of India where incidence of these diseases is among the highest

in the world. Fruit consumption showed a nonsignificant effect,

but was so infrequent that the distribution was dichotomized

into those who ate it once a week and those who ate it less often.

Vegetable consumption, dichotomized on daily rather than

weekly consumption, was significant.

In summary, ofeight studies that have reported on a vitamin

C index in cancer ofthe oral cavity. larynx, or esophagus, every

one has found a statistically significant elevation in risk associated

at TE

L AV

IV U

NIV

ER

SIT

Y on N

ovember 21, 2011

ww

w.ajcn.org

Dow

nloaded from

2725 BLOCK

with low intake. Three of these studies found carotene to be

weak and nonsignificant, whereas in two vitamin A was also

found to be significant but weaker than vitamin C. Of the 12

studies of foods rather than nutrients, 6 found statistically sig-

nificant risk for low fruit intake, 2 found suggestive results or

did not report statistical significance, 3 found extremely low

intakes in both patients and control subjects in extremely high-

risk populations, and 1 found no effect.

Lung cancer

Of the �‘��-30 epidemiologic studies that have examined the

relationship between dietary factors and lung cancer, most have

focused on vitamin A or carotenoids. Because of this focus, in

many cases the diet questionnaire food list was not designed to

assess vitamin C well and the investigators did not attempt to

draw any conclusions about the role ofvitamin C. A few inves-

tigators have examined the role ofvitamin C, however, and these

are reported below, as well as investigations in which the role

of vitamin C-rich foods could be separately assessed.

Fontham et al (45)examined 1253 lung cancer cases and 1274

age-. sex-, and race-matched controls in the New Orleans area.

Approximately 3 1% ofthe study group was black and 26% were

nonsmokers or former smokers who had quit for � 3 y. Results

were adjusted for smoking, income, and other confounders.

Compared with those who consumed 140 mg or more of vitamin

C per day (as estimated by the questionnaire), those who con-

sumed < 90 mg had a risk estimate of 1.5 (P < 0.001). The

vitamin C effect became stronger (RR 1.9) after adjustment

for carotene intake, whereas the apparent protection due to car-

otene disappeared after adjustment for vitamin C intake.

Koo (46) studied 88 Hong Kong women with lung cancer

and 1 37 control subjects to examine risk factors among mdi-

viduals who have never smoked. Risk estimates were adjusted

for age, parity, and education. Only fresh fruit and fresh fish

consumption were found to be significantly related to a reduction

in lung cancer risk. Leafy green vegetables and carrots were in

the protective direction but nonsignificant. By contrast, those

who consumed fresh fruit less than once a week had a risk ratio

of2.4 compared with those who consumed it five to seven times

a week (trend P < 0.002). A vitamin C score yielded a risk

estimate of 2. 1 (trend P < 0.01 5); the effect was strongest for

adenocarcinoma, although the number ofsquamous cell cancers

in these women may have been too small for a stable risk esti-

mate.

In a study conducted in the Netherlands, HoIst et al (47) stud-

ied 49 lung cancer patients in a case-control study. After ad-

justment for smoking and other risk factors, persons with a vi-

tamin C index < 50 mg/d (as estimated by questionnaire) had

a risk 4.3 times greater than those with a higher intake. This

result was highly statistically significant. No effect ofa /3-carotene

score was seen.

Le Marchand et al (48) studied 230 male and 102 female lung

cancer cases and population-based controls in Hawaii. Among

males, both total vitamin C and vitamin C from foods alone

was significantly protective, RR = 2.0 for the contrast between

lowest and highest quartile, a stronger effect for vitamin C con-

trolled for 13-carotene than was seen for /3-carotene controlled

for vitamin C. However, among females a nonsignificant effect

in the protective direction was seen only for intake from foods,

not supplements, whereas a /3-carotene effect was stronger andsignificant. Vegetables were strongly protective, whereas fruit

was not.

Kromhout (49) analyzed the results of a prospective study in

which a dietary history interview was completed in 1960 on 870

men aged 40-59 y; 63 subsequently died of lung cancer over

the 25-y follow-up. Compared with men consuming � 100 mg

vitamin C/d in 1960 (as estimated by the diet history interview),

those consuming < 60 mg had a 2.8-fold increased risk (age but

not smoking adjusted, P < 0.01). Dietary vitamin C intake re-

mained a statistically significant protective factor after appro-

priate multivariate adjustment for age, cigarette smoking, and

vitamin supplement use. A 55-y-old, male, one-pack-a-day

smoker with low dietary and supplement vitamin C had a 25%

chance ofdying from lung cancer in the 25-y follow-up, whereas

the same smoker with high dietary and supplement vitamin C

had only a 7% chance oflung cancer mortality. Fruit intake was

also significantly and inversely related to subsequent lung cancer

mortality, after age and smoking adjustment. /3-carotene intake

was not related to lung cancer mortality.

The above five recent studies of diet in lung cancer have all

found a statistically significant protective effect of vitamin C.

Four of them have found a weak and nonsignificant effect of /3-

carotene; the fifth (48) found an equal effect of 13-carotene and

vitamin C from foods in men, and a stronger effect of /3-carotene

in women. The comparison with vitamin C from foods is the

appropriate one since the 13-carotene is all from foods and since

supplement use is a behavior commonly increased in prodromal

states, particularly among women.

Four additional studies have reported on a vitamin C index

from the diet (50-52) or in plasma (53) and have found suggestive

but nonsignificant results. Investigators in Hawaii (50, 54) con-

ducted a retrospective study in Hawaii of 26 1 male and 103

female patients, including all five ethnic groups of Hawaii, and

population control subjects. The effect of vitamin C was in the

protective direction, RR = 1.6 for males, but was not statistically

significant. The effect of vitamin C from foods alone was not

reported; instead, vitamin C intake included supplements used

in the 3 mo preceding recognition of lung cancer symptoms.

Supplement use is generally much greater for vitamin C than

vitamin A, greater for women than men, and greater for Cau-

casians than other ethnic groups (55). Consequently, inclusion

of supplement use introduces difficulties of interpretation since

individuals may start taking supplements when they experience

malaise or other prodromal symptoms. This may explain why

in LeMarchand’s study in Hawaii vitamin C from supplements

was in the harmful direction, but only among Caucasian women.

In all ethnic and sex groups, vitamin C from foods alone appeared

to reduce the risk of lung cancer.

Kvale et al (5 1) and Bjelke (56) conducted a prospective study

ofrespondents to a mailed diet questionnaire that included 10

ascorbic acid-rich foods. They observed I 53 subsequent lung

cancer deaths and found no effect for a vitamin C score. To-

matoes, however, rich in vitamin C but not /3-carotene, were

one ofonly two foods that approached significance as individual

items (trend P = 0.07 for squamous and small-cell carcinomas).

Shekelle et al (52) also conducted a prospective study and oh-

served a total of 33 lung cancer cases. Unlike the 13-carotene

score, the vitamin C score was based on a 24-h recall. Considering

the limitations ofthat method for vitamin C and the small num-

ber of cases, the P value of 0.20 for the difference between the

mean of 9 1 .8 mg of vitamin C per day among those who later

developed lung cancer and the mean of 10 1.0 mg among those

who did not is suggestive.

St#{228}helinet al (53, 57) collected plasma from respondents in

at TE

L AV

IV U

NIV

ER

SIT

Y on N

ovember 21, 2011

ww

w.ajcn.org

Dow

nloaded from

ASCORBIC ACID IN CANCER PREVENTION 2735

197 1- 1973 and followed them for mortality through 1980. In

that time there were 35 lung cancer deaths for which baseline

plasma vitamins could be analyzed. Persons who subsequently

died of lung cancer had lower plasma ascorbate (0.79 mg/dL)

than control subjects (0.90 mg/dL.), P = 0.2. Plasma /3-carotene

was significantly different. Subsequent analyses by St#{228}helin(58)

of more recent cancer deaths did not reveal any differences in

mean plasma ascorbate between lung cancer patients and control

subjects.Finally, two studies have reported no effect ofa dietary vitamin

C index. Two reports using the 1957 Roswell Park questionnaire

(59, 60) found no effect of vitamin C after control for other

factors. The instrument was not welldesigned to assess vitamin

C as noted by the authors (6 1). Byers et al (62) found no effect

for vitamin C in a 1987 study in upstate New York; however,

three fourths of the eligible cases could not be interviewed be-

cause ofillness or death. Thus, it is possible that a nutrient effect

could be missed if low intake had the effect of increasing the

severity or progression of disease.

Although numerous investigators have studied the role of car-

otene-containing vegetables, in relatively few could the role of

fruits, especially vitamin C-rich fruits or raw vegetables, be cx-

amined separately. These are reported below. Long-dc and

Hammond (63) examined data from a prospective study in which

1 36 000 white males in the United States provided information

on their frequency ofconsumption offruit and green salad; 671

lung cancer deaths were seen over the subsequent 1 1-y follow-

up. Men who consumed fruit less than three times per week had

1 .7 times the lung cancer risk ofthose who ate it 5-7 times per

week, a statistically significant increase. Green salad showed a

lower effect in the same direction. Bond et al (64) reported on

a carotenoid and not a vitamin C index. However, of the six

food items found to be individually significant, five (tomatoes,

melon, pasta with sauce, broccoli, and fortified cereals) are rich

in vitamin C as well as (or in some cases instead of) carotenoids.

Others have reported protective effects of green (65) or green

and yellow (66) vegetables. Whereas the protective factor is gen-

erally assumed to be a carotenoid, it is worth noting that many

green vegetables are rich in vitamin C.

Ofthe 1 1 lung cancer studies that have specifically mentioned

a vitamin C score, 5 have found a statistically significant pro-

tective effect, 4 have found effects in the protective direction but

not significant, and 2 have reported no effect. In two of the

studies that found suggestive but nonsignificant results (52, 57),

the lack of statistical significance was based on I tests involving

small numbers oflung cancer cases, n = 33 and 35, respectively.

Of two additional studies that assessed fruit intake, both have

found significant protective effects. Four studies have found

stronger effects for vitamin C than for carotenoids. Whereas a

large body of evidence suggests an important effect for carot-

enoids in lung cancer protection, the recent data reported above

suggest that there may also be an independent protective effect

of vitamin C intake. The role of vitamin C not only as an an-

tioxidant and free-radical scavenger in its own right (6, 7) but

also in enhancing the action of vitamin E (8-14) is well estab-

lished.

Pancreatic cancer

Five studies have examined the risk of pancreatic cancer in

relation to either a vitamin C index (67) or fruit (68-7 1 ). Falk

et al (67) investigated 363 cases and matched hospital control

subjects in Louisiana. The questionnaire was designed to assess

a wide range ofnutrients and asked about usual adult diet. Those

who consumed less than ‘-70 mg vitamin C/d (as estimated by

the questionnaire) had a relative risk of 2.6 for males, or 1 .8 for

females compared with those who consumed � I 59 mg/d (in

both cases, trend P < 0.05). Fruit consumption (based on fre-

quency of consumption of six fruits, three of which are rich in

vitamin C) conferred a 1.6-fold protective effect after adjustment

for multiple confounding factors.

Norell et al (68), in a case-control study in Sweden of 99

patients and matched hospital and population control subjects,

found a highly significant protective effect both for carrots and

for citrus fruits. For citrus fruits there was a two- to threefold

protective effect for those who ate them daily compared with

those who ate them less than weekly. Mack et al (69) studied

490 pancreas cancer patients and matched neighborhood con-

trols. For those who consumed fresh fruit or vegetables less than

five times a week there was a statistically significant elevated

risk, RR = 1 .4 compared with those who ate them more often.

The diet instrument included only very broad food categories,

the only relevant one here being “fresh fruit or raw vegetables.”

Considering the obvious crudity of the instrument, the obser-

vation of a significant effect is striking. Finally. Gold et al (70)

studied 201 persons with pancreatic cancer and both hospital

and neighborhood control subjects. Adjusted for several con-

founding factors, frequent consumption of raw fruits and veg-

etables was highly protective, RR = 1 .8 for those who ate them

less than five times a week compared with those who ate them

five or more times (P < 0.02).

A prospective study among Seventh Day Adventists (7 1) oh-

served 40 deaths from pancreatic cancer ( 1 7 men and 23 women).

The intake distributions of dietary variables were divided into

thirds and risk within each third was adjusted for age and sex,

a process which, although appropriate, results in small numbers

and reduced statistical power. Frequent consumption of dried

fruit was significantly protective; use of tomatoes, fresh citrus

fruit, and fresh winter fruit showed a “suggestive, though non-

significant, protective relationship.”

Pancreatic cancer is the fifth most common cause of cancer

mortality in the United States (72) and is a disease with extremely

poor prognosis. In the current state of our ability to treat this

disease, prevention is of primary importance. All five pancreas

cancer studies have found statistically significant protective ef-

fects for fruit, and in some instances for vegetables as well. The

one study that calculated a vitamin C index found a significant

twofold reduction in risk associated with high intake.

Stomach cancer

Several studies of an ecological nature have been conducted

(73-75) that suggested a protective role for fruit or for vitamin

C intake in stomach cancer. Ascorbic acid has recently been

demonstrated to be concentrated at three times the plasma level

in the gastric juice of persons with normal gastric histology, but

not in those with chronic gastritis (76). Moreover, in normal

persons it is predominantly in the reduced form, ascorbic acid,

the form required for the antinitrosation reaction; in gastritis

patients, by contrast, it is predominantly in the oxidized form,

dehydroascorbic acid. A similar pattern is seen for gastric tissue

ascorbic acid, which often falls to “immeasurable levels” (77)

in patients with chronic gastritis. The focus below is on studies

at the individual level, but it should be noted that indices of

vitamin C intake have heretofore not differentiated between

at TE

L AV

IV U

NIV

ER

SIT

Y on N

ovember 21, 2011

ww

w.ajcn.org

Dow

nloaded from

2745 BLOCK

ascorbic and dehydroascorbic acid or considered the conversion

of ascorbic to dehydroascorbic acid during cooking, and that

furthermore the correlation between vitamin C intake and gastric

ascorbic acid was low, only r = 0.22 (77).

Several studies have found a statistically significant protective

effect of vitamin C or fruit in gastric cancer. Bjelke (40, 41, 78)

found vitamin C to be “the dietary variable which discriminated

most strongly between the total group of stomach cancer cases

. . . and the controls” (40). The effect was particularly strong

for gastric carcinomas ofthe diffuse and intestinal types. For the

combined group ofdiffuse and intestinal carcinomas, the relative

odds for low vitamin C intake were 1.8 (40). In Norway, a sta-

tistically significant difference was seen in both males and females

(41) and a strong effect was also observed in Minnesota (40).

No effect of vitamin A was seen for stomach cancer. The author

notes (40) that “the data for stomach cancer are consistent with

an increased risk being associated with a classical deficiency af-

fecting a minority of the population.”

In a study ofdietary factors and gastric cancer in Louisiana,

Correa et al (79) found fruits as a group and dietary vitamin C

to be statistically significantly protective in both blacks and whites

after multivariate control for smoking and other factors. Those

in the lower quarter ofthe intake distribution had twice the risk

(whites) or three times the risk (blacks) compared with those in

the upper quarter. A nonsignificant protective effect for carot-

enoids disappeared after control for vitamin C, whereas the vi-

tamin C effect remained after control for carotenoids and other

risk factors.

In a high-risk region ofChina, You et al (80) observed a sta-

tistically significant twofold relative risk of stomach cancer for

the lowest quartile ofvitamin C intake after control for sex, age,

and income. A similar effect was also seen for carotene. Fruit

was also significantly protective among those who ate > 5 kg/y.

A stronger effect was seen for fresh vegetables where the lowest

intake quartile was � 73 kg/y.

La Vecchia et al (8 1 ) conducted a case-control study in Italy

of206 gastric cancer patients and hospital control subjects. Those

in the bottom one third of the distribution of vitamin C intake

had a 2.5-fold elevated risk, P < 0.001 . The effect of /3-carotene

was similar in multiple regression analysis. In Canada, Risch et

al (82) studied 246 gastric cancer patients and population control

subjects. Although the questionnaire included 1 5 good veg-

etable sources ofvitamin C and three fruit sources, two important

sources, citrusjuices and tomatoes, were either omitted or com-

bined with poor vitamin C sources. a characteristic that may

have generated imprecision in the vitamin C estimate. Citrus

fruit, excluding juice. was highly significant, whereas the overall

vitamin C estimate was significant only at P < 0. 10. However,

when ascorbate was examined just from the 2 1 nitrate-containing

vegetables, a strongly protective and statistically significant effect

was seen for increasing ascorbate intake. This analysis tends to

confirm the hypothesis of an antinitrosation mechanism for

protection by ascorbic acid in stomach cancer. Similarly,

Meinsma (83) found strong increased risk with high bacon intake

and a strong protective effect (P < 0.01) of high consumption

ofcitrus fruit and vitamin C.

In a prospective investigation of the relationship between

plasma levels of nutrients and subsequent cancer, St#{228}helinet al

(53, 57, 58) observed lower levels ofvitamin C in the plasma of

20 persons who later developed stomach cancer (42.6 �zmol/L)

than those who did not (52.8 �mol/L). An examination of their

dietary intake (83) revealed approximately a threefold risk of

subsequent stomach cancer among those who reported infre-

quent citrus fruit consumption in 1971-1973 (P = 0.07).

Several investigators did not report a vitamin C index but did

report specifically on fruit intake. In a study ofgastric cancer in

Japan, Kono et al (85) found a significant twofold elevation in

risk for those who consumed fruit less often than daily, after

adjustment for smoking and other dietary factors. Jedrychowski

et al (86), in Poland, found a statistically significant relative risk

of 3.2 for those who ate fruit less often than twice a week com-

pared with those who ate it daily. In Greece, Trichopoulos et al

(87) observed a statistically significant protective effect of fre-

quent consumption of citrus fruits and of raw salad-type vege-

tables in 1 10 patients with adenocarcinoma of the stomach.

Graham et al (88), in a matched-pair reanalysis ofa previously-

reported negative study (8�), observed a reduced risk of gastric

cancer associated with consuming various vegetables in the Un-

cooked state. Coggon et al (90) in a study of 95 patients and

population control subjects in England found statistically sig-

nificant reductions in risk, twofold or greater, for fruit and for

salad vegetables. For fruit intake, the high-consumption category

consisted of those consuming it six or more times per week. In

1972 Haenszel et al (9 1) reported effects for several vegetables,

but only tomatoes (a source ofvitamin C but not of /3-carotene)

showed protective effects in both Issei and Nisei. No effect of

fruit was seen in this Japanese American population. A study

in Japan (92) found no effect of orange consumption. Finally,

in 1966 Higginson (93) found less frequent consumption of fresh

fruit and raw vegetables among stomach cancer patients than

matched hospital controls. Fontham et al (94) found a statistically

significant 2.5-fold elevation in risk of gastric cancer precursor

lesions for those below the median intake ofvitamin C in a high-

risk black population in Louisiana. In a negative study of pre-

cursor lesions, however, Haenszel et al (95) found no difference

between the mean plasma levels of total ascorbate in persons

with atrophic gastritis and control subjects in Colombia.

In summary, seven investigators have reported on vitamin C

dietary intake and stomach cancer risk. All seven have found

statistically significant protective effects of approximately two-

fold, usually in the overall vitamin C score although in one in-

stance statistical significance was limited to the vitamin C con-

tamed in nitrate-containing vegetables. One prospective study

found lower plasma vitamin C levels in those who subsequently

developed stomach cancer, nearly significant at P = 0.06 despite

having only 19 cases. In two studies that also examined 13-car-

otene, that factor was also significantly protective; in two other

studies that simultaneously examined 13-carotene and vitamin

C in multivariate models both remained significant in one study,

whereas the effect of 13-carotene disappeared in the other (79).

Eight studies have reported on the role of fruit intake; seven

have found lower fruit consumption in those who developed

stomach cancer, in most instances a statistically significant two-

to threefold effect. Consumption ofraw salad vegetables is often

found to be protective as well.

Cervical cancer

The role of ascorbic acid in the precancerous condition, cer-

vical dysplasia, was examined by Wassertheil-Smoller et al (96)

and by Romney et al (97). Plasma ascorbic acid levels were 0.36

mg/dL in women with the precursor lesion and 0.75 mg/dL in

at TE

L AV

IV U

NIV

ER

SIT

Y on N

ovember 21, 2011

ww

w.ajcn.org

Dow

nloaded from

ASCORBIC ACID IN CANCER PREVENTION 2755

control subjects, (P < 0.000 1 ) even after control for factors such

as multiple partners, multiple pregnancies, early onset of sexual

activity, and socioeconomic status (97). Dietary factors were

also examined in a different case-control series of 87 women

with cervical dysplasia and matched control subjects (96). Com-

pared with those above the median intake of 88 mg/d, those

below that median had a significant excess risk (RR = 4.35) for

severe dysplasia or carcinoma in situ. Vitamin C intake remained

a significant factor even after control for age and sexual activity.

Brock et al (98) studied 1 17 in situ cervical cancer patients

and matched community controls in Australia. Those in the

lowest quarter of the distribution of intake of vitamin C had a

significantly elevated risk (RR 2.5). After control for eight

sexual activity and dietary variables, including carotene, the risk

estimate was still ‘�2.0, although this was no longer statistically

significant. In contrast, carotene no longer appeared to be pro-

tective after control for vitamin C and the other variables. Fruit

intake more than once per day remained significantly protective

(RR = 2.5). Plasma 13-carotene was significantly protective;

plasma ascorbic acid was not measured.

Verreault et al (99) examined 1 89 women with invasive cer-

vical cancer and population controls. Low intake of vitamin C

was associated with a statistically significant twofold increased

risk. The effect of fruit juices was even stronger (OR = 3, P

< 0.0 1 ). The effect of carotene and dark green vegetables was

in the same direction, although weaker.

Marshall et al (6 1) failed to observe a protective effect of vi-

tamin C in a study of 5 1 3 cervical cancer patients and hospital

control subjects. An effect was seen for the carotene index. The

authors point out that the Roswell Park instrument was “cx-

tremely imprecise in the measurement ofa key source of vitamin

C, fresh fruits.”

Co/orecial cancer

The role of vitamin C in the occurrence of rectal polyps is

unclear. Sporadic or nonsignificant polyp regression (100), re-

ductions in polyp area (101), or nonsignificant reductions in

recurrence rate (102, 103) have been reported. In the latter study,

35% ofpatients who had been receiving vitamins for 2#{189}y were

polyp free compared with 23% of those on placebo, an effect

that the study had insufficient power to detect as significant. A

recent report found minimal effect of vitamin C alone on oc-

currence of polyps in familial polyposis patients (104), whereas

cereal fiber was effective; the most striking effects were seen in

those who had a high compliance to both fiber and vitamin

supplements. Such studies have been quite small and effects of

such manipulations on actual progression to cancer have not

been observed.

A possible role of vitamin C in prevention or reduction of

fecal mutagens in stool has been observed in two studies (105,

106). In a study of diet and fecapentaene levels conducted by

the National Cancer Institute (106), a strong protective effect

was observed for both dietary and supplemental vitamin C in-

take. Consumption of citrus fruit and of supplemental vitamin

E was also significantly negatively associated with fecapentaene

levels. Thus, if fecal mutagens are associated with colon carci-

nogenesis, vitamin C may play an independent role or may act

synergistically with vitamin E in blocking their effect.

Several studies have examined rectal and colon cancer sepa-

rately. An examination ofthese data suggest that vitamin C may

act differently at these two sites. Rectal cancer will be discussed

first below followed by colon cancer and studies addressing only

the combined group, colorectal cancer.

Seven studies have reported on the association of a vitamin

C index, fruit or raw vegetable intake with rectal cancer, often

with statistically significant results. Kune et al (107. 108) cx-

amined rectal and colon cancer separately in a large case-control

study in Australia. A statistically significant protective effect (OR

= I .7) was seen for high intake of dietary vitamin C as well as

for vitamin supplements in rectal cancer. Tuyns et al ( 109-1 1 1)

conducted a study in Belgium of 365 persons with rectal cancer

and population controls. Vitamin C was significantly protective

(RR = 1.5, P < 0.03) after adjustment for age, sex, province,

and total calorie intake (109, 1 1 1). Total fiber was also protective

but 13-carotene was not. Vegetables. both cooked and raw, were

also found to be highly protective ( 1 10) with lesser effects for

fruit. Potter and McMichael (1 12) found a protective effect of

vitamin C in a study of 124 male and 75 female cases of rectal

cancer and appropriate control subjects. The relative risk asso-

ciated with being in the lowest quintile of intake compared with

the highest was 1 .7 among men, 3.3 among women, the latter

being statistically significant. The role of 13-carotene and fiber

was weak and inconsistent.

Heilbrun et al (1 1 3) compared the baseline 24-h recalls of 60

Japanese men in Hawaii who subsequently developed rectal

cancer and control subjects who did not. Vitamin C intake was

generally quite high, and although lower in patients (105 mg/d)

than in control subjects ( 1 16 mg/d), it was not statistically sig-

nificant. No effect was seen in a quintile analysis among the 60

patients and their control subjects, nor was an effect of dietary

fiber or /3-carotene observed. La Vecchia et al (1 14) studied 236

rectal cancer patients and hospital control subjects and observed

an effect in the protective direction but that effect was nonsig-

nificant. Several vitamin C-rich foods were significantly protec-

tive. Bjelke (40) reported data for rectal cancer cases separately

in studies conducted both in Norway and in Minnesota using a

nutrient index based on l00-g portions. For rectal but not for

colon cancer, patients had “a lower consumption of fruits and

berries and, in particular, of vitamin C” in Norway (40). The

same was true in Minnesota where the vitamin C difference

between rectal cancer patients and control subjects achieved sta-

tistical significance.

Finally, Graham et al (1 15) found a significantly increased

risk (RR = 1 .6) among men eating raw vegetables infrequently

and an even stronger effect among women. The role offruit was

not examined.

Thus, there have been four statistically significant and two

suggestive results for vitamin C in rectal cancer and an additional

significant result for raw vegetables. 13-Carotene effects were

weaker and nonsignificant in all but one study.

Numerous studies have examined colon cancer or a combined

category. colorectal, that generally would contain predominantly

colon cancer cases. A few have found a statistically significant

protective effect of vitamin C. Heilbrun et al ( I 1 3) examined

102 colon cancer patients in Hawaii. Vitamin C intake was sig-

nificantly lower among those who developed colon cancer over

the subsequent 16-y period (92 mg/d by 24-h recall) compared

with those who did not (1 16 mg/d). Those in the lowest quintile

of intake had 1 .9 times the risk of those in the upper quintile

(P = 0.01). In Australia, Kune et al(l07, 108)conducted a large

study of 7 15 colorectal cancer patients and matched community

at TE

L AV

IV U

NIV

ER

SIT

Y on N

ovember 21, 2011

ww

w.ajcn.org

Dow

nloaded from

2765 BLOCK

control subjects. Both dietary and supplemental vitamin C were

significantly protective with a threefold risk associated with the

low quintile ofdietary vitamin C after adjustment for other di-

etary factors. Consumption ofa high-fiber diet was also protective

provided that vegetable intake was also high. Macquart-Moulin

et al ( I 16) examined 399 cases ofcolorectal cancer in Marseilles

and matched injury controls. After adjustment for age, sex, ca-

brie intake, and body weight, a highly significant protective effect

was seen for vitamin C intake (RR = 1 .8 for low vs high quartile

of intake). A similar protective effect was seen for several mi-

cronutrients and minerals and for fiber from vegetables, but not

for vitamin A or fiber from flour.

Some investigators have found suggestive results in colon

cancer. La Vecchia et al ( I 14) observed a protective but non-

significant effect ofdietary ascorbic acid in colon cancer adjusted

for age and sex. However, tomatoes, green peppers, and melon

were highly significantly protective. The cutpoint for low intake

suggests a very high intake even in the lower tertile and little

dispersion between the low and high tertile. Bjelke (40, 41, 78)

found no effect for a vitamin C index in Norway. In Minnesota

he found vitamin C to be significantly lower not only in patients

with rectal cancer but also in patients with cancer of the left

colon (40). Although preliminary results by Tuyns (109) mdi-

cated a significant protective effect, a later analysis revealed no

consistent effects of a vitamin C index (1 1 1). Several foods rich

in vitamin C were, however, significantly protective (1 10).

Of the eight groups of investigators who have reported on a

dietary vitamin C index, two have found no evidence of pro-

tection for colon cancer. Potter and McMichael ( 1 12) found no

relationship with colon cancer in a study of 220 patients and

community control subjects in Australia. Canadian investigators

( 1 1 7, 1 1 8) found no evidence of a protective effect of either a

vitamin C index or of citrus fruit in a case-control study of 348

colon cancer patients and control subjects in Canada. Average

vitamin C intake among male cases in that study was 170 mg/

d as estimated by questionnaire. Fiber intake was also not pro-

tective.

Several investigations have reported protective effects of fruit

or vegetables but have not specifically reported on risks associated

with low vitamin C intake. Many have found protective effects

of vegetables, but evidence for an effect of fruit in colon cancer

is inconsistent. Modan et al (1 19) found that several vitamin C-

rich foods, including oranges. tomatoes, and green peppers, were

highly significantly protective. Slattery et al (120) also found

protection conferred by fruit as well as vegetables. Higginson

(92), in 1966, found a lower consumption offresh fruit in co-

lorectal cancer patients than in control subjects (not statistically

significant). Dales et al ( 12 1) found nonsignificant effects in the

protective direction for persons above the median on a fiber

index consisting of6 fruits, 14 vegetables, 3 grain products, and

5 legumes. Some fruits were protective in a study of colorectal

cancer in Hawaiian Japanese ( 122). In a large case-control study

in Wisconsin, Young and Wolf (123) found a significant pro-

tective effect for vegetables and for vitamin supplements, but

not for fruit. Similarly. Manousos et al (1 24) found significant

protection conferred by high intake of vegetables, but found no

effect of fruit. Graham et al (1 1 5) found a significant protective

effect for raw and cooked vegetables.

Finally, in a very small study which examined baseline plasma

ascorbate levels, St#{228}helinet al (53, 57, 84) found no difference

between I 4 colorectal cancer patients and their 32 control sub-

jects in the ascorbic acid level in plasma collected in 1965-1966

or 1971-1973. A later analysis of colorectal cancer patients

found lower plasma ascorbate in patients than in control subjects

(P = 0.06).

For colon cancer, as distinct from rectal cancer, the existing

evidence is less consistent. Four groups found a statistically sig-

nificant protective effect of vitamin C, two found suggestive but

nonsignificant effects, two found no effect ofthe nutrient index,

but one of these observed a significant effect of vitamin C-rich

foods. Overall, for both colon and rectal cancer, of the nine

groups that have reported on a vitamin C index, six have found

statistically significant results in one or the other site, two were

nonsignificant but in the protective direction, and one reported

no effect.

B/adder cancer

Several investigators demonstrated that ascorbic acid inhibits

nitrosation in humans and animals and inhibits carcinogen-in-

duced bladder tumors in animals (125-127). Following up on

this lead, Kolonel et al (54) examined 123 male and 41 female

patients with bladder cancer and population control subjects in

two ethnic groups in Hawaii. Vitamin C intake from supplements

was lower in patients than in control subjects in all four ethnic-

sex groups. The same was true for food sources among females

but not males. For the combined total intake from both sources,

patients averaged 389 mg/d and control subjects 5 13 mg/d. None

of the differences within the four individual ethnic-sex groups

was statistically significant; significance ofthe overall case-control

comparison was not reported. No effects were found for a vitamin

A index. La Vecchia et al (128) examined 163 bladder cancer

patients and hospital control subjects in northern Italy. The

questionnaire included three fruit/vegetable items: carrots, green

vegetables, and fruit. Although there was a significant association

with the former, there appeared to be no association between

case-control status and intake ofthis fruit item in this population.

It is notable that this population reported a high intake of fruit,

averaging more than once a day in both patients and control

subjects, and thus the possible effects of low intake cannot be

examined. Mettlin and Graham (129) and Paganini-Hill et al

(1 30) found an effect for vitamin A, but did not report on the

role ofvitamin C. Table I summarizes the epidemiologic studiesofvitamin C and cancer prevention for non-hormone-dependent

cancer sites.

Childhood brain tumors

and other nonepithelial cancers

Only one study has been conducted on dietary factors in this

cancer. Thomas Sinks and John R Wilkins III (personal corn-

munication, 1989) interviewed parents of 100 children with brain

tumors and 200 matched control subjects, regarding the mothers’

diets during pregnancy. A statistically significant threefold in-

creased risk of delivering a child who later developed brain tu-

mors was associated with low maternal intake ofvitamin C dur-

ing pregnancy, an effect that remained after adjustment for other

factors.

Breast, ovary, endometrium and prostate cancer

For ovarian, endometrial and prostate cancer, the little cvi-

dence that exists does not support an important role for vitamin

at TE

L AV

IV U

NIV

ER

SIT

Y on N

ovember 21, 2011

ww

w.ajcn.org

Dow

nloaded from

ASCORBIC ACID IN CANCER PREVENTION 2775

TABLE ISummary of epidemiologic studies of vitamin C and cancer prevention �

Non-hormone-dependent cancer sites References No of studies

No of statistically

significantly protective

Median

relative risk

All sites in this category See below Vitamin C index: 46

Fruit: 29

33

21Oral cavity (24-26)

(27 29, 32)

Vitamin C index: 3

Fruit: 4

3

3

2.0

1.7Larynx (31)

(32)

Vitamin C index: I

Fruit: 1

1

1

2.4

2.0

Esophagus (33, 35-37)(38-40, 32, 42-44)

Vitamin C index: 4Fruit: 7

4

3t

2.2

IDjLung (45-52, 60, 62)

(63, 64)

Vitamin C index: 10

Fruit: 2St2

1.6l.7�

Pancreas (67)

(68-7 1)

Vitamin C index: 1

Fruit: 4

1

4

2.2

1.6Stomach or precursors (40, 41, 78-83, 94)

(84-87, 90)

Vitamin C index: 7

Fruit: 5

7

5 II2.0�

2.5

Cervix or precursors (61, 96, 98, 99) Vitamin C index: 4 3 2.0Rectum (40, 107, 108) Vitamin C index: 6 4 1.5Colon or colorectal (40, 107, 1 1 1-1 14, 1 16,

(119, 120, 122-124)

1 17) Vitamin C index: 8

Fruit:543

1.11.7

Bladder (54)(128)

Vitamin C index: 1Fruit: I

00

IDID

Brain (Sinks, personalcommunication, 1989)

Vitamin C index: I I 3.0

S Studies that assessed the role of a dietary vitamin C index or of fruit intake. Studies that reported on the role of both are included only under

“Vitamin C Index,” even ifthe fruit effect was stronger or more statistically significant. Because sample sizes in subgroups were sometimes small, in

some cases results are classified as statistically significant ifa major subgroup (eg, males) produced significance. Significance is defined as P < 0.05;in many studies levels of P < 0.01 or stronger were observed, and some of these are noted in the text. Median relative risk refers to the median over

all studies in the group, not simply the statistically significant ones.

t See text.t ID, Insufficient data.

§ Estimate based on only those for which relative risk was reported.

II In one study, P = 0.07, n = 19.

C, whereas recent analyses suggest an important role in breastcancer. Slattery et al (13 1) studied 85 women with ovarian cancer

and population control subjects. An effect in the protective di-

rection was observed for those in the lower one-third of the

distribution (OR = 1.4). However, this did not achieve statistical

significance. La Vecchia et al (132, 133) have reported two studies

of ovarian cancer in Italy. Both found a statistically significant

protective effect for more frequent consumption oftwo vegetable

items, but no effect of reported consumption of a single item,

fruit. In a study in China, Shu et al (1 34) found no relationship

between vitamin C intake and ovarian cancer, although it is

notable that the lowest quartile consisted ofthose who consumed

� 68 mg. Finally, Byers et al (I 35) found no effect of vitamin

C as calculated from the 1957 Roswell Park questionnaire, but

did find a protection from vitamin A calculated from fruits and

vegetables. Thus, the existing studies do not provide evidence

ofa role for vitamin C in ovarian cancer. It is notable, however,

that two of these studies appeared to be in populations with a

very high intake of this nutrient: in Slattery et al (13 1), the low

intake category was defined as vitamin C < 98 mg/d, and in La

Vecchia et al (1 32) the study group consumed an average of 12

servings of fruit per week.

In a single study ofendometrial cancer, La Vecchia et al (136)

investigated 206 patients and age-matched hospital control sub-

jects in Italy. Patients reported significantly lower intake of green

vegetables than control subjects, and lower intake of fruit, al-

though not significantly so. Consumption of fruit in this pop-

ulation appears to be high with both patients and control subjects

averaging 1 #{189}servings/d.

Several investigations have suggested an increased risk for

prostate cancer associated with increased vitamin A and in some

instances vitamin C intake. Reports using the Roswell Park 1957questionnaire and the same study population (I 37, 138) found

an increased risk for increasing vitamin A and vitamin C intake,

trend significant for men aged > 70 y. Investigators in Hawaii

(54, 139, 140) have also reported in the same case series a sig-

nificantly increased risk associated with vitamin A, and the same

direction ofeffect for vitamin C, but not statistically significant.

Two similar studies in Washington, DC (141 , 142) assessed diet

from the distant past (when the patients and control subjects

were aged 30-49 y, and � 50 y). The results suggested an elevated

risk for increasing vitamin A intake, but not for vitamin C. A

prospective study ofJapanese men in Hawaii (143) found similar

results. Two analyses ofa case-control study in Japan (144, 145)

have found a protective effect of 13-carotene in contrast to the

above studies, but no evidence of an effect of vitamin C.

Other studies have found protective effects for increasing con-

sumption of fruits or vegetables or of carotenoid indices calcu-

lated from them, but either did not report on vitamin C or did

not use food lists designed to assess it (146- 148). Thus, with

at TE

L AV

IV U

NIV

ER

SIT

Y on N

ovember 21, 2011

ww

w.ajcn.org

Dow

nloaded from

2785 BLOCK

regard to prostate cancer and vitamin C or foods rich in vitamin

C. the evidence is meager and conflicting with little evidence of

an effect.

For breast cancer, Howe et al ( I 49- 154) recently examined a

series of studies in a major meta-analysis of the role of dietary

factors. Saturated fat was positively associated with breast cancer,

but in addition “vitamin C intake had the most consistent and

statistically significant inverse association with breast cancer

risk.” When 13-carotene, fiber, and vitamin C were examined

simultaneously. only the latter remained significant, and the au-

thors consequently used vitamin C as the marker for the effective

agent(s) in fruit and vegetables. The results for vitamin C re-

mained after control for fat intake. In terms ofattributable risk,

the authors conclude that “ifall postmenopausal women in the

population modify their saturated fat intake to (that ofthe lower

one-fifth of the population), the current rate of breast cancer

would be reduced by 10% in postmenopausal women in North

America. . . . If all postmenopausal women . . . were to in-

crease fruit and vegetable intakes to reach an average daily con-

sumption of vitamin C (equivalent to that of the highest one-

fifth of the population), risk of breast cancer . . . would be re-

duced by 16%.” The effects were approximately additive, and

simultaneously making both changes would reduce the risk by

24%. The authors also point out that misclassification of dietary

intake would lead to underestimation of the relative risk and

therefore of the attributable risk. Thus, for breast cancer the

protective effect of vitamin C appears to be very consistent in

the studies examined by Howe et al ( 149- 1 54), and of a mag-

nitude at least equal to that ofsaturated fat. Table 2 summarizes

the epidemiologic studies of vitamin C and cancer prevention

for hormone-dependent cancer sites.

Summary

In the I 1 non-hormone-dependent cancer sites described

above, 46 studies have specifically reported on a vitamin C index

or plasma ascorbate values; 33 of these found statistically sig-

nificant protective effects, and several more were in the protective

direction but did not achieve significance. None has found el-

evated risk with increasing intake. In addition to those, 29 studies

reported on the effect of fruit consumption, 2 1 of which found

significant protection associated with frequent consumption or

high risk associated with low consumption. For oral, esophageal,

gastric, and pancreatic cancer, the evidence is extremely strong,

with virtually all studies showing a significant protective effect.

The cervical and rectal cancer data are also very strong, with

only a few studies failing to find significant protection. In lung

cancer, five recent studies observed statistically significant pro-

tection, although several earlier studies found nonsignificant re-

sults. In breast cancer, there appears to be a very consistent

protective effect, in the meta-analysis by Howe et al (149). How-

ever, none ofthe studies ofovarian or prostate cancer have found

a statistically significant protective effect of vitamin C.

Dietary estimates involve uncertainties in nutrient content of

foods, in degree of nutrient destruction in cooking and storage,

in portion size quantification, and in the individual’s ability to

estimate his or her frequency of consumption, as well as sub-

sequent altering of blood and tissue levels by host factors. All

ofthese result in substantial misclassification ofthe individual’s

true nutrient status. Misclassification results in a severe bias of

risk estimates toward the null. A great deal has been made of

this problem in discussions of the effects of fat intake, in which

risk estimates have frequently been found to be low and non-

significant. In studies reported here for several cancer sites, how-

ever, particularly the epithelial cancers, the risk estimates have

frequently been substantial and statistically significant. The fact

that misclassification undoubtedly exists can only mean that the

observed effects reported here are underestimates of the true

risk.

These results appear suggestive enough to warrant more in-

tensive investigations in the future. Few studies have examined

plasma vitamin C values and none has examined leukocyte vi-

tamin C values and their relationships with cancer. Leukocyte

values would be preferable because they represent a target tissue

for ascorbate concentration and they are not affected by very

recent dietary intake. Such analyses are needed in prospective

studies or case-control studies nested in a prospective cohort.

These values represent what is actually available to the body for

TABLE 2Summary of epidemiologic studies of vitamin C and cancer prevention

Hormone-dependent

cancer sites References No studies

No statistically

significantly protective

Medianrelative risk

All sites in this category See below Vitamin C index: 9

Fruit: 3

Ot

1

Ovary (131, 134, 135)(132)

VitaminCindex:3Fruit: 1

0j

0

1.3

ID.

Endometrium (136) Fruit: 1 1 1.8

Prostate (137, 139, 141-144) VitaminCindex:6 0� 0.8

Breast (149-154) VitaminCindex:9 Seetextt 1.4

6 Studies that assessed the role of a dietary vitamin C index or of fruit intake. Studies that reported on the role of both are included only under

“Vitamin C Index,” even ifthe fruit effect was stronger or more statistically significant. Because sample sizes in subgroups were sometimes small, in

some cases results are classified as statistically significant if a major subgroup (eg, males) produced significance. Significance is defined as P < 0.05;

in many studies levels of P < 0.0 1 or stronger were observed, and some of these are noted in the text. Median relative risk refers to the median over

all studies in the group. not simply the statistically significant ones.

t Howe et al ( 149), in a meta-analysis, found vitamin C to be consistently inversely associated with breast cancer risk, in nine studies examined.See text.

� None were significant in either the protective or harmful direction.§ Two studies found significantly elevated risk with high intake, in some age subgroups or control comparisons.

at TE

L AV

IV U

NIV

ER

SIT

Y on N

ovember 21, 2011

ww

w.ajcn.org

Dow

nloaded from

ASCORBIC ACID IN CANCER PREVENTION 2795

antioxidant or other functions. Dietary data, on the other hand,

are severely hampered by inaccuracies in food table values, loss

of ascorbate in foods as a result of storage and cooking. and

other factors comprising the chasm ofassumptions between the

estimates of dietary intake and the level of ascorbate actually

reaching the tissues. Very important among these is the role of

other characteristics ofthe individual that alter blood and tissue

levels after the food is consumed. Smoking dramatically lowers

blood ascorbate levels, as does aspirin consumption, oral con-

traceptives, acute and chronic diseases, and a variety of other

stresses. The fact that smoking is usually included in multivariate

models adds another complication to the interpretation, since

they may be in the same causal pathway-smoking may exert

part of its effect by virtue of its reduction of plasma ascorbate

levels. Because of factors such as these, it would appear that

clarification ofthe role ofvitamin C in cancer prevention would

be greatly enhanced by studies that examine blood levels, care-

fully prepared and stored, in relation to cancer outcome.

In addition to assessing blood levels, some additional meth-

odologic features might be considered in future investigations.

First, sample size has been inadequate in several studies described

here; even when apparently adequate, investigators often divide

their sample into quartiles or quintiles and then impose control

for numerous confounders [eg, eight in Brock et al (98). each

with several levels] on each of the quantiles. The result is cx-

tremely unstable estimates and poor statistical power. Prestudy

sample size and power calculations should be based on the real

analyses that will be performed. Second, estimates of the mean

nutrient intake and standard deviation based on several 24-h

recalls or records should be collected in a subset of the study

population. This would permit correction of the mean and dis-

tribution obtained in the food frequency questionnaire so that

the nutrient intake corresponding to the quantile cutpoints could

be more accurately estimated. If this were done, it might ulti-

mately be possible to compare studies in different populations

and with different instruments and to estimate risks for levels

of actual nutrient intake. Third, analyses should take into ac-

count effect modification rather than simply confounding by

other factors. For example, the effect ofvitamin C (or /3-carotene)

may only exist when fat (or meat) intake is high or low. Ross et

al ( 147) found a suggestion that 13-carotene was only protective

for prostate cancer when fat intake was low, and Heilbrun et al

(1 1 3) found similar results with regard to the protective effect

ofdietary fiber in colon cancer at different fat levels. Joint effects

oftwo variables and effect modifications such as these have rarely

been examined, but may explain some ofthe differences in results

seen in different population groups. Future investigators may

wish to design their studies with sample sizes adequate to examine

their data in this way.

Because other nutrients, especially carotenoids and folic acid,

often are obtained in the same foods, with dietary data we cannot

be completely certain that the effect is due to vitamin C and not

to other factors. Nevertheless, the strength and consistency of

the results reported here for several sites suggests that there may

be a real and important effect ofascorbic acid in cancer preven-

tion. It may be more productive to stop thinking in terms of

“or”-”Is it vitamin C or carotenoids?” It is very likely that

both are needed, that all the nutrients packaged together in fruitsand vegetables are synergistic and provide optimal benefit when

all are present in optimal quantities. Vitamin C is apparently

the first line ofantioxidant defense (7), spares vitamin E (8-14),

may be an effective radical scavenger under different partial

pressures of oxygen than those that are optimal for 13-carotene.

and may act synergistically with other biologic antioxidants and

radical scavengers in quenching different elements of a radical

cascade ( 1 55). Future research should focus on analyses of risks

in individuals in which two or more nutrients are low or high.

Public health action should be directed towards increasing the

consumption of fruit, as well as vegetables, in which nature has

packaged a variety of protective nutrients.

I am indebted to the following people for their skillful and committedassistance and support for this paper Lisa Carter, Susana Rosales. VickiHoffman, and Marilyn Apfel.

References

I . Hodges RE. Ascorbic acid. In: Goodhart RS, Shils ME. eds. Modern

nutrition in health and disease. 6th ed. Philadelphia: Lea and Fe-biger, 1978:259-73.

2. Sauberlich HE. Ascorbic acid. In: Olson RE, Broquist HP. Chi-chester CO. Darby WI, Kolbye AC, Stalvey RM, eds. Nutrition

reviews’ present knowledge in nutrition. 5th ed. Washington, DC:The Nutrition Foundation, Inc. 1984:260-72.

3. Levine M. New concepts in the biology and biochemistry of ascorbicacid. N Fngl I Med l986;3l4:892-902.

4. Block G, Menkes M. Ascorbic acid in cancer prevention. In: MoonTE, Micozzi MS. eds. Diet and cancer prevention. Investigating

the role of micronutrients. New York: Marcel Dekker, Inc. 1989:

34 1-88.

5. Hanck AB. Vitamin C and cancer. Prog Clin Biol Res 1988:259:307-20.

6. Frei B, Stocker R, Ames BN. Antioxidant defenses and lipid per-

oxidation in human blood plasma. Proc NatI Acad Sci USAI 988:85:9748-52.

7. Frei B, England L, Ames BN. Ascorbate is an outstanding antiox-idant in human blood plasma. Proc NatI Acad Sci USA 1989:86:

6377-8 1.8. Tappel AL. Vitamin E as the biological lipid antioxidant. Vitam

Horm l962;20:493-S 10.

9. Menzel DB. Toxicity of ozone, oxygen, and radiation. Annu Rev

Pharmacol I 970; 10:379-94.10. Packer IF, Slater TF, Willson RL. Direct observation of a free

radical interaction between vitamin F and vitamin C. NatureI 979:278:737-8.

I 1. Chen LH, Chang ML. Effect ofdietary vitamin F and vitamin C

on respiration and swelling of guinea pig liver mitochondria. INutr l978;108:l6l6-20.

12. Chen LH, Lee MS. Hsing WF, Chen SH. Effect of vitamin C on

tissue antioxidant status of vitamin E deficient rats. Int J Vitam

Nutr Res I 980;50: 1S6-62.I 3. Barclay LRC, Locke SI, MacNeil IM. The autoxidation of unsat-

urated lipids in micelles. Synergism ofinhibitors vitamin C and E.Can I Chem 1983:61:1288-9.

14. Hatch GE, Slade R, Selgrade MK, Stead AG. Nitrogen dioxide

exposure and lung antioxidants in ascorbic acid-deficient guineapigs. Toxicol AppI Pharmacol 1986;82:35 I -9.

15. Anderson R. Ascorbic acid and immune functions: mechanism ofimmunostimulation. In: Counsell IN, Hornig DH. eds. Vitamin

C (ascorbic acid). London: Applied Science Publishers. 198 1:249-72.

16. Bendich A. Vitamin C and immune responses. Food Technoll987;4l:l 12-4.

I 7. Anderson R, Oosthuizen R, Maritz R, Theron A, Van Rensburg

Al. The effects of increasing weekly doses of ascorbate on certain

at TE

L AV

IV U

NIV

ER

SIT

Y on N

ovember 21, 2011

ww

w.ajcn.org

Dow

nloaded from

2805 BLOCK

cellular and humoral immune functions in normal volunteers. AmJ Clin Nutr 1980:33:71-6.

18. Siegel BV. Enhanced interferon response to murine leukemia virus

by ascorbic acid. Infect Immun 1974:10:409-10.

19. Johnston CS. KoIb WP, Haskell BE. The effect of vitamin C flu-triture on complement component Clq concentrations in guineapig plasma. J Nutr 1987:1 17:764-8.

20. Weiss L, Greep RO. Histology. New York: McGraw-Hill, 1977.

2 1 . Cameron E, Pauling L. Ascorbic acid and the glycosaminoglycans:an orthomolecular approach to cancer and other diseases. Oncology

1973:27: 18 1-92.22. Cameron E, Pauling L. The orthomolecular treatment of cancer.

I. The role of ascorbic acid in host resistance. Chem Biol Interact1974:9:273-83.

23. Block G, Hartman AM. Dietary assessment methods. In: Moon

TE. Micozzi MS. eds. Nutrition and cancer prevention. Investigating

the role of micronutrients. New York: Marcel Dekker, Inc. I 989:159-80.

24. Rossing MA, Vaughan T, McKnight B. Diet and pharyngeal cancer.Am J Epidemiol 1989:130:799 (abstr).

25. McLaughlin JK, Gridley G, Block G, et al. Dietary factors in oral

and pharyngeal cancer. JNCI 1988:80:1237-43.26. Marshall I. Graham S. Mettlin C, Shedd D. Swanson M. Diet in

the epidemiology of oral cancer. Nutr Cancer 1982:3:14S-9.27. Franco EL, Kowalski LP, Oliveira BV, et al. Risk factors for oral

cancer in Brazil: a case-control study. Int I Cancer 1989:43:992-

1000.28. Winn DM, Ziegler RG, Pickle LW, Gridley G, Blot WI, Hoover

RN. Diet in the etiology of oral and pharyngeal cancer amongwomen from the southern United States. Cancer Res 1984:44:1216-.,,

29. Wynder EL, Bross Ii, Feldman RM. A study of the etiologicalfactors in cancer of the mouth. Cancer 1957: 10: 1300-2 1.

30. Graham 5, Dayal H, Rohrer T, et al. Dentition, diet, tobacco, and

alcohol in the epidemiology of oral cancer. J NatI Cancer Inst

1977:59:161 1-8.3 1. Graham 5, Mettlin C. Marshall I, Priore R, Rzepka T, Shedd D.

Dietary factors in the epidemiology ofcancer ofthe larynx. Am IEpidemiol 1981:1 13:675-80.

32. Notani PN, Jayant K. Role of diet in upper aerodigestive tractcancer. Nutr Cancer 1987:10:103-13.

33. Brown LM, Blot WI, Schuman SH, et al. Environmental factorsand high risk of esophageal cancer among men in coastal South

Carolina. INCI l988;80: 1620-5.34. Tuyns Al. Protective effect of citrus fruit on esophageal cancer.

Nutr Cancer 1983:5:195-200.

35. Tuyns Al, Riboli F. Doornbos G. Pequignot G. Diet and esophagealcancer in Calvados (France). Nutr Cancer 1987:9:81-92.

36. Ziegler RG, Moms LE, Blot WI, Pottern LM. Hoover R, FraumeniIF. Esophageal cancer among black men in Washington, DC. II.Role ofnutrition. INCI 1981:67:1 199-206.

37. Mettlin C, Graham 5, Priore R, Marshall I, Swanson M. Diet and

cancer of the esophagus. Nutr Cancer 1980;2: 143-7.38. Cook-Mozaffari P1, Azordegan F, Day NE, Ressicaud A, Sabai C,

Aramesh B. Oesophageal cancer studies in the Caspian littoral ofIran: results ofa case-control study. Br I Cancer 1979:39:293-309.

39. Hirayama T. Epidemiological evaluation of the role of naturally

occurring carcinogens and modulators ofcarcinogenesis. In: Miller

FC, Miller IA, Hirono I, Sugimura T, Takayama 5, eds. Naturally

Occurring Carcinogens-Mutagens and Modulators of Carcinogen-esis. Tokyo: Japan Scientific Societies Press, 1979:359-80.

40. Bjelke E. Epidemiologic studies of cancer of the stomach, colon,

and rectum: with special emphasis on the role of diet. Vols I-IV.Doctoral dissertation, 1973. Ann Arbor, MI: University Microfilms

International, 1973.4 1 . Bjelke E. Case-control study of cancer of the stomach, colon, and

rectum. In: Clark RL, Cumley RW, McCay IF, Copeland MM,

eds. Oncology 1970: Being the Proceedings of the Tenth Interna-

tional Cancer Congress. Volume V. A. Environmental causes. B.Epidemiology and demography. C. Cancer education. Chicago:

Yearbook Medical Publishers, Inc, 197 1:320-34.

42. Li I-Y, Ershow AG, Chen Z-J, et al. A case-control study of cancerofthe esophagus and gastric cardia in Linxian. Int I Cancer l989;43:

755-6 1.43. Martinez I. Factors associated with cancer ofthe esophagus, mouth,

and pharynx in Puerto Rico. I NatI Cancer Inst l969;42:l069-94.44. Decarli A, Liati P, Negri E, Franceschi S. La Vecchia C. Vitamin

A and other dietary factors in the etiology of esophageal cancer.Nutr Cancer 1987;lO:29-37.

45. Fontham ETH, Pickle LW, Haenszel W, Correa P. Lin Y, FalkRT. Dietary vitamins A and C and lung cancer risk in Louisiana.

Cancer l988;62:2267-73.46. Koo LC. Dietary habits and lungcancer risk among Chinese females

in Hong Kong who never smoked. Nutr Cancer l988;ll:l55-72.47. HoIst PA, Kromhout D, Brand R. For debate: pet birds as an in-

dependent risk factor for lung cancer. Br Med J I 988;297: 1319-21.

48. Le Marchand L, Yoshizawa CN, KOIOITeI-LN, Hankin JH, Good-man MT. Vegetable consumption and lung cancer risk: a popu-lation-based case-control study in Hawaii. JNCI l989;8l:l 158-64.

49. Kromhout D. Essential micronutrients in relation to carcinogenesis.Am I Clin Nutr l987;45:l36l-7.

50. Hinds MW, Kolonel LN, Hankin JH, Lee I. Dietary vitamin A,

carotene, vitamin C and risk of lung cancer in Hawaii. Am I Ep-idemiol l984;1 19:227-37.

5 1. Kvale 0, Bjelke E, Gart II. Dietary habits and lung cancer risk.

Int J Cancer l983;3l:397-405.52. Shekelle RB, Lepper M, Liu 5, et al. Dietary vitamin A and risk

of cancer in the Western Electric study. Lancet 198l;2: 1 185-90.53. St#{228}helinHB, Gey KF, Brubacher 0. Plasma vitamin C and cancer

death: the Prospective Basel Study. Ann NY Acad Sci l987;498:124-31.

54. Kolonel LN, Hinds MW, Nomura AMY, Hankin JH, Lee I. Re-lationship ofdietary vitamin A and ascorbic acid intake to the risk

for cancers ofthe lung, bladder, and prostate in Hawaii. Natl CancerInst Monogr l985;69:l37-42.

55. Block G, Cox C, Madans I, Schreiber GB, Licitra L, Melia N.Vitamin supplement use, by demographic characteristics. Am I

Epidemiol 1988; 127:297-309.56. Bjelke F. Dietary vitamin A and human lung cancer. Int I Cancer

l975;l5:56l-5.

57. Stahelin HB, Rosel F, Buess F, Brubacher G. Cancer, vitamins, and

plasma lipids: Prospective Basel Study. JNCI l984;73:l463-8.58. St#{228}helinHB, Gey KF, Eichholzer M, LUdin E. /3-carotene and

cancer prevention: The Basel Study. Am I Clin Nutr 199 l;53(suppl):2655-95.

59. Mettlin C, Graham 5, Swanson M. Vitamin A and lung cancer.

JNCI l979;62: 1435-8.60. Byers T, Vena I, Meulin C, Swanson M, Graham S. Dietary vitamin

A and lung cancer risk: an analysis by histologic subtypes. Am I

Epidemiol 1984; 120:769-76.61. Marshall JR. Graham S. Byers T, Swanson M, Brasure I. Diet and

smoking in the epidemiology ofcancer ofthe cervix. JNCI l983;70:

847-51.62. Byers TE, Graham 5, Haughey BP, Marshall JR. Swanson MK.

Diet and lung cancer: findings from the Western New York diet

study. Am I Epidemiol l987;l25:35l-63.

63. Long-dc W, Cuyler Hammond F. Lung cancer, fruit, green salad

and vitamin pills. Chin Med I l985;98:206-lO.64. Bond GO, Thompson FE, Cook RR. Dietary vitamin A and lung

cancer: results of a case-control study among chemical workers.

Nutr Cancer l987;9:109-2l.

65. MacLennan R, Da Costa I, Day NE, Law CH, Ng YK, Shanmu-

garatnam K. Risk factors for lung cancer in Singapore Chinese, a

at TE

L AV

IV U

NIV

ER

SIT

Y on N

ovember 21, 2011

ww

w.ajcn.org

Dow

nloaded from

ASCORBIC ACID IN CANCER PREVENTION 28 1S

population with high female incidence rates. Int I Cancer l977;20:

854-60.

66. Hirayama T. A large scale cohort study on cancer risks by diet-with special reference to the risk reducing effects of green-yellowvegetable consumption. In: Hayashi Y, Nagao M, Sugimura T, et

al, eds. Diet, nutrition and cancer. Proceedings of the 16th Inter-national Symposium ofThe Princess Takamatsu Cancer Research

Fund, Tokyo, 1985. Tokyo: Japan Scientific Societies Press, 1986:4 1-53.

67. Falk RT, Pickle LW, Fontham ET, Correa P. Fraumeni IF. Life-style risk factors for pancreatic cancer in Louisiana: a case-control

study. Am I Epidemiol 1988;128:324-36.68. Norell SE, Ahlbom A, Erwald R, et al. Diet and pancreatic cancer:

a case-control study. Am I Epidemiol l986;l24:894-902.69. Mack TM, Yu MC, Hanisch R, Henderson BE. Pancreas cancer

and smoking, beverage consumption, and past medical history.JNCI l986;76:49-60.

70. Gold EB, Gordis L, Diener MD, et al. Diet and other risk factorsfor cancer of the pancreas. Cancer l985;55:460-7.

7 1 . Mills PK, Beeson WL, Abbey DE, Fraser GE, Phillips RL. Dietaryhabits and past medical history as related to fatal pancreas cancerrisk among Adventists. Cancer l988;6l :2578-85.

72. National Cancer Institute. SEER Program: Cancer incidence and

mortality in the United States 1973-81. Bethesda, MD: National

Institutes of Health, 1984. (NIH Publication No. 85-1837.)73. Kolonel LN, Nomura AMY, Hirohata T, Hankin IH, Hinds MW.

Association ofdiet and place ofbirth with stomach cancer incidence

in Hawaii Japanese and Caucasians. Am I Gin Nutr 198 l;34:2478-85.

74. Correa P. Cuello C, Fajardo LF, Haenszel W, Bolanos 0, de Ra-

mirez B. Diet and gastric cancer� nutrition survey in a high-risk

area. INCI I 983;70:673-8. -

75. Burr ML, Samloff lM, Bates CI, Holliday RM. Atrophic gastritis

and vitamin C status in two towns with different stomach cancerdeath-rates. BrJ Cancer l987;56:l63-7.

76. Rathbone BJ, Johnson AW, Wyatt II, Kelleher I, Heatley RV,Losowsky MS. Ascorbic acid: a factor concentrated in human gastricjuice. Clin Sci 1989;76:237-41.

77. Sobala GM, Schorah CJ, Sanderson M, et al. Ascorbic acid in thehuman stomach. Gastroenterology l989;97:357-63.

78. Bjelke F. Epidemiology ofcolorectal cancer, with emphasis on diet.In: Davis W, Harrap KR, Stathopoulos 0, Maltoni C, eds. Human

cancer. Its characterization and treatment. Advances in tumourprevention, detection and characterization. Vol 5. Proceedings of

the Eighth International Symposium on the Biological Character-ization of Human Tumours. Athens, May 8- 1 1, 1979. Princeton:Excerpta Medica, 1980:1S8-74.

79. Correa P. Fontham E, Pickle LW, Chen V. Lin Y, Haenszel W.Dietary determinants of gastric cancer in south Louisiana inhab-

itants. JNCI 1985;75:645-54.80. You W-C, Blot WI, Chang Y-S, et al. Diet and high risk of stomach

cancer in Shandong, China. Cancer Res 1988;48:3518-23.81. La Vecchia C, Negri E. Decarli A, D’Avanzo B. Franceschi S. A

case-control study ofdiet and gastric cancer in northern Italy. Int

I Cancer l987;40:484-9.82. Risch HA, lain M, Choi NW, et al. Dietary factors and the incidence

ofcancer of the stomach. Am I Epidemiol l985;l22:947-59.83. Meinsma L. Nutrition and cancer. Voeding 1964;25:357-65 (in

German).84. Stahelin HB, Rosel F, Buess E, Brubacher G. Dietary risk factors

for cancer in the Basel Study. Bibl Nutr Dicta l986;37: 144-53.

85. Kono S. Ikeda M, Tokudome S. Kuratsune M. A case-control study

ofgastric cancer and diet in Northern Kyushu, Japan. Ipn I CancerRes 1988:79:1067-74.

86. Iedrychowski W, Wahrendorf I, Popiela 1, Rachtan I. A case-control study ofdietary factors and stomach cancer risk in Poland.Int I Cancer 1986:37:837-42.

87. Trichopoulos D, Ouranos G, Day NE, et al. Diet and cancer of

the stomach: a case-control study in Greece. Int I Cancer 1985:36:29 1-7.

88. Graham S. Schotz W, Martino P. Alimentary factors in the epi-

demiology ofgastric cancer. Cancer l972;30:927-38.89. Graham S. Lilienfeld AM, Tidings IF. Dietary and purgation factors

in the epidemiology ofgastric cancer. Cancer 1967:20:2224-34.

90. Coggon D, Barker DIP, Cole RB, Nelson M. Stomach cancer andfood storage.JNCI l989;81:l 178-82.

91. Haenszel W, Kurihara M, Segi M, Lee RKC. Stomach cancer

among lapanese in Hawaii. I NatI Cancer Inst 1972;49:969-88.92. Tajima K, Tominaga S. Dietary habits and gastro-intestinal cancers:

a comparative case-control study of stomach and large intestinal

cancers in Nagoya, Japan. Ipn I Cancer Res 1985:76:705-16.93. Higginson I. Etiological factors in gastrointestinal cancer in man.

I Natl Cancer Inst l966;37:527-45.94. Fontham F, Zavala D, Correa P. et al. Diet and chronic atrophic

gastritis: a case-control study. INCI l986;76:62l-7.95. Haenszel W, Cuello C, Correa P. Lopez-S A, Zarama G, Zavala

D. Correlations of values of micronutrients in sera with gastric

pathology. NatI Cancer Inst Monogr I 985:69: 1 15-9.96. Wassertheil-Smoller S. Romney SL, Wylie-Rosett I, et al. Dietary

vitamin C and uterine cervical dysplasia. Am I Epidemiol 1981:1 14:

7 14-24.

97. Romney SL, Duttagupta C, Basu I, et al. Plasma vitamin C and

uterine cervical dyplasia. Am I Obstet Gynecol 198�:1�1:976-80.98. Brock KE, Berry G, Mock PA, MacLennan R, Truswell AS, Brinton

LA. Nutrients in diet and plasma and risk ofin situ cervical cancer.JNCI 1988;80:580-5.

99. Verreault R, Chu I, Mandelson M, Shy K. A case-control study

ofdiet and invasive cervical cancer. Int I Cancer 1989:43:1050-4.

100. DeCosse II, Adams MB, Kuzma IF, LoGerfo P. Condon RE. Effectofascorbic acid on rectal polyps ofpatients with familial polyposis.

Surgery 1975;78:608-12.

101. Bussey HJR, DeCosse II, Deschner FE, et al. A randomized trialof ascorbic acid in polyposis coli. Cancer 1982:50:1434-9.

102. McKeown-Fyssen GE, Holloway C. Iazmaji V. Dion P. Bruce WR.A randomized trial of vitamin C and F supplementation in the

prevention of recurrence of colorectal polyps. Prey Med 1987: I 6:

275(abstr).

103. McKeown-Eyssen 0, Holloway C. Iazmaji V. Bright-See F, Dion

P. Bruce WR. A randomized trial of vitamins C and F in the pre-vention of recurrence of colorectal polyps. Cancer Res 1988;48:

4701-5.

104. DeCosse II, Miller HH, Lesser ML. Effect of wheat fiber and vi-

tamins C and F on rectal polyps in patients with familial adeno-matous polyposis. JNCI l989;8 1:1290-7.

105. Dion PW, Bright-See EB, Smith CC, Bruce WR. Theeffect of dietary

ascorbic acid and alpha-tocopheral on fecal mutagenicity. MutatRes l982;l02:27-37.

106. Schiffman MM. Diet and faecal genotoxicity. Cancer Surv 1987:6:

653-72.

107. Kune GA, Kune S. The nutritional causes ofcolorectal cancer: anintroduction to the Melbourne Study. Nutr Cancer 1987:9: I -4.

108. Kune 5, Kune GA, Watson LF. Case-control study ofdietary eti-ological factors: the Melbourne colorectal cancer study. Nutr Cancer

l987;9:21-42.109. Tuyns Al. A case-control study on colorectal cancer in Belgium.

Preliminary results. Soz Praventivmed 1986:31:81-2.

I 10. Tuyns Al. Kaaks R, Haelterman M. Colorectal cancer and the

consumption offoods: a case-control study in Belgium. Nutr Cancer1988;! 1:189-204.

I 1 1 . Tuyns AJ, Haelterman M, Kaaks R. Colorectal cancer and the

intake of nutrients: oligosaccharides are a risk factor. fats are not.A case-control study in Belgium. Nutr Cancer 1987:10:181-96.

1 I 2. Potter ID, McMichael AJ. Diet and cancer ofthe colon and rectum:

a case-control study. INCI l986;76:557-69.

at TE

L AV

IV U

NIV

ER

SIT

Y on N

ovember 21, 2011

ww

w.ajcn.org

Dow

nloaded from

282S BLOCK

1 13. Heilbrun LK, Nomura A. Hankin IH, Stemmermann ON. Dietand colorectal cancer with special reference to fiber intake. Int I

Cancer 1989:44:1-6.

I 14. La Vecchia C, Negri F, Decarli A, et al. A case-control study of

diet and cob-rectal cancer in northern Italy. Int I Cancer I 988:41:492-8.

I I 5. Graham S. Dayal H, Swanson M. Mittelman A. Wilkinson G. Dietin the epidemiology of cancer of the colon and rectum. INCI

1978:6 1:709- I4.

I 16. Macquart-Moulin G, Riboli F, Cornee I, Charnay B, BerthezeneP. Day N. Case-control study on colorectal cancer and diet in Mar-

seilles. lnt I Cancer 1986:38:183-91.

I 17. Jam M. Cook GM. Davis FG, Grace MG, Howe OR, Miller AB.A case-control study of diet and cob-rectal cancer. Int I Cancer

1980:26:757-68.I 18. Miller AB, Howe GR, Jam M, Craib KJP, Harrison L. Food items

and food groups as risk factors in a case-control study ofdiet andcob-rectal cancer. Int I Cancer I 983:32: 155-6 1.

1 19. Modan B. Cuckle H. Lubin F. A note on the role ofdietary retinol

and carotene in human gastro-intestinal cancer. Int I Cancer198 l;28:42l-4.

120. Slattery ML. Sorenson AW, Mahoney AW, French TK, Kritchevsky

D. Street IC. Diet and colon cancer: assessment of risk by fibertype and food source. INCI 1988:80:1474-80.

121. Dales LG, Friedman GD, Ury HK, Grossman 5, Williams SR. A

case-control study of relationships of diet and other traits to co-

lorectal cancer in American blacks. Am I Epidemiol 1979:109:132-44.

122. Haenszel W, Berg JW, Segi M. Kurihara M, Locke FB. Large-bowel cancer in Hawaiian lapanese. J NatI Cancer Inst 1973:51:

1765-79.123. Young TB. Wolf DA. Case-control study of proximal and distal

colon cancer and diet in Wisconsin. Int I Cancer 1988:42:167-75.

124. Manousos 0. Day NE. Trichopoulos D, Gerovassilis F. TzonouA. Polychronopoulou A. Diet and colorectal cancer: a case-control

study in Greece. Int I Cancer 1983:32:1-5.I 25. Pipkin GE. Schlegel IU, Nishimura R, Shultz GN. Inhibitory effect

of L-ascorbate on tumor formation in urinary bladders implantedwith 3-hydroxyanthranilic acid. Proc Soc Exp Biol Med 1969:131:

522-4.126. Schlegel JU. Pipkin GE, Nishimura R, Duke GA. Studies in the

etiology and prevention of bladder carcinoma. I Urol l969;lOl:3 17-24.

127. Schlegel IU. Pipkin GE. Nishimura R. Shultz GN. The role of

ascorbic acid in the prevention ofbladder tumor formation. TransAm Assoc Genitourin Surg 1969:61:85-9.

128. La Vecchia C. Negri E, Decarli A. D’Avanzo B. Liberati C, Fran-

ceschi S. Dietary factors in the risk ofbladder cancer. Nutr Cancer1989:12:93-101.

129. Mettlin C, Graham S. Dietary risk factors in human bladder cancer.Am J Epidemiol 1979:1 10:255-63.

130. Paganini-Hill A, Chao A. Ross RK. Henderson BE. Vitamin A,

beta-carotene. and the risk of cancer: a prospective study. INCII 987:79:443-8.

131. Slattery ML. Schuman KL. West DW. French TK. Robison LM.Nutrient intake and ovarian cancer. Am I Epidemiol 1989:130:

497-502.I 32. La Vecchia C, Decarli A. Negri E, et al. Dietary factors and the

risk ofepithelial ovarian cancer. INCI 1987:79:663-9.133. La Veechia C. Decarli A, Franceschi 5, Gentile A. Negri F, Parazzini

F. Dietary factors and the risk of breast cancer. Nutr Cancer

I987:10:205-14.

134. Shu XO, Gao YT, Yuan IM, Ziegler RG, Brinton LA. Dietary

factors and epithelial ovarian cancer. Br I Cancer l989;59:92-6.135. Byers T, Marshall I, Graham 5, Mettlin C, Swanson M. A case-

control study of dietary and nondietary factors in ovarian cancer.

INCI 1983:71:681-6.136. La Veechia C, Decarli A, Fasoli M, Gentile A. Nutrition and diet

in the etiology ofendometrial cancer. Cancer l986;57:l248-53.137. Graham S. Haughey B, Marshall I, et al. Diet in the epidemiology

ofcarcinoma ofthe prostate gland. JNCI l983;70:687-92.138. Middleton B, Byers T, Marshall I, Graham S. Dietary vitamin A

and cancer-a multisite case-control study. Nutr Cancer l986;8:

107- 16.139. Kolonel LN, Yoshizawa CN, Hankin IH. Diet and prostatic cancer

a case-control study in Hawaii. Am I Epidemiol l988;l27:999-

1012.

140. Kolonel LN, Hankin IH, Yoshizawa CN. Vitamin A and prostate

cancer in elderly men: enhancement of risk. Cancer Res l987;47:2982-5.

141 . Heshmat MY, Kaul L, Kovi I, et al. Nutrition and prostate cancer:a case-control study. Prostate l985;6:7- 17.

142. Kaul L, Heshmat MY, Kovi I, et al. The role ofdiet in prostatecancer. Nutr Cancer 1987:9:123-8.

143. Severson RK, Nomura AMY, Grove IS, Stemmermann GN. Aprospective study ofdemographics, diet, and prostate cancer among

men of Japanese ancestry in Hawaii. Cancer Res l989;49: 1857-

60.144. Ohno Y, Yoshida 0, Oishi K, Okada K, Yamabe H, Schroeder

FM. Dietary beta-carotene and cancer of the prostate: a case-control

study in Kyoto, Japan. Cancer Res l988;48:l33l-6.

145. Oishi K, Okada K, Yoshida 0, et al. A case-control study of prostatic

cancer with reference to dietary habits. Prostate l988;l2:l79-90.146. Shuman LM, Mandel IS, Radke A, Seal U, Halberg F. Some selected

features of the epidemiology of prostatic cancer Minneapolis-St.

Paul, Minnesota case-control study, 1976-1979. In: Magnus K,

ed. Trends in cancer incidence: causes and practical implications.Washington, DC: Hemisphere, 1982:345-54.

147. Ross RK, Shimizu H, Paganini-Hill A, Honda 0, Henderson BE.

Case-control studies of prostate cancer in blacks and whites insouthern California. JNCI 1987;78:869-74.

148. Ross RK, Paganini-Hill A, Henderson BE. The etiology of prostatecancer: what does the epidemiology suggest? Prostate l983;4:333-44.

149. Howe GR, Hirohata T, Hislop TG, et al. Dietary factors and risk

of breast cancer combined analysis of 12 case-control studies. JNCI1990;82:56 1-9.

150. Katsouyanni K, Willett W, Trichopoulos D, et al. Risk of breast

cancer among Greek women in relation to nutrient intake. Cancerl988;6 1: 18 1-5.

15 1. Rohan TE, McMichael Al, Baghurst PA. A population-based case-control study ofdiet and breast cancer in Australia. Am I Epidemiol1988; 128:478-89.

152. La Vecchia C, Decarli A, Franceschi S. Gentile A, Negii E, ParazziniF. Dietary factors and the risk of breast cancer. Nutr Cancer1987;lO:205-l4.

153. Brisson I, Verreault R, Morrison AS, Tennina 5, Meyer F. Diet,mammographic features of breast tissue, and breast cancer risk.

Am I Epidemiol I 989; 130:14-24.154. Toniolo P. Riboli F, Protta F, Charrel M, Cappa APM. Calorie-

providing nutrients and risk of breast cancer. JNCI l989;8l:278-

86.

155. Demopoulos HB. Control of free radicals in biologic systems. FedProc l973;32: 1903-8.

at TE

L AV

IV U

NIV

ER

SIT

Y on N

ovember 21, 2011

ww

w.ajcn.org

Dow

nloaded from