Full text of DOI:10.1207/S15327914nc391_3

Nutrition and Cancer
ISSN: 0163-5581 (Print) 1532-7914 (Online) Journal homepage: http://www.tandfonline.com/loi/hnuc20
Intakes of Selected Nutrients and Food Groups
and Risk of Ovarian Cancer
Susan E. McCann , Kirsten B. Moysich & Curtis Mettlin
To cite this article: Susan E. McCann , Kirsten B. Moysich & Curtis Mettlin (2001) Intakes of
Selected Nutrients and Food Groups and Risk of Ovarian Cancer, Nutrition and Cancer, 39:1,
19-28, DOI: 10.1207/S15327914nc391_3
To link to this article: http://dx.doi.org/10.1207/S15327914nc391_3
Published online: 18 Nov 2009.
Submit your article to this journal
Article views: 41
View related articles
Citing articles: 38 View citing articles
Full Terms & Conditions of access and use can be found at
http://www.tandfonline.com/action/journalInformation?journalCode=hnuc20
Download by: [University of Prince Edward Island]
Date: 05 November 2015, At: 13:59

NUTRITION AND CANCER, 39(1), 19–28
Copyright © 2001, Lawrence Erlbaum Associates, Inc.
Intakes of Selected Nutrients and Food
Groups and Risk of Ovarian Cancer
Susan E. McCann, Kirsten B. Moysich, and Curtis Mettlin
Abstract: In a hospital-based case-control study, we exam-
number of ovulatory cycles from higher parity, resulting in
ined dietary intakes of selected nutrients and food groups
and ovarian cancer risk among 496 women with primary,
histologically confirmed epithelial ovarian cancer and
1,425 women with nonneoplastic diagnoses, ages 20–87
years, admitted to Roswell Park Cancer Institute between
1982 and 1998. Data on diet and other relevant risk factors
in the few years before admission were collected with a self-
administered questionnaire. Odds ratios (OR) and 95%
confidence intervals (CI) were estimated by unconditional
logistic regression adjusting for age, education, region of
residence, regularity of menstruation, family history of
ovarian cancer, parity, age at menarche, oral contraceptive
use, and energy intake. Women in the highest vs. the lowest
quartile of total energy had a weak increase in risk (OR =
1.25, 95% CI = 0.90–1.73). Significantly reduced risks were
associated with higher intakes of dietary fiber (OR = 0.57,
95% CI = 0.38–0.87), vitamin A (OR = 0.66, 95% CI =
0.45–0.98), carotenoid (OR = 0.64, 95% CI = 0.43–0.93),
vitamin E (OR = 0.58, 95% CI = 0.38–0.88), b-carotene
(OR = 0.68, 95% CI = 0.46–0.98), and total fruit and vege-
table intake (OR = 0.62, 95% CI = 0.42–0.92). Our findings
suggest that a diet high in plant foods may be important in
reducing risk of ovarian cancer.
reduced mitotic events in the ovary (2).
Diet could contribute to the etiology of ovarian cancer
through modulation of the endogenous hormonal milieu
(3–13) or through antioxidant and anticarcinogenic mecha-
nisms (14). The relationship between nutrient and food in-
take and ovarian cancer has been addressed in case-control
and prospective designs, although the results have been in-
conclusive. In general, somewhat increased risks have been
reported for higher intakes of total fat and animal fat, choles-
terol and lactose intakes (15–17), and higher intakes of meat,
eggs, and whole milk (17–19). Conversely, inverse associa-
tions have been reported for fiber, vitamin A, b-carotene
(20–22), and fruit and vegetable intakes (15–17).
Beginning in 1982, all individuals admitted to Ros-
well Park Cancer Institute (RPCI), a comprehensive cancer
center in Western New York, were asked to complete a com-
prehensive epidemiological questionnaire that included re-
productive and medical histories, family history of cancer,
occupational and environmental exposures, tobacco and al-
cohol consumption, and diet. Since 1982, >40,000 patients
with cancer and noncancer diagnoses have completed the
questionnaire. Among these, >500 women were diagnosed
with ovarian cancer, providing a unique opportunity to in-
vestigate the impact of diet on risk of ovarian cancer.
Introduction
Methods
Subjects
Ovarian cancer is an important cause of morbidity and
mortality among women. Over 25,000 new cases were esti-
mated in 1999, with >14,000 estimated deaths (1). The dis-
ease is difficult to detect and often quite advanced by
diagnosis. The etiology of ovarian cancer has not been
clearly elucidated, but risk factors tend to be related to those
affecting hormonal and reproductive events. Higher parity
and oral contraceptive use have been associated with re-
duced risks (2). Putative mechanisms for ovarian cancer re-
lated to hormonal factors are excessive levels of circulating
gonadotropins associated with lower parity and decreased
The study population included women admitted to RPCI
between 1982 and 1998 who agreed to complete a compre-
hensive epidemiological questionnaire. Informed consent
was obtained from all participants. Only women with com-
plete data for the present analyses were included. Cases were
496 women with primary, incident epithelial ovarian cancer,
identified from the RPCI tumor registry and diagnostic in-
dex. Cases were predominantly Caucasian (98%) and ranged
in age between 20 and 87 years. Controls included 1,425
S. E. McCann is affiliated with the Department of Social and Preventive Medicine, University at Buffalo, Buffalo, NY 14214. K. B. Moysich and C. Mettlin
are affiliated with the Department of Cancer Prevention, Epidemiology and Biostatistics, Roswell Park Cancer Institute, Buffalo, NY 14263.

women who received medical services at RPCI for non-
neoplastic conditions, randomly selected from a pool of
5,700 eligible women and frequency matched to the cases by
five-year age intervals. The most frequently utilized services
for the control patients included the breast clinic (28%), gy-
necology clinic (17%), surgery (15%), and dermatology
clinic (11%). Among the control women, nonneoplastic di-
agnoses were distributed among the following disease cate-
gories: circulatory system (21%), respiratory system (18%),
digestive system (16%), musculoskeletal system and con-
nective tissue (12%), endocrine disorder (11%), central
nervous system (6%), and other (16%). Similar to cases,
women in the control group were predominantly Caucasian
and ranged in age between 25 and 90 years.
First, nutrients were computed for the WNY Diet Study us-
ing 140 foods, reported frequency of consumption, portion
size, cooking method, seasonality, and US Department of
Agriculture food composition values. With use of the foods
in the WNY Diet Study that represent the 44 foods included
on the shorter questionnaire, multiple linear regression mod-
els were calculated with each nutrient as the dependent vari-
able and the 44 foods as the independent variables. The
resulting regression coefficient for each food and nutrient
was then multiplied by the frequency of use reported on the
PEDS FFQ summed across foods to obtain an index of each
nutrient. Regression weights were available for calculation
of mean daily intakes of energy, protein, carbohydrates, total
fat, saturated fat, monounsaturated fat, polyunsaturated fat,
animal and vegetable fat, cholesterol, dietary fiber, vitamins
A, C, and E, retinol, total carotenoids, a-carotene, b-caro-
tene, cryptoxanthin, lutein + zeaxanthin, and lycopene.
In addition to investigating risk of ovarian cancer with
nutrient intake, risk associated with monthly frequency of
use of several food groups was of interest. Because the FFQ
was limited to 44 food items (predominantly fruits and vege-
tables), we examined the following food groups only: fruits
and fruit juices, vegetables, total fruits and vegetables, and
meats, categorized according to US Department of Agricul-
ture classification definitions. We were unable to examine
frequency of use of dairy products, breads and cereals, or
snacks and sweets, inasmuch as these items were not in-
cluded on the questionnaire.
Questionnaire
All participants agreed to complete the Patient Epidemi-
ology Data System (PEDS) questionnaire, which is offered
to all new patients as part of the admission process and is
completed by ~50% of new patients within three months of
diagnosis. The self-administered 16-page instrument covers
information on reproductive and medical histories, family
history of cancer, occupational and environmental expo-
sures, tobacco and alcohol consumption, and diet.
Diet in the few years before diagnosis (or admission for
controls) was queried using a 44-item food frequency ques-
tionnaire (FFQ) and did not include information regarding
portion size. The five response categories range from never
to five to seven times per week. The questionnaire was de-
signed to provide an assessment of intakes of fruits and veg-
etables, cruciferous vegetables, and foods providing good
sources of vitamins A, C, and E, fat, and fiber (23). Foods in-
cluded on the questionnaire were chosen to provide a profile
for each respondent for these nutrients. The foods were se-
lected from data from a detailed diet interview, lasting ~2.5
hours, with each of 282 healthy individuals, participants in
the Western New York (WNY) Diet Study I (1975–86). The
interview included detailed questions regarding frequency
of consumption, portion size, seasonality of intake, and
preparation methods for 140 foods. From the food frequency
data, an index of nutrient intake was calculated. Multiple lin-
ear regression models were then calculated, with nutrient in-
dex as the dependent variable and intake of individual foods
as the independent variables to obtain the list of foods that
explained the largest amount of variance for each nutrient.
Although brief, the resultant 44-item food list constituted
those foods most highly correlated with the nutrient profiles
in question and explained 90% of the total variability in in-
gestion of vitamins A, C, and E, fat, and dietary fiber (23).
Because of the brevity of the FFQ, nutrient intake was
calculated using regression weights derived from the
process used to obtain the food list included on the question-
naire. The gender-specific regression weights were com-
puted using dietary data from 1,475 male and 780 female
controls, participants of the WNY Diet Study I (1975–86).
Statistical Analyses
Descriptive analyses included Student’s t-tests of means
for cases and controls for continuous variables, and c2 tests
for categorical variables. For estimation of risk associated
with each nutrient, daily nutrient intake was divided into
quartiles on the basis of the intake distribution of the con-
trols. Risk of ovarian cancer in each quartile relative to the
lowest (referent) quartile of nutrient intake was estimated by
odds ratios (ORs) and 95% confidence intervals (CIs) calcu-
lated with unconditional logistic regression. ORs were ad-
justed for potential confounders, including age, education,
region of residence, age at menarche, parity, oral contracep-
tive use, presence of irregular menses, family history of
ovarian cancer, and total energy intake. Covariates were in-
cluded in the final regression model only if they were estab-
lished risk factors in these data or changed the observed risk
estimates by ³15%.
Analyses of risk associated with specific food groups
were conducted in a similar manner. Monthly frequency of
use of each group was divided into quartiles on the basis of
the intake distribution of the controls. ORs were adjusted for
the same covariates as for nutrient intake, with the exception
of total energy. To account for differences in diet composi-
tion, each food group was further adjusted for monthly fre-
quency of use of the remaining major food groups. Tests for
trend for each nutrient and food group were computed from
20
Nutrition and Cancer 2001

the P value (2-tailed) of the logistic regression of the contin-
uous nutrient or food group variable within the specified
models.
Risk of ovarian cancer associated with mean daily energy
and macronutrient intake is shown in Table 2. Compared
with women in the lowest quartile of intake, women in the
highest quartile of total energy had a weak, but statistically
nonsignificant, increase in risk (OR = 1.25, 95% CI =
0.90–1.73). Risk was not related to daily intakes of protein,
carbohydrates, total fat, or any fat component in these data,
although we observed a borderline statistically significant
trend in increasing risk associated with fat from animal
sources (p = 0.05).
Results
The descriptive characteristics of the ovarian cancer
cases and controls are shown in Table 1. Consistent with the
literature, women with ovarian cancer tended to have a later
age at first pregnancy, be nulliparous, be less likely to have
breast-fed, be less likely to have used oral contraceptives or
have a tubal ligation, be less likely to have regular menstrual
cycles, and be more likely to have a family history of ovar-
ian cancer than women without ovarian cancer. Cases were
also less likely than controls to reside in Western New York.
Age at menarche did not differ for women with and without
ovarian cancer.
Risk of ovarian cancer associated with mean daily micro-
nutrient intake is shown in Table 3. We observed reductions
in risk of ~30–40% for women in the highest vs. lowest
quartiles of dietary fiber (OR = 0.57, 95% CI = 0.38–0.87),
total vitamin A (OR = 0.66, 95% CI = 0.45–0.98), total ca-
rotenoids (OR = 0.64, 95% CI = 0.43–0.93), vitamin E (OR
= 0.58, 95% CI = 0.38–0.88), and b-carotene (OR = 0.68,
95% CI = 0.46–0.98). Somewhat weaker reductions in risk
were also observed for vitamin C (OR = 0.69, 95% CI =
0.47–1.03), retinol (OR = 0.71, 95% CI = 0.48–1.04), a-car-
otene (OR = 0.75, 95% CI = 0.53–1.06), lycopene (OR =
0.81, 95% CI = 0.57–1.15), and lutein + zeaxanthin (OR =
0.76, 95% CI = 0.52–1.10), with vitamin C and lutein +
zeaxanthin exhibiting statistically significant trends for re-
duced risks associated with increasing intake (p = 0.02).
Consistent with the reduced risks observed for the micro-
nutrients, as shown in Table 4, risk of ovarian cancer was re-
duced by ~40% for women in the highest vs. the lowest
quartile of total fruit and vegetable intake (OR = 0.62, 95%
CI = 0.42–0.92). Although risks were reduced for women in
the highest quartiles of fruits (OR = 0.85, 95% CI =
0.59–1.21) and vegetables (OR = 0.76, 95% CI = 0.52–1.10)
examined separately, there did not seem to be a significant
improvement in contribution to risk reduction by either indi-
vidual group. Risk was not related to monthly frequency of
meat consumption (OR = 1.17, 95% CI = 0.80–1.71).
Table 1. Descriptive Characteristics of Ovarian Cancer
Cases and Hospital Controls: Roswell Park Cancer
Institute, Buffalo, NY (1982–96)a
Cases
Controls
(n = 496)
(n = 1,425)
Mean SD
55.1 13.4
Age, yr
54.7 13.6
Age at menarche, yr
Age at 1st pregnancy, yr
13.1
23.2
5.7
4.8*
13.2
22.7
6.2
5.1
Number (%)
Education
Up to high school
College
258 (52.0)
238 (48.0)
781 (54.8)
644 (45.2)
Place of residence
Western New York
Outside Western New York
234 (47.2)†
262 (52.8)
1200 (84.2)
225 (15.8)
Tubal ligation
Yes
No
Discussion
67 (13.5)†
429 (86.5)
253 (17.8)
1172 (82.2)
Relatively little is known concerning the etiology of ovar-
ian cancer, although hormonal exposures appear to be im-
portant (2). There have been few investigations of the impact
of food and nutrient intake on ovarian cancer risk, but these,
in general, support a protective effect of diets higher in fruits
and vegetables, as well as nutrients commonly associated
with these foods, such as vitamin C, dietary fiber, and b-car-
otene (15–17,20–22), and a risk-elevating effect for foods
from animal sources, fat, and cholesterol (15–19). Our find-
ings confirm previously reported associations of reduced
ovarian cancer risks with vitamin C, dietary fiber, and b-car-
otene, although we observed no increases in risk associated
with any nutrient or food group. Several mechanisms could
account for our findings. Vitamins C and E and several of
the carotenoids have been shown to be potent antioxidants
and, therefore, could affect cancer risk. Diet composition
can affect steroid hormone levels, and, therefore, may affect
Oral contraceptive use
Ever
Never
Regularity of menses
Regular
Irregular
Parity
Nulliparous
1–2
3–4
³5
Breast feeding
Never
Ever
158 (31.9)†
338 (68.1)
494 (34.7)
931 (65.3)
432 (87.1)†
64 (12.9)
1154 (81.0)
271 (19.0)
110 (22.2)†
180 (36.3)
168 (33.9)
38 (7.7)
255 (17.9)
503 (53.3)
496 (34.8)
171 (12.0)
311 (64.0)†
175 (36.0)
789 (56.6)
605 (43.4)
Family history of ovarian cancer
No
Yes
460 (92.7)†
36 (7.3)
1377 (96.6)
48 (3.4)
a: Statistical significance is as follows: *, p < 0.05 (t-test); †, p < 0.05 (c2
test).
Vol. 39, No. 1
21

Table 2. ORs and 95% CIs for Risk of Ovarian Cancer With Daily Macronutrient Intake: Roswell Park Cancer Institute,
Buffalo, NY (1982–96)a
n
OR
P Value
Cases
Controls
(95% CI)
for Trend
Energy
Q1 (£1,346 kcal)
102
140
358
354
1.00
1.26
(0.92–1.73)
1.17
(0.85–1.62)
1.25
(0.90–1.73)
0.45
0.16
0.81
Q2 (1,347–1,754 kcal)
Q3 (1,755–2,256 kcal)
Q4 (>2,256 kcal)
128
126
361
352
Proteinb
Q1 (£56 g)
Q2 (57–76 g)
109
129
357
359
1.00
1.08
(0.77–1.51)
1.10
(0.74–1.61)
1.20
(0.69–2.08)
Q3 (77–102 g)
129
129
361
348
Q4 (>102 g)
Carbohydratesc
Q1 (£167 g)
Q2 (168–216 g)
108
142
354
354
1.00
1.10
(0.78–1.57)
0.95
(0.62–1.44)
0.86
Q3 (217–273 g)
124
122
366
351
Q4 (>273 g)
(0.47–1.59)
Total fatd
Q1 (£49 g)
Q2 (50–67 g)
109
131
355
362
1.00
1.17
(0.83–1.64)
1.22
(0.82–1.80)
1.48
(0.85–2.60)
1.00
0.11
0.05
Q3 (68–89 g)
127
129
360
348
Q4 (>90 g)
Fat from animal sourcese
Q1 (£34 g)
111
132
359
360
Q2 (35–49 g)
1.16
(0.83–1.62)
1.18
(0.80–1.73)
1.34
Q3 (50–68 g)
124
129
357
349
Q4 (>68 g)
(0.78–2.32)
Fat from vegetable sourcese
Q1 (£14 g)
Q2 (15–17 g)
117
131
356
352
1.00
1.10
0.68
(0.80–1.52)
1.01
(0.72–1.43)
0.90
Q3 (18–22 g)
123
125
360
357
Q4 (>22 g)
(0.60–1.34)
Saturated fate
Q1 (£16 g)
Q2 (17–24 g)
112
122
357
361
1.00
1.06
0.70
(0.76–1.49)
1.22
(0.8–1.79)
1.36
Q3 (25–34 g)
129
133
359
348
Q4 (>34 g)
(0.79–2.35)
(Continued)
22
Nutrition and Cancer 2001

Table 2. (Continued)
n
OR
P Value
Cases
Controls
(95% CI)
for Trend
Monounsaturated fate
Q1 (£15 g)
Q2 (16–21 g)
109
128
358
361
1.00
1.18
0.30
(0.84–1.65)
1.22
(0.82–1.80)
1.50
Q3 (22–28 g)
123
136
354
352
Q4 (>28 g)
(0.87–2.61)
Polyunsaturated fate
Q1 (£8 g)
Q2 (9–10 g)
115
131
362
351
1.00
1.17
0.40
(0.84–1.61)
1.05
(0.74–1.49)
0.92
Q3 (11–13 g)
127
123
360
352
Q4 (<13 g)
(0.60–1.42)
a: Q = quartile, adjusted for age, education, region of residence, regularity of menstruation, family history of ovarian cancer, parity, age at menarche, and oral
contraceptive use. OR, odds ratio; CI, confidence interval.
b: Further adjusted for carbohydrates and fat.
c: Further adjusted for protein and fat.
d: Further adjusted for carbohydrates and protein.
e: Further adjusted for total fat.
Table 3. ORs and 95% CIs for Risk of Ovarian Cancer With Daily Micronutrient Intake: Roswell Park Cancer Institute,
Buffalo, NY (1982–96)
n
ORa
P Value
Cases
Controls
(95% CI)
for Trend
Cholesterol
Q1 (£ 229 mg)
Q2 (230–323 mg)
120
129
355
364
1.00
0.88
(0.64–1.22)
0.86
(0.60–1.23)
0.99
(0.63–1.55)
0.51
0.03
0.02
Q3 (324–460 mg)
117
130
356
350
Q4 (>460 mg)
Dietary fiber
Q1 (£16 g)
Q2 (17–22 g)
117
141
350
359
1.00
1.06
(0.77–1.45)
0.85
(0.60–1.19)
0.57
(0.38–0.87)
Q3 (23–30 g)
137
101
362
354
Q4 (>30 g)
Vitamin C
Q1 (£112 mg)
Q2 (113–173 mg)
111
136
351
354
1.00
1.14
(0.83–1.58)
1.12
(0.80–1.56)
0.69
Q3 (174–250 mg)
153
96
360
360
Q4 (>250 mg)
(0.47–1.03)
(Continued)
Vol. 39, No. 1
23

Table 3. (Continued)
n
ORa
P Value
Cases
Controls
(95% CI)
for Trend
Vitamin E
Q1 (£4.9 mg)
Q2 (5.0–6.8 mg)
128
129
354
355
1.00
0.88
(0.64–1.22)
0.82
(0.58–1.16)
0.58
(0.38–0.88)
0.002
0.81
0.04
0.01
0.10
0.02
0.97
Q3 (6.9–9.4 mg)
135
104
362
354
Q4 (>9.4 mg)
Retinol
Q1 (£1,299 IU)
Q2 (1,300–2,117 IU)
139
130
362
362
1.00
0.92
(0.67–1.25)
0.84
(0.60–1.17)
0.71
(0.48–1.04)
Q3 (2,117–3,909 IU)
120
107
354
347
Q4 (>3,909 IU)
Vitamin A
Q1 (£5,917 IU)
Q2 (5,918–9,225 IU)
135
116
353
363
1.00
0.73
(0.53–1.01)
0.82
(0.58–1.14)
0.66
(0.45–0.98)
Q3 (9,226–14,204 IU)
131
114
361
348
Q4 (>14,204 IU)
Total carotenoids
Q1 (£3,808 IU)
Q2 (3,809–6,378 IU)
123
125
356
356
1.00
0.86
(0.62–1.19)
0.98
(0.71–1.36)
0.64
(0.43–0.93)
Q3 (6,379–10,311 IU)
145
103
357
356
Q4 (>10,311 IU)
a-Carotene
Q1 (£445 mg)
Q2 (446–801 mg)
134
130
356
357
1.00
0.95
(0.69–1.30)
0.76
(0.55–1.06)
0.75
(0.53–1.06)
Q3 (802–1,576 mg)
Q4 (>1,576 mg)
118
114
356
356
b-Carotene
Q1 (£3,043 mg)
Q2 (3,044–5,195 mg)
124
129
356
357
1.00
0.90
(0.65–1.23)
1.00
(0.72–1.39)
0.68
(0.46–0.98)
Q3 (5,196–8,349 mg)
Q4 (>8,349 mg)
141
102
356
356
Cryptoxanthin
Q1 (£20 mg)
Q2 (21–81 mg)
126
123
356
357
1.00
1.06
(0.77–1.45)
0.95
(0.69–1.30)
1.11
Q3 (82–262 mg)
Q4 (>262 mg)
116
131
356
356
(0.81–1.53)
(Continued)
24
Nutrition and Cancer 2001

Table 3. (Continued)
n
ORa
P Value
Cases
Controls
(95% CI)
for Trend
Lutein + zeaxanthin
Q1 (£2,748 mg)
Q2 (2,749–4,535 mg)
123
104
356
357
1.00
0.87
0.02
(0.62–1.20)
1.10
(0.80–1.51)
0.76
Q3 (4,536–7,101 mg)
Q4 (>7,101 mg)
162
107
356
356
(0.52–1.10)
Lycopene
Q1 (£2,362 mg)
Q2 (2,363–4,319 mg)
119
117
356
357
1.00
0.88
0.16
(0.64–1.21)
1.02
(0.74–1.41)
0.81
Q3 (4,320–6,684 mg)
Q4 (>6,684 mg)
145
115
356
356
(0.57–1.15)
a: Adjusted for age, education, region of residence, regularity of menstruation, family history of ovarian cancer, parity, age at menarche, oral contraceptive
use, and total energy intake.
Table 4. ORs and 95% CIs for Risk of Ovarian Cancer With Monthly Frequency of Use of Selected Food Groups:
Roswell Park Cancer Institute, Buffalo, NY (1982–96)
n
ORa
P Value
Cases
Controls
(95% CI)
for Trend
Fruits and fruit juices
Q1 (£48)
Q2 (49–72)
109
130
353
357
1.00
1.13
0.40
(0.81–1.56)
1.25
(0.90–1.74)
0.85
Q3 (73–101)
147
110
354
361
Q4 (>101)
(0.59–1.21)
Vegetables
Q1 (£24)
Q2 (25–42)
106
134
351
364
1.00
1.06
0.07
(0.77–1.47)
1.21
(0.88–1.68)
0.76
Q3 (43–66)
155
101
355
355
Q4 (>66)
(0.52–1.10)
Total fruits and vege-
tables
Q1 (£80)
Q2 (81–117)
107
137
347
362
1.00
1.11
0.09
(0.81–1.53)
1.17
(0.84–1.63)
0.62
Q3 (118–164)
162
90
364
352
Q4 (>164)
(0.42–0.92)
Meats
Q1 (£9)
Q2 (10–16)
111
128
365
359
1.00
1.04
0.12
(0.76–1.44)
1.11
(0.80–1.54)
1.17
Q3 (17–25)
132
125
357
344
Q4 (>25)
(0.80–1.71)
a: Each food group adjusted for age, education, region of residence, regularity of menstruation, family history of ovarian cancer, parity, age at menarche, oral
contraceptive use, and remaining food groups.
Vol. 39, No. 1
25

ovarian cancer risk through effects on the endogenous hor-
monal milieu. Research has shown that levels of estriol, total
estrogens, prolactin, and sex hormone-binding globulins can
be affected through changes in fat, protein sources, and di-
etary fiber intakes (3–13).
many epidemiological studies; our study obtained response
rates similar to those commonly reported in the literature.
Nevertheless, previous studies that utilized the PEDS data-
base and faced the same methodological issue consistently
replicated established epidemiological associations for a
variety of cancer sites, including ovary (24–26), colon
(27–30), breast (31,32), prostate (33,34), and lung (35,36).
There were marked differences between cases and con-
trols with respect to place of residence, with ovarian cancer
patients being more likely to live a greater distance from
RPCI than controls. This is likely due to the fact that RPCI
houses the Gilda Radner Familial Ovarian Cancer Registry,
which acts as a referral center for patients with a suspected
genetic predisposition for the development of ovarian cancer
(37). However, other than residential history, observed epi-
demiological differences between cases and controls were
consistent with those reported in the ovarian cancer litera-
ture (38–40).
Recall bias is always a problem in case-control studies of
cancer. However, in this investigation, it may have been less
of an issue because of our use of hospital controls. The ques-
tionnaire used in this investigation places no particular em-
phasis on any specific item. Thus there is little reason to
believe that cases were more motivated than controls to re-
call diet. Exposure misclassification may also have affected
our results, inasmuch as we based our analyses on self-
reported diet and were not able to independently verify this
information. Yet it is unlikely that this potential misclassi-
fication was differential in nature. In fact, our results con-
firm previous literature suggesting reduced risks of ovarian
cancer associated with diets high in fruits and vegetables and
associated nutrients (15–17,20–22).
Finally, the FFQ used to assess diet was limited, querying
frequency of use of only 44 foods and beverages. Although
these foods were selected to provide an adequate assessment
of intakes of fruits and vegetables, dietary fiber, vitamins A
and C, and the carotenoids, several important food groups
and nutrients may have been neglected. The use of regres-
sion weights for nutrient calculation allows for estimation of
nutrients not directly obtainable from the brief food list, but
the relative validity of these estimates varies by nutrient. In a
comparison of the 44-food questionnaire with two longer
FFQs [the National Cancer Institute Health Habits and His-
tory (“Block”) FFQ and the Harvard Semiquantitative FFQ],
the brief FFQ provided estimates of dietary fiber, vitamins A
and C, and carotenoids comparable to those obtained with
the other two instruments (41). Estimates of total energy, fat,
and fat components were less comparable (41). Notwith-
standing these limitations, we nevertheless obtained esti-
mates of risks for several nutrients that are consistent with
previously reported studies (15–17,20–22).
There is a great deal of interest in the role of phyto-
estrogens in the etiology of hormone-sensitive cancers. It has
been hypothesized that these compounds, available through
dietary intake, could affect endogenous hormone metabolism
through competitive inhibition of estrogen receptors or
through substitution of less active estrogen metabolites (14).
Although it was not possible to directly estimate because of
the limited number of foods included in the FFQ, plant foods
are good sources of phytoestrogens such as lignans. We ob-
served significant reductions in risk associated with the
highest intakes of dietary fiber. In a study of dietary
phytochemical intake and breast cancer in Western New
York, intakes of dietary fiber were highly correlated with in-
takes of lignans (r = 0.84; unpublished data). It is likely that
the women in the present study also consume significant
amounts of lignans, an exposure of potential importance in a
hormone-sensitive disease such as ovarian cancer.
Several methodological issues should be considered in
interpreting these results. As in all case-control studies, bias
could have affected the validity of the findings. Selection
bias may have occurred in this investigation. The ovarian
cancer patient group was restricted to women who were
treated at RPCI, a large regional cancer treatment center, and
may not represent the general population of ovarian cancer
patients in the Western New York region. However, it is un-
likely that self-reported diet would be different for RPCI pa-
tients or women treated in different facilities. Furthermore,
in this study, the control group consisted of female patients
who received medical services at RPCI for a large variety of
nonneoplastic conditions. The use of hospital controls might
introduce bias because of the possibility that some controls
were suffering from conditions that could affect dietary in-
take. However, it is unlikely that the hospital controls would
be any more likely than the cases to alter their diet because
of prevalent conditions. In addition, hospital controls were
selected from a large pool of eligible participants with a
wide variety of diagnostic groups, minimizing bias arising
from potential overrepresentation of patients with character-
istics that may be associated with exposure. Finally, as seen
in Table 1, many previously identified risk factors for ovar-
ian cancer were evident in our sample as well, suggesting
that this source of bias was not substantial.
Selection bias may have also been introduced because of
the low participation rate in this study. Only ~50% of eligi-
ble cases and controls agreed to complete the PEDS ques-
tionnaire. It is possible that women with more serious illness
did not complete the questionnaire; these women may differ
importantly from those who participated. We have no way to
ascertain whether those women who refused to complete the
instrument differed from participants with respect to the ex-
posures of interest. Nonresponse is a serious problem in
It is unlikely that the lack of portion size information for
the FFQ affected our results. In fact, the regression weights
used to calculate nutrient intake are obtained from food fre-
quency data that include portion size and frequency of use.
This method is analogous to the use of a standard portion
26
Nutrition and Cancer 2001

11. Hill, P, Garbaczewski, L, Helman, P, Huskisson, J, Sporangisa, E, et
al.: Diet, lifestyle, and menstrual activity. Am J Clin Nutr 33,
1192–1198, 1980.
12. Bennett, FC, and Ingram, DM: Diet and female sex hormone concen-
trations: an intervention study for the type of fat consumed. Am J Clin
Nutr 52, 808–812, 1990.
13. Rose, DP, Boyar, AP, Cohen, C, and Strong, LE: Effect of a low-fat
diet on hormone levels in women with cystic breast disease. I. Serum
steroids and gonadotropins. JNCI 78, 623–626, 1987.
14. Lampe, JW: Health effects of vegetables and fruit: assessing mecha-
nisms of action in human experimental studies. Am J Clin Nutr 70
Suppl, 475S–490S, 1999.
size in nutrient calculation. Furthermore, the availability of
portion size information has not been shown to substantially
improve nutrient intake estimates, inasmuch as subjects do
not estimate portion size accurately and tend to ignore ques-
tions regarding portion size, and portion sizes vary less
among individuals than do frequencies of use of specific
foods (42–46).
A plant-based diet continues to be important in lowering
risks of many cancers (1). Plant foods contain many sub-
stances, nutrient and nonnutrient, with potential to impact
cancer development through effects on hormone metabo-
lism, antioxidant activities, and other pathways (14). In con-
clusion, our results suggest that a diet high in fruits and
vegetables and, consequently, dietary fiber, vitamin C, and
carotenoids may play an important role in reducing the risk
of ovarian cancer, independent of nondietary risk factors.
15. LaVecchia, C, Decarli, A, and Negri, E: Dietary factors and the risk of
epithelial ovarian cancer. JNCI 79, 663–669, 1987.
16. Shu, XO, Gao, YT, Yuan, JM, Ziegler, RG, and Brinton, LA: Dietary
factors and epithelial ovarian cancer. Br J Cancer 59, 92–96, 1989.
17. Kushi, LH, Mink, PJ, Folsom, AR, Anderson, KE, Zhang, W, et al.:
Prospective study of diet and ovarian cancer. Am J Epidemiol 149,
21–31, 1999.
18. Mori, M, and Miyake, H: Dietary and other risk factors of ovarian can-
cer among elderly women. Jpn J Cancer Res 79, 997–1004, 1988.
19. Mettlin, CJ, and Piver, MS: A case-control study of milk-drinking and
ovarian cancer risk. Am J Epidemiol 132, 871–876, 1990.
20. Byers, T, Marshall, J, Graham, S, Mettlin, C, and Swanson, M: A case-
control study of dietary and nondietary factors in ovarian cancer. JNCI
71, 681–686, 1983.
21. Tzonou, A, Hsieh, CC, Polychronopoulou, A, Kaprinis, G, Toupadaki,
N, et al.: Diet and ovarian cancer: a case-control study in Greece. Int J
Cancer 55, 411–414, 1993.
Acknowledgments and Notes
Address correspondence to Susan E. McCann, Dept. of Social and Pre-
ventive Medicine, University at Buffalo, 270 Farber Hall, Buffalo, NY
14214. Phone: (716) 829-2975, ext. 605. FAX: (716) 829-2979. E-mail:
mccann@acsu.buffalo.edu.
Submitted 24 August 2000; accepted in final form 4 January 2001.
22. Slattery, ML, Schuman, KL, West, DW, French, TK, and Robison,
LM: Nutrient intake and ovarian cancer. Am J Epidemiol 130,
497–502, 1989.
References
23. Byers, T, Marshall, J, Fiedler, R, Zielezny, M, and Graham, S: As-
sessing nutrient intake with an abbreviated dietary interview. Am J
Epidemiol 122, 41–50, 1985.
1. American Cancer Society: Cancer Facts and Figures. Atlanta, GA:
Am Cancer Soc, 1999.
2. American Institute for Cancer Research: Food, Nutrition and the Pre-
vention of Cancer: A Global Perspective. II. Cancers, Nutrition and
Food. Ovary. Washington, DC: American Institute for Cancer Re-
search, 1997, chapt 4.12.
3. Thomas, HV, Davey, GK, and Key, TJ: Oestradiol and sex hormone-
binding globulin in premenopausal and postmenopausal meat-eaters,
vegetarians, and vegans. Br J Cancer 80, 1470–1475, 1999.
4. Boyd, NF, Lockwood, GA, Greenberg, CV, Martin, LJ, and Tritchler,
DL: Effects of a low-fat high-carbohydrate diet on plasma sex hor-
mones in premenopausal women: results from a randomized controlled
trial. Canadian Diet and Breast Cancer Prevention Study Group. Br J
Cancer 76, 127–135, 1997.
24. Mettlin, CJ, and Piver, MS: A case-control study of milk-drinking and
ovarian cancer risk. Am J Epidemiol 132, 871–876, 1990.
25. Hempling, RE, Wong, C, Piver, MS, Natarajan, N, and Mettlin, CJ:
Hormone replacement therapy as a risk factor for epithelial ovarian
cancer: results of a case-control study. Obstet Gyncecol 89, 1012–
1016, 1997.
26. Cornelison, TL, Natarajan, N, Piver, MS, and Mettlin, CJ: Tubal liga-
tion and the risk of ovarian carcinoma. Cancer Detect Prev 21, 1–6,
1997.
27. Freedman, AN, Michalek, AM, Weiss, HA, Zhang, ZF, Marshall, JR,
et al.: Aspirin use and p53 expression in colorectal cancer. Cancer De-
tect Prev 22, 213–218, 1998.
5. Woods, MN, Barnett, JB, Spiegelman, D, Trail, N, Hertzmark, E, et al.:
Hormone levels during dietary changes in premenopausal African-
American women. JNCI 88, 1369–1374, 1996.
6. Gates, JR, Parpia, B, Campbell, TC, and Junshi, C: Association of di-
etary factors and selected plasma variables with sex hormone-binding
globulin in rural Chinese women. Am J Clin Nutr 63, 22–31, 1996.
7. Bagga, D, Ashley, JM, Geffrey, SP, Wang, HJ, Barnard, RJ, et al.: Ef-
fects of a very low fat, high fiber diet on serum hormones and men-
strual function. Implications for breast cancer prevention. Cancer 76,
2491–2496, 1996.
8. Jacques, H, Noreau, L, and Moorjani, S: Effects on plasma lipoproteins
and endogenous sex hormones of substituting lean white fish for other
animal-protein sources in diets of postmenopausal women. Am J Clin
Nutr 55, 896–901, 1992.
9. Bennett, FC, and Ingram, DM: Diet and female sex hormone concen-
trations: an intervention study for the type of fat consumed. Am J Clin
Nutr 52, 808–812, 1990.
28. Freedman, AN, Michalek, AM, Marshall, JR, Mettlin, CJ, Petrelli, NJ,
et al.: Familial and nutritional risk factors for p53 overexpression in
colorectal cancer. Cancer Epidemiol Biomarkers Prev 5, 285–291,
1996.
29. Freedman, AN, Michalek, AM, Marshall, JR, Mettlin, CJ, Petrelli, NJ,
et al.: The relationship between smoking exposure and p53
overexpression in colorectal cancer. Br J Cancer 73, 902–908, 1996.
30. Suh, O, Mettlin, C, and Petrelli, NJ: Aspirin use, cancer, and polyps of
the large bowel. Cancer 72, 1171–1177, 1993.
31. Scanlon, EF, Suh, O, Murthy, SM, Mettlin, C, Reid, SE, et al.: Influ-
ence of smoking on the development of lung metastases from breast
cancer. Cancer 75, 2693–2699, 1995.
32. Mettlin, C, Croghan, I, Natarajan, N, and Lane, W: The association of
age and familial risk in a case-control study of breast cancer. Am J
Epidemiol 131, 973–983, 1990.
33. Mettlin, C, Selenskas, S, Natarajan, N, and Huben, R: b-Carotene and
animal fats and their relationship to prostate cancer risk. A case-control
study. Cancer 64, 605–612, 1989.
34. Mettlin, C, Natarajan, N, and Huben, R: Vasectomy and prostate can-
cer risk. Am J Epidemiol 132, 1056–1061, 1990.
10. London, S, Willett, W, Longcope, C, and McKinlay, S: Alcohol and
other dietary factors in relation to serum hormone concentrations in
women at climacteric. Am J Clin Nutr 53, 166–171, 1991.
Vol. 39, No. 1
27

35. Mettlin, C: Milk drinking, other beverage habits, and lung cancer risk.
Int J Cancer 43, 608–612, 1989.
36. Ambrosone, CB, Rao, U, Michalek, AM, Cummings, KM, and
Mettlin, CJ: Lung cancer histologic types and family history of cancer.
Analysis of histologic subtypes of 872 patients with primary lung can-
cer. Cancer 72, 1192–1198, 1993.
37. Piver, MS, Goldberg, JM, Tsukada, Y, Mettlin, CJ, Jishi, MF, et al.:
Characteristics of familial ovarian cancer: a report of the first 1,000
families in the Gilda Radner Familial Ovarian Cancer Registry. Eur J
Gynaecol Oncol 17, 169–176, 1996.
38. Riman, T, Persson, I, and Nilsson, S: Hormonal aspects of epithelial
ovarian cancer: review of epidemiological evidence. Clin Endocrinol
49, 695–707, 1998.
41. McCann, SE, Trevisan, M, Priore, R, Muti, P, Markovic, N, et al.:
Comparability of nutrient estimation by three food frequency question-
naires for use in epidemiologic studies. Nutr Cancer 35, 3–8, 1999.
42. Guthrie, HA: Selection and quantification of typical food portions by
young adults. J Am Diet Assoc 84, 1440–1444, 1984.
43. Subar, AF, Thompson, FE, Smith, AF, Jobe, JB, Ziegler, RG, et al.:
Improving food frequency questionnaires: a qualitative approach using
cognitive interviewing. J Am Diet Assoc 95, 781–788, 1995.
44. Hernandez-Avila, M, Master, C, Hunter, DJ, Buring, J, Philips, J, et al.:
Influence of additional portion size data on the validity of a semi-
quantitative food frequency questionnaire (abstr). Am J Epidemiol 128,
891, 1988.
45. Samet, JM, Humble, CG, and Skipper, BE: Alternatives in the collec-
tion and analysis of food frequency interview data. Am J Epidemiol
120, 572–581, 1984.
39. Daly, M, and Obrams, GI: Epidemiology and risk assessment for ovar-
ian cancer. Semin Oncol 25, 255–264, 1998.
40. Gajewski, W, and Legare, RD: Ovarian cancer. Surg Oncol Clin North
Am 7, 317–333, 1998.
46. Pickle, LW, and Hartman, AM: Indicator foods for vitamin A assess-
ment. Nutr Cancer 7, 3–23, 1985.
28
Nutrition and Cancer 2001