AJCN. First published ahead of print November 7, 2012 as doi: 10.3945/ajcn.112.039438.

See corresponding editorial on page 1249.

Dietary intakes of carbohydrates in relation to prostate cancer risk:

a prospective study in the Malmo Diet and Cancer cohort13
Isabel Drake, Emily Sonestedt, Bo Gullberg, Goran Ahlgren, Anders Bjartell, Peter Wallstrom, and Elisabet Wirfalt
ABSTRACT
Background: Dietary carbohydrates have been implicated in relation to prostate cancer.
Objective: Our objective was to examine the associations between
dietary intakes of carbohydrates, fiber, and their food sources and
risk of prostate cancer, overall and by case severity, in the Malmo
Diet and Cancer cohort.
Design: The analysis included 8128 men aged 4573 y without
a history of cancer, cardiovascular disease, or diabetes and who were
classified as adequate energy reporters. After a median follow-up time
of 15 y, prostate cancer was diagnosed in 817 men. We used Cox
proportional hazards regression to model associations between
energy-adjusted nutrient and food intakes with risk of incident prostate cancer, with competing risk of death from nonprostate cancer
causes taken into account.
Results: After adjustment for age and other known or potential risk
factors, we observed no associations between total carbohydrates or
dietary fiber and prostate cancer. We observed positive associations
between the intake of low-fiber cereals with overall and low-risk
prostate cancer and between intakes of cake and biscuits and rice
and pasta with low-risk prostate cancer (all P-trend , 0.05). A high
intake compared with zero consumption of sugar-sweetened beverages was associated with increased risk of symptomatic prostate
cancer (HR: 1.38; 95% CI: 1.04, 1.84).
Conclusions: Results from this large study with high-validity dietary data suggest that a high intake of refined carbohydrates may
be associated with increased risk of prostate cancer. However we
observed no significant associations with high-risk prostate cancer,
and not all foods that are typically high in refined carbohydrates were
associated with prostate cancer.
Am J Clin Nutr doi: 10.
3945/ajcn.112.039438.
INTRODUCTION

Prostate cancer has been shown to be affected by genetic

factors to a greater extent than many other tumors (1). However,
migration studies and between-country variation in incidence and
mortality rates have suggested that lifestyle and environmental
factors play an important role (2, 3). Several dietary factors have
been suggested to influence prostate cancer risk, but results are
inconclusive because of inherent difficulties in nutrition epidemiology and challenges in understanding prostate cancer cause
(4). Thus, established risk factors for prostate cancer include age,
ethnicity, and family history of prostate cancer (2, 5). Alterations
in the insulin and insulin-like growth factor endocrine axis are
one mechanism through which diet may influence prostate cancer

risk and progression (6, 7). Both insulin (8) and insulin-like
growth factor I (9) stimulate prostate cancer growth in vitro and
have been associated with prostate cancer risk in epidemiologic
studies (1013). Because one of the most potent stimulants for
insulin production is carbohydrate consumption, it has been
proposed that diets high in carbohydrates may affect prostate
cancer risk. The effect of carbohydrates on blood glucose and
insulin concentrations depend on the quantity and the quality of
carbohydrates, which can be characterized in part by the content
of refined or complex carbohydrates (eg, dietary fiber) (14).
Whole grains are an important source of complex carbohydrates,
and a high intake of whole grains is thought to be protective
against several cancers (15); however, recent studies have suggested that there is no relation between whole grains or dietary
fiber and prostate cancer risk (1618). Several reviews have been
published on the association between various dietary factors and
prostate cancer (19, 20), but the evidence regarding the role of
carbohydrates and their food sources appears inconclusive (4).
Therefore, the objective of this study was to investigate the association between dietary carbohydrates, fiber, and their food
sources and risk of prostate cancer in Swedish men participating
in the Malmo Diet and Cancer (MDC)4 cohort.
SUBJECTS AND METHODS

MDC cohort
The MDC cohort is a population-based prospective cohort set
in the South of Sweden. In 1991, the source population was
1
From the Research Group in Nutritional Epidemiology (ID, BG, PW, and EW),
Diabetes and Cardiovascular Disease, Genetic Epidemiology (ES), Department of
Clinical Sciences in Malmo, Lund University, Malmo, Sweden, and the Department of Urology, Skane University Hospital, Malmo, Sweden (GA and AB).
2
Supported by the Swedish Council for Working Life and Social Research, the Swedish Cancer Society, the Albert Pahlsson Foundation for
Scientific Research, the Gunnar Nilsson Cancer Foundation, Skane University Hospital Foundations and Donations, the Malmo General Hospital
Foundation for the Combating of Cancer, and the Ernhold Lundstrom Foundation for Scientific Research.
3
Address correspondence to I Drake, Research Group in Nutrition Epidemiology, Department of Clinical Sciences in Malmo, Lund University, Clinical Research Center Entrance 72, 60:13, SE-205 02 Malmo, Sweden.
E-mail: isabel.drake@med.lu.se.
4
Abbreviations used: MDC, Malmo Diet and Cancer; NPCR, National Prostate
Cancer Register; PSA, prostate-specific antigen; SCR, Swedish Cancer Register.
Received March 19, 2012. Accepted for publication September 7, 2012.
doi: 10.3945/ajcn.112.039438.

Am J Clin Nutr doi: 10.3945/ajcn.112.039438. Printed in USA. 2012 American Society for Nutrition

Copyright (C) 2012 by the American Society for Nutrition

1 of 10

2 of 10

DRAKE ET AL

defined as all persons living in the city of Malmo who were born
between 1926 and 1945 and had Swedish reading and writing
skills. In May 1995, the cohort was extended to include all
women born between 1923 and 1950 and all men born between
1923 and 1945. With this extension, 74,138 persons constituted
the source population. Details of recruitment procedures and the
cohort are described elsewhere (2124). With a participation
rate of w40%, the cohort consists of 28,098 participants, of
whom 11,063 are men.
Dietary assessment
Dietary information was collected by using a modified diethistory method, which combined a 168-item quantitative diet
history questionnaire (by using exact frequencies and a picture
booklet to assess portion sizes), a 7-d menu book (in which
descriptions of prepared meals, nutrient supplements, and cold
beverages were collected), and a 1-h dietary interview. Data on
the validity (25, 26) and reproducibility (27) of the method have
been published. Energy and nutrient intakes were computed from
the reported food intake by using the MDC Food and Nutrient
Database, which mainly originates from the PC Kost2-93 food
database of the Swedish National Food Administration. For men
in the MDC cohort, the relative validity coefficients (compared
with 18 d of weighed food records) were 0.66 for carbohydrates,
0.60 for sugar, 0.74 for dietary fiber, 0.65 for vegetables, 0.60 for
fruit, 0.69 for potatoes, 0.50 for bread, 0.74 for cereals, and 0.35 for
rice and pasta (25). The nutrient and food variables investigated in
this study were total carbohydrates, monosaccharides, sucrose, dietary fiber, whole grains, vegetables, fruit and berries, fruit juices,
potatoes, low-fiber cereals, low-fiber bread, high-fiber bread, cakes
and biscuits, rice and pasta, sweets and sugar, and sugar-sweetened
beverages. Potential dietary confounders were identified from the
literature and included energy, alcohol, calcium, selenium, vitamin
E, total protein, red meat, processed meat, total fat, SFAs, PUFAs,
fish and shellfish, and dairy products.
Anthropometric and other variables
Anthropometric measures were taken by trained project staff
and obtained from subjects who wore light indoor clothing
without shoes. BMI (in kg/m2) was defined as weight divided
by height squared. Waist circumference (in cm) was measured
midway between the lowest rib margin and iliac crest. Information
on age was obtained through the Swedish personal identification
number, which includes the date of birth. A structured questionnaire that covered socioeconomics, lifestyle, and disease history
was completed by the participants. The agreement between the
baseline questionnaire and the same questionnaire when repeated after 3 wk was high for most variables (k . 0.75) (23).
Smoking was defined as nonsmokers, exsmokers, or smokers
(which included irregular smokers). Educational level was defined according to the number of years of education completed
or degree of educational level attained as elementary, primary
and secondary, upper secondary, additional education without
a degree, or university degree. Total physical activity was obtained by combining work activity (sedentary, moderate, heavy,
or very heavy), domestic activity (quartiles of hours of household work), and leisure-time physical activity (quartiles of
score). The leisure-time physical activity score was obtained by

having participants estimate the number of minutes per week

spent on different physical activities for each of the 4 seasons.
The leisure-time physical activity score was calculated by
multiplying an intensity factor with the duration of each activity.
A diabetes diagnosis (yes or no) was based on a self-reported
diagnosis or the use of antidiabetic drugs. A history of a cardiovascular event (yes or no) was identified from national and
local registries and defined as the occurrence of a myocardial
infarction or stroke before study entry. Self-reported information
on the country of birth was used to create a variable that indicated birth in Sweden (yes or no). In September 1994, coding
routines for the dietary assessment were slightly changed to reduce the interview time (24). To reduce the potential bias caused
by this change and to account for the recruitment of slightly older
individuals during the last 2 y of study recruitment, the calendar
year of study entry was included as a categorical variable. A
categorical variable that indicated the season of data collection
(winter, spring, summer, or autumn) was also created. Participants
have previously been classified as potential underreporters, adequate reporters, or overreporters of energy intake (28) by using the
approach described by Goldberg et al (29) and later refined by
Black et al (30). Individuals with potentially unstable dietary
habits were identified according to their response to the questionnaire item Have you substantially changed your food habits
in the past due to illness or other reason? (31).
Clinical data and case ascertainment
In this study, we excluded men with cancer (except for nonmelanoma skin cancers) at baseline (n = 485), which resulted in
a study population of 10,578 men. The follow-up time was accrued from entry into the study until the date of prostate cancer
diagnosis, date of death from any cause, emigration, or end of
follow-up (31 December 2009), whichever came first. Incident
prostate cancer cases were identified by linkage of personal
identification numbers with the Swedish Cancer Register (SCR).
Information on vital status of the study population was obtained
from the Swedish Cause of Death Registry and the Swedish Tax
Agency. In this population, sufficient statistical power (ie, 80%
and a = 0.05) to observe a risk gradient of 11.75 over quintiles
of nutrient intake (given a true risk gradient of 13 and a validity
coefficient of the dietary variable of 0.6) was reached when 283
cancer cases had accumulated. Because of the high relative
validity of dietary data (r . 0.6 for many dietary variables), the
power to examine dietary hypotheses was satisfactory. By 31
December 2009, a total of 1016 prostate cancer cases were
identified by linkage of the MDC cohort to the SCR. Data on the
clinical and histologic characteristics of the tumors were collected from the National Prostate Cancer Register (NPCR). Data
on cases (n = 54) that occurred between 1991 and 1995 were
manually extracted from medical records by using standard
routines. The SCR is known to be $98% complete. A validation
of NPCR data from another region showed high validity for all
variables, including the variables used in the classification of
tumors by case severity in this study (32). High-risk prostate
cancer cases (n = 356) were defined as follows: local clinical
tumor stage T3 or T4, the presence of lymph node metastasis
(N1) or bone metastasis (M1), a Gleason score $8, or serum
prostate-specific antigen (PSA) concentrations .50 ng/mL (33).
Tumors were also classified as high-risk cases if the WHO grade

CARBOHYDRATES AND PROSTATE CANCER

was 3 and the Gleason score was unavailable (n = 6). All other
tumors were classified as low-risk cases (n = 651). Clinical data
were insufficient for the classification of 9 cases as low- or highrisk cases. In addition, health-conscious men may be prone to
attend regular health checks including PSA testing and, thus
more likely to be diagnosed with prostate cancer at an early
stage. This health-conscious behavior may cause a potential
bias when associations between diet and prostate cancer are
investigated. Therefore, we also investigated symptomatic cases (defined as men who presented with lower urinary tract
symptoms or other malignancy-related symptoms) separately
(n = 469).
Statistical analysis
Differences in characteristics of the study population across
quintiles of energy-adjusted carbohydrate intakes were tested by
using ANOVA and the chi-square test. Food and nutrient variables were energy adjusted by regressing the intake (logetransformed, dependent variable) on the total energy intake
(loge-transformed, independent variable), and the individuals
were divided into quintiles depending on their residual rankings
(34). A small number (0.01) was added to handle zero intakes
before log transformation when the proportion of participants
with a zero intake was less than 20%. Individuals with the
lowest intake (ie, first quintile) were used as the reference category. When .20% of participants reported a zero intake of
a food or food group, these individuals were categorized as zero
consumers ,and the remaining individuals were divided into
tertiles according to their energy-adjusted intakes. Individuals
with zero intakes were used as the reference category. For the
main analyses, we excluded men previously classified as potential energy misreporters and men with previous history of
cardiovascular disease or diabetes at baseline (n = 2450), which
resulted in a study population of 8128 men. Analyses of the
association between dietary variables and prostate cancer were
based on the Cox proportional hazards model. To ensure that the
estimation procedure was based on comparison of individuals at
the same age, we used age as the time scale (35). Study entry
was defined as the age at baseline, and study exit was defined as
the age at the end of follow-up. The proportional hazards assumption was evaluated graphically and tested by using scaled
Schoenfelds residuals. HRs and 95% CIs were estimated for
total, low-risk, high-risk, and symptomatic prostate cancers. We
selected a priori potential confounders, including known or
suspected risk factors for prostate cancer diagnosis. Multivariate
models were adjusted for the calendar year of study entry
(categorical), season of data collection (categorical), and the
following potential prostate cancer risk factors: total energy
intake (loge-transformed kcal; continuous), height (cm; continuous), waist (cm; continuous), physical activity (categorical),
smoking (categorical), educational level (categorical), and birth
in Sweden (categorical). Because of limited evidence on the role
of most dietary factors in prostate cancer development (4) and
the high colinearity between dietary factors, the multivariate
analysis included adjustment for calcium (quintiles of energyadjusted intake; categorical), selenium (quintiles of energyadjusted intake; categorical), and alcohol (quintiles of energy
adjusted intake; categorical). Because the inclusion of dietary
confounders did not greatly affect risk estimates, only the full

3 of 10

multivariate model that included these dietary covariates is

presented. To test for linear trends in risk of prostate cancer,
quintile ranks of explanatory variables were included as continuous predictors. We used Walds statistic (36) to evaluate
whether linear associations differed by case severity (low
compared with high risk). To evaluate the potential effect of
competing risks on the observed results, the multivariate model
was also used for competing risk regression, whereby all deaths
from causes other than prostate cancer (n = 1347) were considered competing events. An additional sensitivity analysis
excluded men who reported a substantial past food-habit change.
Stata/SE 12.0 software for Windows (StataCorp LP) was used
for all statistical analyses. All tests were 2-sided, and P , 0.05
was considered statistically significant.

RESULTS

Study population
The study included 1016 cases of prostate cancer that was
diagnosed between April 1992 and December 2009. The median
follow-up time (10th90th percentile) was 14.9 (range: 5.717.6) y.
Baseline characteristics of the study population are shown by
quintiles of energy-adjusted carbohydrate intake in Table 1. The
median daily intake of total carbohydrates ranged from 214 g in
the lowest quintile to 325 g in the highest quintile. Men with
a higher total carbohydrate intake were slightly older, were
shorter, had lower BMI and waist circumference, and were more
likely to be nonsmokers and physically active than were men
with a lower intake. Diabetes was less common in men with
a higher intake of carbohydrates, whereas having a history of
cardiovascular disease was more common, than in men with
a lower intake. Total carbohydrate intake correlated positively
with intakes of total selenium (r = 0.33) and calcium (r = 0.53)
and inversely with alcohol (r = 20.10).
Dietary intakes and risk of prostate cancer
For the main analysis, men classified as potential energy
misreporters and men with diabetes or previous cardiovascular
illness at baseline were excluded (Tables 2 and 3). No significant
associations were seen for total carbohydrates, dietary fiber,
whole grains, vegetables, fruit and berries, high-fiber bread, and
fruit juices. In the multivariate model, a high intake of cake and
biscuits was associated with increased risk of low-risk prostate
cancer (P-trend = 0.044), and risk estimates and the trend across
quintiles was strengthened in the competing risk model (P-trend =
0.034). The positive association was not seen for high-risk
prostate cancer, but there was no significant heterogeneity by
case severity (low compared with high risk; P-heterogeneity =
0.828). A high intake of low-fiber cereals was associated with
increased risk of total (P-trend = 0.034) and low-risk (P-trend =
0.013) prostate cancer, and the results were further strengthened in
the competing risk model. No association was seen with high-risk
prostate cancer risk (P-heterogeneity = 0.885). In addition, a high
intake of rice and pasta was associated with increased risk of lowrisk prostate cancer (HR: 1.33; 95% CI: 1.04, 1.70; P-trend = 0.044).
Increased risk remained in the competing risk model; however, the
trend was only borderline significant (P = 0.051). No association
was seen with high-risk prostate cancer (P-heterogeneity = 0.844).

4 of 10

DRAKE ET AL

TABLE 1
Characteristics of men (n = 10,578) in the MDC cohort (19911996) by quintiles of carbohydrate intake1
Quintiles of total carbohydrate intake (median intake)

Cases/participants (n)
Age (y)
Height (cm)
BMI (kg/m2)
Waist circumference (cm)
Alcohol (g/d)
Selenium (mg/d)
Calcium (mg/d)
Smoking status [n (%)]
Smokers
Exsmokers
Nonsmokers
Educational level [n (%)]
Elementary
Primary and secondary
Upper secondary
Further education without degree
University degree
Physical activity [n (%)]
Quartile 1
Quartile 2
Quartile 3
Quartile 4
Diabetes diagnosis [n (%)]
No
Yes
History of cardiovascular event [n (%)]
No
Yes
Born in Sweden [n (%)]
Yes
No
Past food habit change [n (%)]
Yes
No
Energy reporting [n (%)]
Under
Adequate
Over

1 (214 g/d)

2 (249 g/d)

3 (269 g/d)

4 (290 g/d)

5 (325 g/d)

200/2115
58.2 6 6.72
177 6 7
26.6 6 3.7
95.2 6 10.6
27.1 6 22.2
45.8 6 17.2
1124 6 478

200/2116
58.8 6 7.0
177 6 6
26.4 6 3.4
94.4 6 9.9
18.8 6 15.1
43.2 6 15.3
1181 6 479

213/2116
59.0 6 7.1
176 6 7
26.2 6 3.6
93.4 6 10.2
13.9 6 12.5
41.9 6 14.5
1194 6 449

195/2116
59.6 6 7.0
176 6 7
26.2 6 3.3
93.2 6 9.7
11.0 6 11.0
40.2 6 13.3
1217 6 452

208/2115
59.4 6 7.1
176 6 7
26.0 6 3.3
92.4 6 9.9
7.2 6 8.4
38.1 6 13.2
1212 6 454

,0.0001
,0.0001
,0.0001
,0.0001
,0.0001
,0.0001
,0.0001
,0.0001

846 (40.0)
844 (39.9)
424 (20.1)

682 (32.2)
912 (43.1)
522 (24.7)

600 (28.4)
922 (43.6)
592 (28.0)

472 (22.3)
958 (45.3)
685 (32.4)

453 (21.4)
901 (42.6)
759 (35.9)

888
406
280
216
323

(42.0)
(19.2)
(13.3)
(10.2)
(15.3)

922
451
241
202
296

(43.7)
(21.4)
(11.4)
(9.6)
(14.0)

995
438
242
176
263

(47.1)
(20.7)
(11.5)
(8.3)
(12.4)

1029
397
228
181
274

(48.8)
(18.8)
(10.8)
(8.6)
(13.0)

1015
377
262
199
253

(48.2)
(17.9)
(12.4)
(9.5)
(12.0)

622
425
609
312

(31.6)
(21.6)
(31.0)
(15.9)

578
398
672
358

(28.8)
(19.8)
(33.5)
(17.9)

551
390
682
380

(27.5)
(19.5)
(34.1)
(19.0)

520
346
706
430

(26.0)
(17.3)
(35.3)
(21.5)

407
349
754
481

(20.4)
(17.5)
(37.9)
(24.2)

,0.0001

,0.0001

,0.0001
1995 (94.3)
120 (5.7)

2017 (95.3)
99 (4.7)

2049 (96.8)
67 (3.2)

2038 (96.3)
78 (3.7)

2052 (97.0)
63 (3.0)

2030 (96.0)
85 (4.0)

2020 (95.5)
96 (4.5)

2007 (94.9)
109 (5.1)

1988 (94.0)
128 (6.0)

1959 (92.6)
156 (7.4)

1903 (90.0)
212 (10.0)

1889 (89.3)
226 (10.7)

1890 (89.4)
225 (10.6)

1848 (87.3)
268 (12.7)

1757 (83.1)
358 (16.9)

376 (17.8)
1735 (82.2)

431 (20.4)
1682 (79.6)

443 (21.0)
1670 (79.0)

502 (23.7)
1613 (76.3)

671 (31.8)
1441 (68.2)

281 (13.3)
1760 (83.2)
74 (3.5)

255 (12.1)
1792 (84.7)
60 (3.3)

243 (11.5)
1800 (85.1)
73 (3.5)

272 (12.9)
1768 (83.5)
76 (3.6)

249 (11.8)
1790 (84.6)
76 (3.6)

,0.0001
,0.0001
,0.0001

0.754

Quintile ranking was based on energy-adjusted carbohydrate intakes by using the residual method. ANOVA was used to calculate P values for age,
height, BMI, waist circumference, alcohol, selenium, and calcium, and chi-square tests of association were used for smoking, educational level, physical
activity, diabetes diagnosis, history of cardiovascular event, born in Sweden, past food habit change, and energy reporting. Individuals with missing values
were excluded from the analysis. Percentages may not add up to 100% because of rounding. MDC, Malmo Diet and Cancer.
2
Mean 6 SD (all such values).

Significantly increased risk of symptomatic prostate cancer was seen

for a high intake of sugar-sweetened beverages (HR: 1.41; 95%
CI: 1.06, 1.88; P-trend = 0.050). This association was slightly
attenuated in the competing risk model, and the trend was no
longer significant (P = 0.066). We also observed that a very low
intake of monosaccharides (quintile 1 compared with quintiles
25) was associated with decreased risk of symptomatic prostate
cancer (HR: 0.69; 95% CI: 0.48, 0.97). When men classified as
energy misreporters and with comorbidities at baseline (cardiovascular disease or diabetes) were included, associations seen
were similar but attenuated (data not shown). The exclusion of
men who reported a past food-habit change did not substantially
change the results (data not shown). In addition, the inclusion of

additional potential dietary confounders did not affect observed

associations.
DISCUSSION

Overall, results from this study suggest that relative intakes of total
carbohydrates and dietary fiber are not associated with prostate
cancer risk. However, the quality of carbohydrates, as indicated by
the increased risk seen with high intakes of certain foods that are high
in refined carbohydrates (ie, cakes and biscuits, low-fiber cereals, rice
and pasta, and sugar-sweetened beverages), may be associated with
incident prostate cancer. Although we observed no significant heterogeneity by case severity, most of the observed associations were

Total carbohydrates
Median intake (g/d)
Cases (n)
Total person-years
Model 1
Model 2
Model 3
Monosaccharides
Median intake (g/d)
Cases (n)
Total person-years
Model 1
Model 2
Model 3
Sucrose
Median intake (g/d)
Cases (n)
Total person-years
Model 1
Model 2
Model 3
Dietary fiber
Median intake (g/d)
Cases (n)
Total person-years
Model 1
Model 2
Model 3
Whole grains
Median intake (g/d)
Cases (n)
Total person-years
Model 1
Model 2
Model 3
Vegetables
Median intake (g/d)
Cases (n)
Total person-years
Model 1
Model 2
Model 3
Fruit and berries
Median intake (g/d)
Cases (n)
Total person-years
Model 1

333.7
178
21,719
1.06 (0.86, 1.32)
1.07 (0.83, 1.39)
1.10 (0.84, 1.42)

61.5
246
21,634
1.18 (0.93, 1.48)
1.15 (0.89, 1.48)
1.18 (0.92, 1.52)

81.3
158
21,439
0.90 (0.72, 1.12)
0.89 (0.70, 1.13)
0.90 (0.71, 1.15)

30.0
174
21,166
1.13 (0.91, 1.42)
1.11 (0.86, 1.42)
1.15 (0.89, 1.49)

51.5
170
21,228
1.04 (0.83, 1.30)
0.98 (0.77, 1.24)
1.00 (0.78, 1.28)

296.1
148
21,103
0.97 (0.78, 1.22)
1.03 (0.81, 1.31)
1.06 (0.83, 1.34)

335
181
21,214
1.17 (0.93, 1.45)

21.5
123
21,094
1.00 (reference)
1.00 (reference)
1.00 (reference)

23.3
151
20,369
1.00 (reference)
1.00 (reference)
1.00 (reference)

14.0
140
21,428
1.00 (reference)
1.00 (reference)
1.00 (reference)

0
146
20,842
1.00 (reference)
1.00 (reference)
1.00 (reference)

70.2
163
20,813
1.00 (reference)
1.00 (reference)
1.00 (reference)

44.9
139
22,168
1.00 (reference)

Quintile 5

219.4
153
21,466
1.00 (reference)2
1.00 (reference)
1.00 (reference)

Quintile 1

Total prostate cancer

(no. of cases = 817)

0.496

0.549
0.361
0.250

0.527
0.967
0.803

0.075
0.210
0.134

0.651
0.818
0.838

0.623
0.771
0.590

0.623
0.620
0.514

P-trend

89

1.00 (reference)

102

1.00 (reference)
1.00 (reference)
1.00 (reference)

91

1.00 (reference)
1.00 (reference)
1.00 (reference)

92

1.00 (reference)
1.00 (reference)
1.00 (reference)

91

1.00 (reference)
1.00 (reference)
1.00 (reference)

81

1.00 (reference)
1.00 (reference)
1.00 (reference)

98

1.00 (reference)
1.00 (reference)
1.00 (reference)

Quintile 1

119

1.24 (0.94, 1.64)

100

1.03 (0.78, 1.36)

1.13 (0.84, 1.53)
1.16 (0.86, 1.56)

113

1.14 (0.86, 1.50)

1.09 (0.80, 1.47)
1.11 (0.82, 1.50)

120

1.22 (0.93, 1.60)

1.21 (0.89, 1.64)
1.24 (0.90, 1.70)

93

0.91 (0.68, 1.22)

0.91 (0.67, 1.25)
0.92 (0.67, 1.25)

117

1.25 (0.94, 1.66)

1.20 (0.88, 1.64)
1.22 (0.89, 1.67)

127

1.13 (0.86, 1.48)

1.20 (0.88, 1.66)
1.21 (0.88, 1.68)

Quintile 5

Low-risk prostate cancer

(no. of cases = 531)

0.630

0.643
0.408
0.325

0.415
0.705
0.609

0.046
0.117
0.088

0.610
0.869
0.823

0.446
0.693
0.585

0.620
0.485
0.471

P-trend

48

1.00 (reference)

59

1.00 (reference)
1.00 (reference)
1.00 (reference)

54

1.00 (reference)
1.00 (reference)
1.00 (reference)

47

1.00 (reference)
1.00 (reference)
1.00 (reference)

57

1.00 (reference)
1.00 (reference)
1.00 (reference)

41

1.00 (reference)
1.00 (reference)
1.00 (reference)

53

1.00 (reference)
1.00 (reference)
1.00 (reference)

Quintile 1

62

1.08 (0.74, 1.57)

47

0.89 (0.61, 1.31)

0.87 (0.57, 1.31)
0.90 (0.60, 1.35)

56

0.88 (0.60, 1.28)

0.81 (0.54, 1.20)
0.84 (0.55, 1.28)

53

0.98 (0.66, 1.45)

0.91 (0.60, 1.40)
0.97 (0.61, 1.52)

65

0.91 (0.64, 1.30)

0.89 (0.61, 1.30)
0.91 (0.61, 1.34)

55

1.03 (0.69, 1.55)

1.05 (0.68, 1.62)
1.09 (0.70, 1.68)

59

0.96 (0.67, 1.40)

0.88 (0.57, 1.35)
0.89 (0.57, 1.40)

Quintile 5

High-risk prostate cancer

(no. of cases =278)

0.510

0.615
0.616
0.495

0.891
0.556
0.711

0.788
0.884
0.915

0.795
0.836
0.836

0.951
0.955
0.894

0.860
0.900
0.982

P-trend

TABLE 2
Association of extreme categories of nutrient and food intakes with prostate cancer risk in men (n = 8128) in the Malmo Diet and Cancer cohort (19912009)1

53

1.00 (reference)

75

1.00 (reference)
1.00 (reference)
1.00 (reference)

65

1.00 (reference)
1.00 (reference)
1.00 (reference)

55

1.00 (reference)
1.00 (reference)
1.00 (reference)

59

1.00 (reference)
1.00 (reference)
1.00 (reference)

46

1.00 (reference)
1.00 (reference)
1.00 (reference)

62

1.00 (reference)
1.00 (reference)
1.00 (reference)

Quintile 1

84

1.33 (0.94, 1.88)

61

0.91 (0.65, 1.27)

0.93 (0.64, 1.33)
0.96 (0.67, 1.38)

81

1.06 (0.77, 1.47)

0.94 (0.66, 1.35)
0.98 (0.68, 1.41)

78

1.24 (0.88, 1.75)

1.07 (0.73, 1.58)
1.13 (0.75, 1.70)

67

0.92 (0.64, 1.30)

0.87 (0.59, 1.26)
0.87 (0.60, 1.27)

83

1.40 (0.98, 2.01)

1.39 (0.93, 2.07)
1.43 (0.95, 2.15)

81

1.14 (0.82, 1.59)

1.00 (0.68, 1.47)
1.01 (0.68, 1.51)

Quintile 5

(Continued)

0.347

0.388
0.441
0.324

0.457
0.948
0.891

0.093
0.541
0.384

0.769
0.834
0.860

0.215
0.405
0.299

0.182
0.651
0.597

P-trend

Symptomatic prostate cancer

(no. of cases = 362)

CARBOHYDRATES AND PROSTATE CANCER

5 of 10

1.11 (0.88, 1.41)

1.15 (0.90, 1.46)

255.5
154
20,694
0.87 (0.70, 1.07)
0.86 (0.68, 1.08)
0.87 (0.69, 1.09)

172.5
135
20,672
0.92 (0.73, 1.15)
0.97 (0.76, 1.24)
0.95 (0.75, 1.21)

138.4
168
21,076
1.08 (0.86, 1.36)
1.04 (0.82, 1.32)
1.08 (0.85, 1.37)

77.5
191
20,917
1.15 (0.92, 1.45)
1.17 (0.91, 1.51)
1.21 (0.94, 1.56)

80.7
164
21,715
0.92 (0.73, 1.15)
0.91 (0.71, 1.16)
0.93 (0.73, 1.19)

64.7
175
22,099
1.00 (reference)
1.00 (reference)
1.00 (reference)

15.4
166
21,439
1.00 (reference)
1.00 (reference)
1.00 (reference)

3.1
138
21,405
1.00 (reference)
1.00 (reference)
1.00 (reference)

3.3
121
20,427
1.00 (reference)
1.00 (reference)
1.00 (reference)

10.7
140
19,596
1.00 (reference)
1.00 (reference)
1.00 (reference)

Quintile 5

1.00 (reference)
1.00 (reference)

Quintile 1

Total prostate cancer

(no. of cases = 817)

0.476
0.433
0.634

0.352
0.349
0.230

0.465
0.653
0.427

0.022
0.087
0.059

0.275
0.322
0.373

0.880
0.735

P-trend

86

1.00 (reference)
1.00 (reference)
1.00 (reference)

72

1.00 (reference)
1.00 (reference)
1.00 (reference)

85

1.00 (reference)
1.00 (reference)
1.00 (reference)

110

1.00 (reference)
1.00 (reference)
1.00 (reference)

121

1.00 (reference)
1.00 (reference)
1.00 (reference)

1.00 (reference)
1.00 (reference)

Quintile 1

92

0.87 (0.65, 1.17)

0.88 (0.64, 1.20)
0.89 (0.65, 1.22)

121

1.34 (1.00, 1.79)

1.42 (1.03, 1.97)
1.45 (1.03, 2.02)

110

1.19 (0.89, 1.58)

1.12 (0.83, 1.51)
1.15 (0.85, 1.55)

84

0.85 (0.64, 1.13)

0.92 (0.68, 1.24)
0.90 (0.66, 1.22)

99

0.81 (0.62, 1.06)

0.82 (0.61, 1.09)
0.82 (0.62, 1.10)

1.17 (0.87, 1.57)

1.19 (0.88, 1.61)

Quintile 5

Low-risk prostate cancer

(no. of cases = 531)

0.602
0.639
0.743

0.087
0.044
0.034

0.230
0.383
0.265

0.014
0.085
0.065

0.105
0.175
0.210

0.882
0.973

P-trend

52

1.00 (reference)
1.00 (reference)
1.00 (reference)

49

1.00 (reference)
1.00 (reference)
1.00 (reference)

52

1.00 (reference)
1.00 (reference)
1.00 (reference)

55

1.00 (reference)
1.00 (reference)
1.00 (reference)

54

1.00 (reference)
1.00 (reference)
1.00 (reference)

1.00 (reference)
1.00 (reference)

Quintile 1

69

0.96 (0.67, 1.38)

0.95 (0.65, 1.40)
0.99 (0.68, 1.45)

69

0.88 (0.61, 1.28)

0.86 (0.58, 1.29)
0.91 (0.61, 1.34)

54

0.87 (0.60, 1.28)

0.87 (0.59, 1.30)
0.92 (0.61, 1.38)

50

1.06 (0.72, 1.55)

1.07 (0.71, 1.59)
1.03 (0.69, 1.54)

54

0.97 (0.66, 1.41)

0.95 (0.64, 1.41)
0.95 (0.64, 1.41)

1.05 (0.70, 1.57)

1.09 (0.73, 1.64)

Quintile 5

High-risk prostate cancer

(no. of cases =278)

0.647
0.588
0.781

0.431
0.296
0.451

0.495
0.486
0.700

0.590
0.535
0.453

0.763
0.906
0.876

0.582
0.505

P-trend

56

1.00 (reference)
1.00 (reference)
1.00 (reference)

49

1.00 (reference)
1.00 (reference)
1.00 (reference)

59

1.00 (reference)
1.00 (reference)
1.00 (reference)

64

1.00 (reference)
1.00 (reference)
1.00 (reference)

70

1.00 (reference)
1.00 (reference)
1.00 (reference)

1.00 (reference)
1.00 (reference)

Quintile 1

71

0.93 (0.65, 1.32)

0.86 (0.59, 1.25)
0.89 (0.62, 1.28)

84

1.10 (0.77, 1.57)

1.07 (0.73, 1.59)
1.11 (0.75, 1.65)

81

1.16 (0.83, 1.63)

1.06 (0.74, 1.51)
1.11 (0.77, 1.59)

54

0.98 (0.68, 1.41)

1.04 (0.71, 1.52)
1.00 (0.68, 1.47)

72

0.99 (0.72, 1.38)

1.03 (0.73, 1.45)
1.04 (0.74, 1.46)

1.23 (0.84, 1.79)

1.28 (0.87, 1.88)

Quintile 5

0.532
0.357
0.476

0.712
0.813
0.634

0.844
0.681
0.915

0.083
0.206
0.136

0.915
0.618
0.570

0.757
0.648

P-trend

Symptomatic prostate cancer

(no. of cases = 362)

1
Quintiles of energy-adjusted intake are shown. Men classified as potential energy misreporters and men with a history of cardiovascular disease or diabetes diagnosis at baseline were excluded from the
analysis. All P-trend values were obtained by treating categories of intake as a continuous variable. Model 1 was estimated by using Cox proportional hazards regression with age as the time metric. Model 2 was
estimated as for model 1 and adjusted for year of study entry (categorical), season of data collection (summer, spring, autumn, or winter), energy intake (kcal/d), height (cm), waist (cm), physical activity
(quartiles of score), smoking (nonsmoker, former smoker, or current smoker), educational level (elementary, primary and secondary, upper secondary, additional education without a degree, or university
degree), birth in Sweden (yes or no), alcohol (quintiles of energy-adjusted intake), calcium (quintiles of energy-adjusted intake), and selenium (quintiles of energy-adjusted intake). Model 3 was estimated and
adjusted as for model 2 and with competing risk by death from all causes except prostate cancer taken into account.
2
HR; 95% CI in parentheses (all such values).

Model 2
Model 3
Potatoes
Median intake (g/d)
Cases (n)
Total person-years
Model 1
Model 2
Model 3
Low-fiber bread
Median intake (g/d)
Cases (n)
Total person-years
Model 1
Model 2
Model 3
High-fiber bread
Median intake (g/d)
Cases (n)
Total person-years
Model 1
Model 2
Model 3
Cakes and biscuits
Median intake (g/d)
Cases (n)
Total person-years
Model 1
Model 2
Model 3
Sweets and sugar
Median intake (g/d)
Cases (n)
Total person-years
Model 1
Model 2
Model 3

TABLE 2 (Continued )

6 of 10
DRAKE ET AL

P-trend

Zero intake

Tertile 3

P-trend

Zero intake

Tertile 3

High-risk prostate cancer

(no. of cases = 278)
P-trend

Zero intake

Tertile 3

P-trend

Symptomatic prostate cancer

(no. of cases = 362)

0
297.8

365
148

240
92

120
56

161
77

47,357
19,637

1.00 (reference) 1.05 (0.87, 1.28) 0.626 1.00 (reference) 0.98 (0.77, 1.25) 0.819 1.00 (reference) 1.24 (0.90, 1.70) 0.151 1.00 (reference) 1.27 (0.97, 1.66) 0.193
1.00 (reference) 1.14 (0.93, 1.39) 0.205 1.00 (reference) 1.08 (0.83, 1.39) 0.612 1.00 (reference) 1.30 (0.94, 1.81) 0.092 1.00 (reference) 1.41 (1.06, 1.88) 0.050
1.00 (reference) 1.13 (0.92, 1.38) 0.221 1.00 (reference) 1.06 (0.82, 1.37) 0.661 1.00 (reference) 1.28 (0.92, 1.78) 0.103 1.00 (reference) 1.38 (1.04, 1.84) 0.066

0
29.5

264
184

160
130

102
52

126
76

35,502
24,920

1.00 (reference) 1.17 (0.97, 1.41) 0.154 1.00 (reference) 1.31 (1.04, 1.65) 0.024 1.00 (reference) 0.92 (0.66, 1.29) 0.686 1.00 (reference) 1.08 (0.81, 1.44) 0.422
1.00 (reference) 1.16 (0.95, 1.41) 0.188 1.00 (reference) 1.33 (1.04, 1.70) 0.044 1.00 (reference) 0.87 (0.61, 1.23) 0.497 1.00 (reference) 1.00 (0.74, 1.35) 0.848
1.00 (reference) 1.15 (0.94, 1.40) 0.213 1.00 (reference) 1.31 (1.03, 1.67) 0.051 1.00 (reference) 0.86 (0.60, 1.24) 0.529 1.00 (reference) 0.99 (0.73, 1.34) 0.859

0
200.0

418
131

255
93

160
36

183
65

55,244
17,363

1.00 (reference) 1.06 (0.87, 1.29) 0.928 1.00 (reference) 1.21 (0.96, 1.54) 0.194 1.00 (reference) 0.78 (0.55, 1.13) 0.074 1.00 (reference) 1.23 (0.93, 1.63) 0.383
1.00 (reference) 0.99 (0.81, 1.22) 0.639 1.00 (reference) 1.12 (0.87, 1.43) 0.553 1.00 (reference) 0.76 (0.52, 1.11) 0.081 1.00 (reference) 1.14 (0.85, 1.54) 0.759
1.00 (reference) 0.99 (0.81, 1.22) 0.625 1.00 (reference) 1.12 (0.87, 1.44) 0.532 1.00 (reference) 0.75 (0.52, 1.09) 0.067 1.00 (reference) 1.14 (0.85, 1.52) 0.784

0
37.7

234
207

141
135

90
69

100
93

38,034
24,315

1.00 (reference)2 1.32 (1.09, 1.59) 0.002 1.00 (reference) 1.45 (1.15, 1.84) 0.001 1.00 (reference) 1.10 (0.80, 1.51) 0.621 1.00 (reference) 1.34 (1.01, 1.78) 0.026
1.00 (reference) 1.24 (1.01, 1.52) 0.034 1.00 (reference) 1.35 (1.04, 1.76) 0.013 1.00 (reference) 1.05 (0.75, 1.47) 0.902 1.00 (reference) 1.22 (0.89, 1.66) 0.191
1.00 (reference) 1.27 (1.04, 1.55) 0.015 1.00 (reference) 1.38 (1.07, 1.79) 0.006 1.00 (reference) 1.08 (0.77, 1.51) 0.770 1.00 (reference) 1.25 (0.92, 1.70) 0.124

Tertile 3

Low-risk prostate cancer

(no. of cases = 531)

Men with zero intake were compared with men in the highest tertile of energy-adjusted intake. Men classified as potential energy misreporters and men with a history of cardiovascular disease or diabetes
diagnosis at baseline were excluded from the analysis. All P-trend values were obtained by treating categories of intake as a continuous variable. Model 1 was estimated by using Cox proportional hazards
regression with age as the time metric. Model 2 was estimated as for model 1 and adjusted for year of study entry (categorical), season of data collection (summer, spring, autumn, or winter), energy intake
(kcal/d), height (cm), waist (cm), physical activity (quartiles of score), smoking (nonsmoker, former smoker, or current smoker), educational level (elementary, primary and secondary, upper secondary,
additional education without a degree, or university degree), birth in Sweden (yes or no), alcohol (quintiles of energy-adjusted intake), calcium (quintiles of energy-adjusted intake), and selenium (quintiles of
energy-adjusted intake). Model 3 was estimated and adjusted as for model 2 and with competing risk by death from all causes except prostate cancer taken into account.
2
HR; 95% CI in parentheses (all such values).

Low-fiber cereals
Median intake (g/d)
Cases (n)
Total person-years
Model 1
Model 2
Model 3
Fruit juices
Median intake (g/d)
Cases (n)
Total person-years
Model 1
Model 2
Model 3
Rice and pasta
Median intake (g/d)
Cases (n)
Total person-years
Model 1
Model 2
Model 3
Sugar-sweetened beverages
Median intake (g/d)
Cases (n)
Total person-years
Model 1
Model 2
Model 3

Zero intake

Total prostate cancer

(no. of cases = 817)

TABLE 3
Association of extreme categories of food intakes with prostate cancer risk in men (n = 8128) in the Malmo Diet and Cancer cohort (19912009)1

CARBOHYDRATES AND PROSTATE CANCER

7 of 10

8 of 10

DRAKE ET AL

seen for low-risk prostate cancer, and only sugar-sweetened beverages were associated with symptomatic prostate cancer.
The null associations seen in this study for dietary fiber and
whole grains were in line with recent reports (1618). However,
clinical and animal studies have suggested that dietary fiber or
whole grains may play a role in prostate cancer progression.
Landberg et al (37) showed that whole grain and bran from rye
resulted in significantly lower plasma PSA than did a cellulosesupplemented refined wheat diet in patients with prostate cancer.
In addition, it has been shown that rye bran delayed growth and
increase apoptosis in prostate tumors in animal studies (38).
Although several studies suggested possible adverse effects
(primarily increased tumor growth and reduced apoptosis of
prostate cancer cells) of an increased carbohydrate consumption
on prostate cancer, most evidence has stemmed from small
preclinical studies (3941). Epidemiologic studies that investigated the association between dietary carbohydrates and
prostate cancer risk have generally shown no significant associations (42, 43). Two case-control studies showed a positive
association between high glycemic foods and risk of prostate
cancer (44, 45), and one case-control study showed a positive
association with the intake of starch (46). However, several prospective cohort studies have shown no association between glycemic index or load and prostate cancer risk (17, 42). We showed
that some, but not all, foods high in refined carbohydrates were
associated with prostate cancer (primarily low-risk prostate cancer). Although no association was seen with total carbohydrates
or sucrose, high intake of sugar-sweetened beverages was shown
to be associated with w40% increased risk of symptomatic
prostate cancer in our study population. A tendency for increased
risk was also seen with a high intake of monosaccharides; however, this result was only significant when very low intakes
(quintile 1) were compared with higher intakes (quintiles 25),
which possibly indicated a potential threshold value.
There is a possibility that men with a more health-conscious
behavior, including higher intakes of dietary fiber, whole grains,
and fruit and vegetables, are exposed to a higher degree of diagnostic intensity, such as PSA testing (4, 47). Although the
actual prevalence has not been quantified, there is strong, indirect
evidence that the use of PSA testing is increasing in Sweden after
having been introduced during the early 1990s (48, 49). Some of
this bias may be reduced by the adjustment for age and the calendar
year of entry into the study. Because the use of PSA testing on
nonsymptomatic men may vary by region, it is a strength that our
study population was restricted to a specific region (ie, Malmo,
Sweden), and thus, the bias from varying local and regional policy
regarding prostate cancer screening was minimized (48). In addition, with data from the NPCR, we were able to investigate
symptomatic tumors separately. To reduce the potential effect of
a detection bias, we adjusted all analyses for several markers of
health-conscious behaviors (eg, educational level, physical activity, smoking, and alcohol consumption). Also, because prostate
cancer develops later in life, men with poor dietary habits are more
likely to die of other causes before being diagnosed with prostate
cancer (50). It is possible that the effect of dietary factors on
prostate cancer incidence seen in epidemiologic studies may be an
effect of competing risk if these dietary factors are related to comorbidities associated with early death. In this study, the taking
into account of competing risk of causes of death other than
prostate cancer had only a minor effect on observed associations.

Strengths of this study include the prospective design, large

number of prostate cancer cases, complete and valid identification of prostate cancer cases through the SCR, and virtually no
loss to follow-up. In addition, a major strength of the study was
the dietary data with high relative validity and the ability to
exclude individuals who were likely to have misreported energy
intakes. Although this exclusion could have caused a potential
reduction in power (especially for high-risk prostate cancer), the
strengthened results seen after this exclusion highlighted the
difficulties in dealing with self-reported dietary data. Few cohort
studies have the possibility of excluding individuals who are
likely to be misreporting energy intake. Because the measurement error for dietary data tends to attenuate associations toward
the null, the associations we saw may, in reality, have been stronger,
and the null results may have been due to random measurement
error and an insufficiently large range in intake between individuals. Because of multiple testing, there wa a possibility that the
findings of this study were due to chance; however, we did
observe a pattern of increased risk with high intakes of refined
carbohydrates as seen with several food groups. In addition,
because of the observational nature of this study, there is always
the possibility that the observed associations were due to some
unmeasured confounders, but residual confounding by the
covariates we adjusted for was unlikely because of the similar
results in age-adjusted and multivariate analyses. Because dietary
components are highly correlated, adjustment for other nutrients
or foods may bias (if prone to measurement error) or attenuate
results. Therefore, in our analyses, we limited the number of
dietary covariates, although the inclusion of other potential dietary confounders (eg, red and processed meat, fish, and dietary
fat) had no effect on the observed associations. There is increasing awareness that localized and low-grade prostate cancer
tumors may have different causes than those of advanced and
high-grade tumors (51). This study separated high-risk prostate
cancer cases (advanced and/or aggressive) from low-risk prostate
cancer. However, the power of this subgroup analysis was reduced by the smaller number of high-risk cases. In a larger
population, additional subclassification of tumors according to
stage and grade might help clarify some associations.
In conclusion, we observed no association between relative
intakes of total carbohydrates, dietary fiber, or fiber-rich foods such
as whole grains with risk of prostate cancer in this prospective
study of Swedish men. Although results were not significant for
high-risk prostate cancer, our findings suggest that high intake of
foods typically high in refined carbohydrates (including cakes and
biscuits, low-fiber cereals, rice and pasta, and sugar-sweetened
beverages) may increase risk of this disease. Because of the
heterogeneous nature of prostate cancer and limited power to
investigate tumors by stage and grade, results of this study should
be interpreted with caution.
The authors responsibilities were as followsID, EW, and PW: designed
the study; ID and BG: planned the statistical analysis; ID: performed all
statistical analyses and wrote the manuscript; and all authors: helped with
the interpretation of results and read and approved the final version of the
manuscript. None of the authors declared a conflict of interest.

REFERENCES
1. Lichtenstein P, Holm NV, Verkasalo PK, Iliadou A, Kaprio J, Koskenvuo M,
Pukkala E, Skytthe A, Hemminki K. Environmental and heritable

CARBOHYDRATES AND PROSTATE CANCER

2.
3.
4.

5.
6.

7.

8.

9.

10.

11.

12.

13.

14.

15.
16.

17.

18.

19.
20.
21.
22.

23.

24.

factors in the causation of canceranalyses of cohorts of twins from

Sweden, Denmark, and Finland. N Engl J Med 2000;343:7885.
Hsing AW, Chokkalingam AP. Prostate cancer epidemiology. Front
Biosci 2006;11:1388413.
Wolk A. Diet, lifestyle and risk of prostate cancer. Acta Oncol 2005;44:
27781.
World Cancer Research Fund, American Institute for Cancer Research.
Food, nutrition, physical activity and the prevention of cancer: a global
perspective. Washington, DC: American Institute for Cancer Research, 2007.
Leitzmann MF, Rohrmann S. Risk factors for the onset of prostatic cancer:
age, location, and behavioral correlates. Clin Epidemiol 2012;4:111.
Kaaks R. Nutrition, insulin, IGF-1 metabolism and cancer risk:
a summary of epidemiological evidence. Novartis Found Symp 2004;
262:24760, discussion 608.
Heald AH, Cade JE, Cruickshank JK, Anderson S, White A, Gibson
JM. The influence of dietary intake on the insulin-like growth factor
(IGF) system across three ethnic groups: a population-based study.
Public Health Nutr 2003;6:17580.
McKeehan WL, Adams PS, Rosser MP. Direct mitogenic effects of
insulin, epidermal growth factor, glucocorticoid, cholera toxin, unknown pituitary factors and possibly prolactin, but not androgen, on
normal rat prostate epithelial cells in serum-free, primary cell culture.
Cancer Res 1984;44:19982010.
Iwamura M, Sluss PM, Casamento JB, Cockett AT. Insulin-like growth
factor I: action and receptor characterization in human prostate cancer
cell lines. Prostate 1993;22:24352.
Chan JM, Stampfer MJ, Ma J, Gann P, Gaziano JM, Pollak M,
Giovannucci E. Insulin-like growth factor-I (IGF-I) and IGF binding
protein-3 as predictors of advanced-stage prostate cancer. J Natl Cancer
Inst 2002;94:1099106.
Hsing AW, Chua S Jr, Gao YT, Gentzschein E, Chang L, Deng J,
Stanczyk F. Prostate cancer risk and serum levels of insulin and leptin:
a population-based study. J Natl Cancer Inst 2001;93:7839.
Chan JM, Stampfer MJ, Giovannucci E, Gann PH, Ma J, Wilkinson P,
Hennekens CH, Pollak M. Plasma insulin-like growth factor-I and
prostate cancer risk: a prospective study. Science 1998;279:5636.
Stattin P, Rinaldi S, Biessy C, Stenman UH, Hallmans G, Kaaks R.
High levels of circulating insulin-like growth factor-I increase prostate
cancer risk: a prospective study in a population-based nonscreened
cohort. J Clin Oncol 2004;22:310412.
Riccardi G, Rivellese AA, Giacco R. Role of glycemic index and
glycemic load in the healthy state, in prediabetes, and in diabetes. Am
J Clin Nutr 2008;87:269S74S.
Jacobs DR Jr, Marquart L, Slavin J, Kushi LH. Whole-grain intake and
cancer: an expanded review and meta-analysis. Nutr Cancer 1998;30:8596.
Suzuki R, Allen NE, Key TJ, Appleby PN, Tjonneland A, Johnsen NF,
Jensen MK, Overvad K, Boeing H, Kaaks R, et al. A prospective
analysis of the association between dietary fiber intake and prostate
cancer risk in EPIC. Int J Cancer 2009;124:2459.
Nimptsch K, Kenfield S, Jensen MK, Stampfer MJ, Franz M, Sampson L,
Brand-Miller JC, Willett WC, Giovannucci E. Dietary glycemic index,
glycemic load, insulin index, fiber and whole-grain intake in relation to
risk of prostate cancer. Cancer Causes Control 2011;22:5161.
Egeberg R, Olsen A, Christensen J, Johnsen NF, Loft S, Overvad K,
Tjonneland A. Intake of whole-grain products and risk of prostate
cancer among men in the Danish Diet, Cancer and Health cohort study.
Cancer Causes Control 2011;22:11339.
Chan JM, Gann PH, Giovannucci EL. Role of diet in prostate cancer
development and progression. J Clin Oncol 2005;23:815260.
Hori S, Butler E, McLoughlin J. Prostate cancer and diet: food for
thought? BJU Int 2011;107:134859.
Berglund G, Elmstahl S, Janzon L, Larsson SA. The Malmo Diet and
Cancer Study. Design and feasibility. J Intern Med 1993;233:4551.
Manjer J, Carlsson S, Elmstahl S, Gullberg B, Janzon L, Lindstrom M,
Mattisson I, Berglund G. The Malmo Diet and Cancer Study: representativity, cancer incidence and mortality in participants and nonparticipants. Eur J Cancer Prev 2001;10:48999.
Manjer J, Elmstahl S, Janzon L, Berglund G. Invitation to a populationbased cohort study: differences between subjects recruited using various strategies. Scand J Public Health 2002;30:10312.
Wirfalt E, Mattisson I, Johansson U, Gullberg B, Wallstrom P,
Berglund G. A methodological report from the Malmo Diet and Cancer
study: development and evaluation of altered routines in dietary data
processing. Nutr J 2002;1:3.

9 of 10

25. Elmstahl S, Riboli E, Lindgarde F, Gullberg B, Saracci R. The Malmo

Food Study: the relative validity of a modified diet history method and
an extensive food frequency questionnaire for measuring food intake.
Eur J Clin Nutr 1996;50:14351.
26. Riboli E, Elmstahl S, Saracci R, Gullberg B, Lindgarde F. The Malmo
Food Study: validity of two dietary assessment methods for measuring
nutrient intake. Int J Epidemiol 1997;26(suppl 1):S16173.
27. Elmstahl S, Gullberg B, Riboli E, Saracci R, Lindgarde F. The Malmo
Food Study: the reproducibility of a novel diet history method and an
extensive food frequency questionnaire. Eur J Clin Nutr 1996;50:13442.
28. Mattisson I, Wirfalt E, Aronsson CA, Wallstrom P, Sonestedt E,
Gullberg B, Berglund G. Misreporting of energy: prevalence, characteristics of misreporters and influence on observed risk estimates in the
Malmo Diet and Cancer cohort. Br J Nutr 2005;94:83242.
29. Goldberg GR, Black AE, Jebb SA, Cole TJ, Murgatroyd PR, Coward
WA, Prentice AM. Critical evaluation of energy intake data using
fundamental principles of energy physiology: 1. Derivation of cut-off
limits to identify under-recording. Eur J Clin Nutr 1991;45:56981.
30. Black AE, Goldberg GR, Jebb SA, Livingstone MB, Cole TJ, Prentice
AM. Critical evaluation of energy intake data using fundamental
principles of energy physiology: 2. Evaluating the results of published
surveys. Eur J Clin Nutr 1991;45:58399.
31. Sonestedt E, Wirfalt E, Gullberg B, Berglund G. Past food habit change
is related to obesity, lifestyle and socio-economic factors in the Malmo
Diet and Cancer Cohort. Public Health Nutr 2005;8:87685.
32. Sandblom G, Dufmats M, Olsson M, Varenhorst E. Validity of a population-based cancer register in Swedenan assessment of data reproducibility in the South-East Region Prostate Cancer Register. Scand
J Urol Nephrol 2003;37:1129.
33. Stocks T, Lukanova A, Rinaldi S, Biessy C, Dossus L, Lindahl B,
Hallmans G, Kaaks R, Stattin P. Insulin resistance is inversely related
to prostate cancer: a prospective study in Northern Sweden. Int
J Cancer 2007;120:267886.
34. Willett WC, Howe GR, Kushi LH. Adjustment for total energy intake
in epidemiologic studies. Am J Clin Nutr 1997;65(4 suppl):1220S8S;
discussion 9S31S.
35. Thiebaut AC, Benichou J. Choice of time-scale in Coxs model analysis of
epidemiologic cohort data: a simulation study. Stat Med 2004;23:380320.
36. Greenland S, Rothman KJ. Introduction to stratified analysis. In:
Rothman KJ, Greenland S, Lash TL, eds. Modern epidemiology. 3rd
ed. Philadelphia, PA: Wolters Kluwer Health/Lippincott Williams &
Wilkins, 2008:25882.
37. Landberg R, Andersson SO, Zhang JX, Johansson JE, Stenman UH,
Adlercreutz H, Kamal-Eldin A, Aman P, Hallmans G. Rye whole grain
and bran intake compared with refined wheat decreases urinary
C-peptide, plasma insulin, and prostate specific antigen in men with
prostate cancer. J Nutr 2010;140:21806.
38. Bylund A, Zhang JX, Bergh A, Damber JE, Widmark A, Johansson A,
Adlercreutz H, Aman P, Shepherd MJ, Hallmans G. Rye bran and soy
protein delay growth and increase apoptosis of human LNCaP prostate
adenocarcinoma in nude mice. Prostate 2000;42:30414.
39. Venkateswaran V, Haddad AQ, Fleshner NE, Fan R, Sugar LM, Nam
R, Klotz LH, Pollak M. Association of diet-induced hyperinsulinemia
with accelerated growth of prostate cancer (LNCaP) xenografts. J Natl
Cancer Inst 2007;99:1793800.
40. Freedland SJ, Mavropoulos J, Wang A, Darshan M, Demark-Wahnefried
W, Aronson WJ, Cohen P, Hwang D, Peterson B, Fields T, et al. Carbohydrate restriction, prostate cancer growth, and the insulin-like growth
factor axis. Prostate 2008;68:119.
41. Mavropoulos JC, Buschemeyer WC 3rd, Tewari AK, Rokhfeld D,
Pollak M, Zhao Y, Febbo PG, Cohen P, Hwang D, Devi G, et al. The
effects of varying dietary carbohydrate and fat content on survival in
a murine LNCaP prostate cancer xenograft model. Cancer Prev Res
(Phila) 2009;2:55765.
42. Shikany JM, Flood AP, Kitahara CM, Hsing AW, Meyer TE, Willcox
BJ, Redden DT, Ziegler RG. Dietary carbohydrate, glycemic index,
glycemic load, and risk of prostate cancer in the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial (PLCO) cohort. Cancer
Causes Control 2011;22:9951002.
43. Hu J, La Vecchia C, Gibbons L, Negri E, Mery L. Nutrients and risk of
prostate cancer. Nutr Cancer 2010;62:7108.
44. Hardin J, Cheng I, Witte JS. Impact of consumption of vegetable, fruit,
grain, and high glycemic index foods on aggressive prostate cancer
risk. Nutr Cancer 2011;63:86072.

10 of 10

DRAKE ET AL

45. Augustin LS, Galeone C, Dal Maso L, Pelucchi C, Ramazzotti V,

Jenkins DJ, Montella M, Talamini R, Negri E, Franceschi S, et al.
Glycemic index, glycemic load and risk of prostate cancer. Int J Cancer
2004;112:44650.
46. Bidoli E, Talamini R, Bosetti C, Negri E, Maruzzi D, Montella M,
Franceschi S, La Vecchia C. Macronutrients, fatty acids, cholesterol
and prostate cancer risk. Ann Oncol 2005;16:1527.
47. Baade PD, Youlden DR, Krnjacki LJ. International epidemiology of
prostate cancer: geographical distribution and secular trends. Mol Nutr
Food Res 2009;53:17184.

48. Stattin P, Johansson R, Lodnert R, Andren O, Bill-Axelsson A, Bratt O,

Damber JE, Hellstrom M, Hugosson J, Lundgren R, et al. Geographical
variation in incidence of prostate cancer in Sweden. Scand J Urol
Nephrol 2005;39:3729.
49. Hemminki K, Rawal R, Bermejo JL. Prostate cancer screening,
changing age-specific incidence trends and implications on familial
risk. Int J Cancer 2005;113:3125.
50. Gronberg H. Prostate cancer epidemiology. Lancet 2003;361:85964.
51. Freedland SJ, Giovannucci E, Platz EA. Are findings from studies of obesity
and prostate cancer really in conflict? Cancer Causes Control 2006;17:59.