Beruflich Dokumente
Kultur Dokumente
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
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
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
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).
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
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
100
113
120
93
117
127
Quintile 5
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
47
56
53
65
55
59
Quintile 5
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
61
81
78
67
83
81
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
5 of 10
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
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
121
110
84
99
Quintile 5
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
69
54
50
54
Quintile 5
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
84
81
54
72
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
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
Zero intake
Tertile 3
P-trend
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
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
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
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.
REFERENCES
1. Lichtenstein P, Holm NV, Verkasalo PK, Iliadou A, Kaprio J, Koskenvuo M,
Pukkala E, Skytthe A, Hemminki K. Environmental and heritable
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
9 of 10
10 of 10
DRAKE ET AL