Supplemental Material can be found at: http://jn.nutrition.org/content/suppl/2014/03/12/jn.113.18967 0.DCSupplemental.html
The Journal of Nutrition Nutritional Epidemiology
Dietary Total and Insoluble Fiber Intakes Are Inversely Associated with Prostate Cancer Risk1–3 M´elanie Deschasaux,4* Camille Pouchieu,4 Mathilde His,4 Serge Hercberg,4,5 Paule Latino-Martel,4 and Mathilde Touvier4
Abstract Although experimental data suggest a potentially protective involvement of dietary fiber in prostate carcinogenesis, very few prospective studies have investigated the relation between dietary fiber intake and prostate cancer risk, and those have had inconsistent results. Our objective was to study the association between dietary fiber intake (overall, insoluble, soluble, and from different sources, such as cereals, vegetables, fruits, and legumes) and prostate cancer risk. Stratifications by excess weight status, insulin-like growth factors, and amount of alcohol intake were also considered. ´ ´ This prospective analysis included 3313 men from the Supplementation en Vitamines et Mineraux Antioxydants (SU.VI. MAX) cohort who completed at least 3 24-h dietary records. One hundred thirty-nine incident prostate cancers were diagnosed between 1994 and 2007 (median follow-up of 12.6 y). Associations between quartiles of energy-adjusted dietary fiber intake and prostate cancer risk were characterized by multivariate Cox proportional hazards models. Prostate cancer risk was inversely associated with total dietary fiber intake (HR of quartile 4 vs. quartile 1 = 0.47; 95% CI: 0.27, 0.81; P = 0.001), insoluble (HR = 0.46; 95% CI: 0.27, 0.78; P = 0.001), and legume (HR = 0.55; 95% CI: 0.32, 0.95; P = 0.04) fiber intakes. In contrast, we found no association between prostate cancer risk and soluble (P = 0.1), cereal (P = 0.7), vegetable (P = 0.9), and fruit (P = 0.4) fiber intakes. In conclusion, dietary fiber intake (total, insoluble, and from legumes but not soluble or from cereals, vegetables, and fruits) was inversely associated with prostate cancer risk, consistent with mechanistic data. This trial was registered at clinicaltrials.gov as NCT00272428. J. Nutr. 144: 504–510, 2014.
Introduction Several mechanisms are involved in prostate cancer development. First, experimental (1–3) and epidemiologic (3,4) data suggest that chronic inflammation stimulates prostate carcinogenesis. Second, insulin resistance and hyperinsulinemia, which lead to several metabolic modifications, such as higher insulinlike growth factor (IGF)6-1 and IGF-2 bioactivity and lower sexhormone binding globulin (SHBG) concentrations (1,5,6), have 1
ˆ le Ile-de-France (public funding from the Paris ´ Supported by the Canceropo region; to M.D.). The funders had no role in the design, implementation, analysis, or interpretation of the data. 2 Author disclosures: M. Deschasaux, C. Pouchieu, M. His, S. Hercberg, P. Latino-Martel, and M. Touvier, no conflicts of interest. 3 Supplemental Tables 1–4 are available from the ‘‘Online Supporting Material’’ link in the online posting of the article and from the same link in the online table of contents at jn.nutrition.org. 6 Abbreviations used: EPIC, European Prospective Investigation into Cancer and Nutrition; IGF, insulin-like growth factor; IGFBP, insulin-like growth factor binding protein; PSA, prostate-specific antigen; SHBG, sex-hormone binding globulin; ´ ´ SU.VI.MAX, Supplementation en Vitamines et Mineraux Antioxydants; WCRF, World Cancer Research Fund. * To whom correspondence should be addressed. E-mail: m.deschasaux@uren. smbh.univ-paris13.fr.
504
been associated with increased risk of prostate cancer in experimental (5,7) and epidemiologic (7–11) studies. Finally, increased concentrations of circulating estrogens and androgens could be related to a greater risk of prostate cancer (12). Experimental data suggested that dietary fiber may do the following: 1) have anti-inflammatory properties (13–15); 2) improve insulin sensitivity (16,17) and reduce IGF bioactivity, notably by increasing insulin-like growth factor binding protein (IGFBP) 3 concentrations (18); 3) upregulate SHBG concentration (19); and 4) influence steroid hormone concentrations by decreasing circulating estrogens and androgens (20,21), thereby supporting the hypothesis that dietary fiber has a protective effect against prostate carcinogenesis. In 2007, the World Cancer Research Fund (WCRF)/American Institute for Cancer Research considered that the epidemiologic evidence of an association between dietary fiber intake and prostate cancer risk was too limited to allow a conclusion (22), because only 1 unadjusted null-result prospective cohort study was available (23). Since then, 3 prospective studies on dietary fiber and prostate cancer risk have been published (24–26). Overall, they did not detect any relation (24–26), except for an inverse association observed in the European Prospective
ã 2014 American Society for Nutrition. Manuscript received December 10, 2013. Initial review completed January 8, 2014. Revision accepted January 27, 2014. First published online February 19, 2014; doi:10.3945/jn.113.189670.
Downloaded from jn.nutrition.org at UNIVERSITY OF PITTSBURGH FALK LIBRARY on January 17, 2015
4 Sorbonne Paris Cit´e Research Center, Nutritional Epidemiology Research Team, U1153 National Institute of Health and Medical Research, U1125 National Institute for Agricultural Research, National Conservatory of Arts and Crafts, Paris 13 University, Bobigny, France; and 5Public Health Department, Avicenne Hospital, Bobigny, France
Materials and Methods Participants. The Suppl´ementation en Vitamines et Min´eraux Antioxydants (SU.VI.MAX) study was at first designed as a randomized, double-blind, placebo-controlled primary prevention trial (registered at clinicaltrials.gov as NCT00272428) aiming to assess the effect of daily supplementation with nutritional doses of antioxidants on the incidence of cardiovascular diseases and cancers (32). In 1994–1995, 13,017 individuals (7876 women and 5141 men) were enrolled for an 8-y intervention study. Follow-up of health events lasted until September 2007. The SU.VI.MAX study was conducted according to the Declaration of Helsinki guidelines and was approved by the Paris-Cochin Hospital Ethics Committee for Studies with Human Subjects (CCPPRB nos. 706 and 2364, respectively) and the French National Commission for Computed Data and Individual Freedom (CNIL nos. 334641 and 907094, respectively). Written informed consent was obtained from all participants. Baseline data collection. Information about sociodemographics, smoking status, physical activity, and family history of prostate cancer were collected by self-administered questionnaires for all participants at enrollment. Anthropometric measures were obtained during a medical examination by the study nurses and physicians. Every 2 mo during the trial phase (1994–2002), participants were invited to complete a 24-h dietary record via the Minitel Telematic Network, a French telephone-based terminal equivalent to an Internet prototype used widely at the beginning of the study. The records were randomly distributed between weeks and weekends and over seasons to take into account intra-individual variability. Participants assessed portion sizes with the aid of a validated picture booklet (33), and the amounts consumed from composite dishes were estimated using French recipes validated by food and nutrition professionals. The mean daily energy, alcohol, and nutrient intakes were estimated using a published
French food composition table (34). Total dietary fiber and soluble fiber contents were obtained using the Association of Official Analytical Chemists method (35). Dietary fiber intakes in the SU.VI.MAX study were described previously (36). During the trial phase, participants were advised against taking any self-prescribed supplementation. Fiber intake from supplements is therefore most likely negligible in our study. In the SU.VI.MAX cohort, a study was conducted to validate an FFQ that was completed by the participants in 2007. This FFQ was compared with 12 dietary records across 1 y (considered as the reference method). Energy-adjusted deattenuated correlation coefficients for total dietary fiber intake between this FFQ and the 12 dietary records were 0.56 for men and 0.63 for women (37). These coefficients were similar to those found in the EPIC cohort (0.54) (38). A 35 mL fasting venous blood sample was collected in vacutainer tubes. Blood samples were centrifuged immediately after blood draw, and plasma aliquots were then preserved in sodium heparin. Less than 1 h after blood draw, plasma aliquots were stored at 220°C in dry ice for shipment to the central biobank (maximum of 24 h), where they were stored frozen in liquid nitrogen. These plasma samples were used to determine the baseline concentrations of selenium, total prostate-specific antigen (PSA) measured by immunoassay (Roche Diagnostics), IGF1 measured by immunometry on a Immulite analyzer (Diagnostics Products Corporation), and IGF-2 measured by immunoradiometric assays (Diagnostic Systems Laboratories). Coefficients of variation for the laboratory assays were 5.5% for total PSA, 5.3% for IGF-1, and 6.8% for IGF-2. Total PSA, IGF-1, and IGF-2 baseline concentrations were measured in 2002–2003 as part of a post hoc research protocol (39,40). The SU.VI.MAX investigators contacted the participants and their treating physicians if their baseline PSA value was >4.0 mg/L. Additional medical investigations were left to the discretion of the treating physician (40). Case ascertainment. During the follow-up period, participants were invited to self-report health events (through a monthly questionnaire). Investigations were then conducted to obtain medical data from participants, physicians, and/or hospitals. All information was reviewed by an independent physician expert committee, and cases were validated with the pathologic reports and classified using the International Chronic Diseases Classification, 10th Revision, Clinical Modification (41). Cancers were validated by histologic reports, and all prostate cancer cases were documented by a pathology report (32,39). All first-incident primary prostate cancers were considered as cases in this study. Statistical analyses. From the 5141 male participants in the SU.VI. MAX study, we excluded 41 men who reported a cancer diagnosis before the start of the follow-up. Among the remaining participants, 3313 provided at least 3 valid 24-h dietary records within the first 2 years of follow-up and thus remained available for analysis. Participants contributed person-time until the date of diagnosis of prostate cancer, the date of last completed questionnaire, the date of death, or September 2007, whichever occurred first. Participants who reported a cancer other than prostate cancer during the study period (n = 163) were included and censored at the date of diagnosis (except those with basal cell skin carcinoma, which was not considered to be cancer). Nutrient intakes were estimated by the mean intake calculated from all dietary records for each man. Baseline characteristics of participants were compared between quartiles of total dietary fiber intake using x2 tests or Fisher tests (ANOVA models). HRs and 95% CIs obtained from Cox proportional hazards models, with age as the primary time variable, were used to characterize the association between quartiles of dietary fiber intake and incident of prostate cancer. We confirmed that the assumptions of proportionality were satisfied through examination of the log–log (survival) vs. log–time plots. Different categories of dietary fiber were tested according to their chemical properties (soluble and insoluble fibers) and according to their food sources (cereal, vegetable, fruit, and legume fibers). Tests for linear trend were performed using the ordinal score on quartiles of fiber intake. Correlations between the different types of dietary fiber intakes were analyzed using Spearman correlation coefficients. Multivariate models were adjusted for factors constitutive Dietary fiber intake and prostate cancer risk
505
Downloaded from jn.nutrition.org at UNIVERSITY OF PITTSBURGH FALK LIBRARY on January 17, 2015
Investigation into Cancer and Nutrition (EPIC) cohort study for fruit fiber among men aged $65 y (26). However, these studies (25,26) used FFQs to assess dietary fiber intakes, which may weaken associations between food intakes and health outcomes (27,28). Thus, new prospective studies are needed to further investigate the relation between dietary fiber intake and prostate cancer risk. In addition, ‘‘dietary fiber’’ refers to a large category of molecules with various biochemical properties (such as solubility) and fermentation products, probably resulting in differential physiologic effects on the management of inflammation, insulin resistance, or circulating steroid hormones and, thus, prostate carcinogenesis (29,30). Nonetheless, only 1 prospective study on prostate cancer risk has investigated separately the different sources of dietary fiber (26), and none has distinguished soluble and insoluble fibers. Furthermore, to our knowledge, previous prospective studies on dietary fiber and prostate cancer risk did not take into account IGF concentrations or stratify for alcohol intake, and only 1 considered stratification by BMI with null result (25). Because overweight and obesity are linked to a low-grade chronic inflammation status (2,6), IGF-1 and IGF-2 concentrations are related to prostate cancer risk (7–10), and alcohol intake is associated with higher concentrations of circulating estrogens (31), we hypothesized that excess weight, IGF-1 and IGF-2 concentrations, and alcohol intake may modulate the association between dietary fiber intake and prostate cancer risk. Hence, our objective was to prospectively investigate the association between different types of dietary fiber (overall, insoluble, soluble, and from different food sources, such as cereals, vegetables, fruits, and legumes) and prostate cancer risk using repeated dietary records, and explore its potential modulation by overweight/obesity status, IGF-1 and IGF-2 plasma concentrations, and alcohol intake.
type of dietary fiber were tested by introducing the product of the 2 variables in the models. Interaction with the SU.VI.MAX trial antioxidant supplementation group was also tested. All tests were 2 sided, and P < 0.05 was considered statistically significant. SAS version 9.3 (SAS Institute) was used for analyses.
Results During a median follow-up of 12.6 y (37,877 person-years), 139 men developed a first primary prostate cancer with a mean age at diagnosis of 63 y. A total of 5.2% of the participants were lost to follow-up. Table 1 presents the characteristics of the participants according to quartiles of total dietary fiber intake. Participants in the upper quartile tended to be older, taller, nonsmokers, more active physically, and had lower BMIs. The mean dietary fiber intake was 21 6 7.3 g/d. Only 25% of the study population reached the 25 g/d minimal French recommendation (43). Main contributors to dietary fiber intake were cereals
TABLE 1 Baseline characteristics of participants (n = 3313) according to quartiles of total dietary fiber intake, SU.VI.MAX Cohort, France, 1994–20071
Age, y BMI, kg/m2 $25 kg/m2, n (%) Height, cm Intervention group, n (%) yes Smoking status Never Former Current Physical activity, n (%) Irregular ,1 h/d walking or equivalent $1 h/d walking or equivalent Educational level, n (%) Primary Secondary University Family history of prostate cancer,3 n (%) yes PSA, mg/L $3 mg/L, n (%) Energy intake, kcal/d Alcohol intake, g/d Total dietary fiber intake, g/d Insoluble fiber, g/d Soluble fiber, g/d Cereal fiber, g/d Vegetable fiber, g/d Fruit fiber, g/d Legume fiber,4 g/d Calcium intake, mg/d Vitamin E intake, mg/d Processed meat intake, g/d Tomato products intake, g/d Legume intake, g/d Blood Selenium, mmol/L
Quartile 1 (n = 828)
Quartile 2 (n = 828)
Quartile 3 (n = 829)
Quartile 4 (n = 828)
P2
51.2 6 4.6 25.9 6 3.2 501 (60.5) 173 6 6.2 424 (51.2)
51.7 6 4.6 25.6 6 3.2 441 (53.3) 173 6 6.3 438 (52.9)
51.8 6 4.8 25.2 6 2.9 407 (49.1) 173 6 6.3 415 (50.1)
52.5 6 4.9 24.8 6 3.0 360 (43.5) 174 6 6.4 413 (49.9)
,0.0001 ,0.0001 ,0.0001 ,0.0001 0.6 ,0.0001
226 (27.3) 423 (51.1) 179 (21.6)
265 (32.0) 460 (55.6) 103 (12.4)
293 (35.3) 435 (52.5) 101 (12.2)
331 (40.0) 428 (51.7) 69 (8.3)
243 (29.4) 181 (21.9) 404 (48.8)
196 (23.7) 208 (25.1) 424 (51.2)
195 (23.5) 212 (25.6) 422 (50.9)
149 (18.0) 181 (21.9) 498 (60.1)
217 (26.2) 305 (36.8) 306 (37.0) 45 (5.4) 1.2 6 1.4 39 (4.7) 2020 6 453 33.5 6 28.0 13.0 6 2.3 10.3 6 1.9 2.7 6 0.7 5.4 6 1.8 2.9 6 1.4 2.0 6 1.4 0.8 6 1.0 818 6 286 10.6 6 3.7 44.5 6 30.7 26.2 6 26.8 13.2 6 16.3 1.1 6 0.2
187 (22.6) 278 (33.6) 363 (43.8) 41 (5.0) 1.3 6 1.4 49 (5.9) 2350 6 428 30.3 6 24.4 18.1 6 1.2 14.4 6 1.1 3.7 6 0.6 7.2 6 2.1 3.9 6 1.6 3.4 6 2.0 1.1 6 1.3 959 6 284 12.9 6 3.8 46.6 6 30.6 32.4 6 27.7 18.8 6 20.6 1.2 6 0.2
180 (21.7) 303 (36.6) 346 (41.7) 49 (5.9) 1.3 6 1.7 54 (6.5) 2593 6 455 28.8 6 22.3 22.4 6 1.4 18.0 6 1.3 4.4 6 0.7 8.7 6 2.5 4.7 6 1.8 4.5 6 2.2 1.5 6 1.5 1064 6 301 14.7 6 4.2 47.0 6 29.1 36.7 6 29.1 24.7 6 22.4 1.1 6 0.2
180 (21.7) 289 (34.9) 359 (43.4) 36 (4.4) 1.3 6 1.7 53 (6.4) 2899 6 560 23.8 6 22.3 30.7 6 5.8 24.8 6 4.9 5.9 6 1.3 12.1 6 4.2 6.0 6 2.3 6.7 6 3.4 2.1 6 2.2 1216 6 343 17.6 6 5.5 45.3 6 30.0 42.2 6 37.0 34.2 6 31.3 1.1 6 0.2
,0.0001
0.05
0.5 0.07 0.4 ,0.0001 ,0.0001 ,0.0001 ,0.0001 ,0.0001 ,0.0001 ,0.0001 ,0.0001 ,0.0001 ,0.0001 ,0.0001 0.9 ,0.0001 ,0.0001 0.009
´ ´ Values are means 6 SDs or n (%). PSA, prostate-specific antigen; SU.VI.MAX, Supplementation en Vitamines et Mineraux Antioxydants. P value for the comparison between quartiles of dietary fiber intake using x2 tests or Fisher tests (P-trend) as appropriate. Data for dietary and biomarker variables were logtransformed to improve normality. All statistical tests were 2 sided. 3 Among first-degree relatives. 4 Including fiber from soy and soy products. 1 2
506
Deschasaux et al.
Downloaded from jn.nutrition.org at UNIVERSITY OF PITTSBURGH FALK LIBRARY on January 17, 2015
to the study design [initial SU.VI.MAX trial intervention group (yes/no), number of dietary records (continuous)], major sociodemographic variables [age (timescale) and educational level (primary, secondary, or university)], lifestyle factors [smoking status (never, former, or current), physical activity (irregular,
The Journal of Nutrition Nutritional Epidemiology
Dietary Total and Insoluble Fiber Intakes Are Inversely Associated with Prostate Cancer Risk1–3 M´elanie Deschasaux,4* Camille Pouchieu,4 Mathilde His,4 Serge Hercberg,4,5 Paule Latino-Martel,4 and Mathilde Touvier4
Abstract Although experimental data suggest a potentially protective involvement of dietary fiber in prostate carcinogenesis, very few prospective studies have investigated the relation between dietary fiber intake and prostate cancer risk, and those have had inconsistent results. Our objective was to study the association between dietary fiber intake (overall, insoluble, soluble, and from different sources, such as cereals, vegetables, fruits, and legumes) and prostate cancer risk. Stratifications by excess weight status, insulin-like growth factors, and amount of alcohol intake were also considered. ´ ´ This prospective analysis included 3313 men from the Supplementation en Vitamines et Mineraux Antioxydants (SU.VI. MAX) cohort who completed at least 3 24-h dietary records. One hundred thirty-nine incident prostate cancers were diagnosed between 1994 and 2007 (median follow-up of 12.6 y). Associations between quartiles of energy-adjusted dietary fiber intake and prostate cancer risk were characterized by multivariate Cox proportional hazards models. Prostate cancer risk was inversely associated with total dietary fiber intake (HR of quartile 4 vs. quartile 1 = 0.47; 95% CI: 0.27, 0.81; P = 0.001), insoluble (HR = 0.46; 95% CI: 0.27, 0.78; P = 0.001), and legume (HR = 0.55; 95% CI: 0.32, 0.95; P = 0.04) fiber intakes. In contrast, we found no association between prostate cancer risk and soluble (P = 0.1), cereal (P = 0.7), vegetable (P = 0.9), and fruit (P = 0.4) fiber intakes. In conclusion, dietary fiber intake (total, insoluble, and from legumes but not soluble or from cereals, vegetables, and fruits) was inversely associated with prostate cancer risk, consistent with mechanistic data. This trial was registered at clinicaltrials.gov as NCT00272428. J. Nutr. 144: 504–510, 2014.
Introduction Several mechanisms are involved in prostate cancer development. First, experimental (1–3) and epidemiologic (3,4) data suggest that chronic inflammation stimulates prostate carcinogenesis. Second, insulin resistance and hyperinsulinemia, which lead to several metabolic modifications, such as higher insulinlike growth factor (IGF)6-1 and IGF-2 bioactivity and lower sexhormone binding globulin (SHBG) concentrations (1,5,6), have 1
ˆ le Ile-de-France (public funding from the Paris ´ Supported by the Canceropo region; to M.D.). The funders had no role in the design, implementation, analysis, or interpretation of the data. 2 Author disclosures: M. Deschasaux, C. Pouchieu, M. His, S. Hercberg, P. Latino-Martel, and M. Touvier, no conflicts of interest. 3 Supplemental Tables 1–4 are available from the ‘‘Online Supporting Material’’ link in the online posting of the article and from the same link in the online table of contents at jn.nutrition.org. 6 Abbreviations used: EPIC, European Prospective Investigation into Cancer and Nutrition; IGF, insulin-like growth factor; IGFBP, insulin-like growth factor binding protein; PSA, prostate-specific antigen; SHBG, sex-hormone binding globulin; ´ ´ SU.VI.MAX, Supplementation en Vitamines et Mineraux Antioxydants; WCRF, World Cancer Research Fund. * To whom correspondence should be addressed. E-mail: m.deschasaux@uren. smbh.univ-paris13.fr.
504
been associated with increased risk of prostate cancer in experimental (5,7) and epidemiologic (7–11) studies. Finally, increased concentrations of circulating estrogens and androgens could be related to a greater risk of prostate cancer (12). Experimental data suggested that dietary fiber may do the following: 1) have anti-inflammatory properties (13–15); 2) improve insulin sensitivity (16,17) and reduce IGF bioactivity, notably by increasing insulin-like growth factor binding protein (IGFBP) 3 concentrations (18); 3) upregulate SHBG concentration (19); and 4) influence steroid hormone concentrations by decreasing circulating estrogens and androgens (20,21), thereby supporting the hypothesis that dietary fiber has a protective effect against prostate carcinogenesis. In 2007, the World Cancer Research Fund (WCRF)/American Institute for Cancer Research considered that the epidemiologic evidence of an association between dietary fiber intake and prostate cancer risk was too limited to allow a conclusion (22), because only 1 unadjusted null-result prospective cohort study was available (23). Since then, 3 prospective studies on dietary fiber and prostate cancer risk have been published (24–26). Overall, they did not detect any relation (24–26), except for an inverse association observed in the European Prospective
ã 2014 American Society for Nutrition. Manuscript received December 10, 2013. Initial review completed January 8, 2014. Revision accepted January 27, 2014. First published online February 19, 2014; doi:10.3945/jn.113.189670.
Downloaded from jn.nutrition.org at UNIVERSITY OF PITTSBURGH FALK LIBRARY on January 17, 2015
4 Sorbonne Paris Cit´e Research Center, Nutritional Epidemiology Research Team, U1153 National Institute of Health and Medical Research, U1125 National Institute for Agricultural Research, National Conservatory of Arts and Crafts, Paris 13 University, Bobigny, France; and 5Public Health Department, Avicenne Hospital, Bobigny, France
Materials and Methods Participants. The Suppl´ementation en Vitamines et Min´eraux Antioxydants (SU.VI.MAX) study was at first designed as a randomized, double-blind, placebo-controlled primary prevention trial (registered at clinicaltrials.gov as NCT00272428) aiming to assess the effect of daily supplementation with nutritional doses of antioxidants on the incidence of cardiovascular diseases and cancers (32). In 1994–1995, 13,017 individuals (7876 women and 5141 men) were enrolled for an 8-y intervention study. Follow-up of health events lasted until September 2007. The SU.VI.MAX study was conducted according to the Declaration of Helsinki guidelines and was approved by the Paris-Cochin Hospital Ethics Committee for Studies with Human Subjects (CCPPRB nos. 706 and 2364, respectively) and the French National Commission for Computed Data and Individual Freedom (CNIL nos. 334641 and 907094, respectively). Written informed consent was obtained from all participants. Baseline data collection. Information about sociodemographics, smoking status, physical activity, and family history of prostate cancer were collected by self-administered questionnaires for all participants at enrollment. Anthropometric measures were obtained during a medical examination by the study nurses and physicians. Every 2 mo during the trial phase (1994–2002), participants were invited to complete a 24-h dietary record via the Minitel Telematic Network, a French telephone-based terminal equivalent to an Internet prototype used widely at the beginning of the study. The records were randomly distributed between weeks and weekends and over seasons to take into account intra-individual variability. Participants assessed portion sizes with the aid of a validated picture booklet (33), and the amounts consumed from composite dishes were estimated using French recipes validated by food and nutrition professionals. The mean daily energy, alcohol, and nutrient intakes were estimated using a published
French food composition table (34). Total dietary fiber and soluble fiber contents were obtained using the Association of Official Analytical Chemists method (35). Dietary fiber intakes in the SU.VI.MAX study were described previously (36). During the trial phase, participants were advised against taking any self-prescribed supplementation. Fiber intake from supplements is therefore most likely negligible in our study. In the SU.VI.MAX cohort, a study was conducted to validate an FFQ that was completed by the participants in 2007. This FFQ was compared with 12 dietary records across 1 y (considered as the reference method). Energy-adjusted deattenuated correlation coefficients for total dietary fiber intake between this FFQ and the 12 dietary records were 0.56 for men and 0.63 for women (37). These coefficients were similar to those found in the EPIC cohort (0.54) (38). A 35 mL fasting venous blood sample was collected in vacutainer tubes. Blood samples were centrifuged immediately after blood draw, and plasma aliquots were then preserved in sodium heparin. Less than 1 h after blood draw, plasma aliquots were stored at 220°C in dry ice for shipment to the central biobank (maximum of 24 h), where they were stored frozen in liquid nitrogen. These plasma samples were used to determine the baseline concentrations of selenium, total prostate-specific antigen (PSA) measured by immunoassay (Roche Diagnostics), IGF1 measured by immunometry on a Immulite analyzer (Diagnostics Products Corporation), and IGF-2 measured by immunoradiometric assays (Diagnostic Systems Laboratories). Coefficients of variation for the laboratory assays were 5.5% for total PSA, 5.3% for IGF-1, and 6.8% for IGF-2. Total PSA, IGF-1, and IGF-2 baseline concentrations were measured in 2002–2003 as part of a post hoc research protocol (39,40). The SU.VI.MAX investigators contacted the participants and their treating physicians if their baseline PSA value was >4.0 mg/L. Additional medical investigations were left to the discretion of the treating physician (40). Case ascertainment. During the follow-up period, participants were invited to self-report health events (through a monthly questionnaire). Investigations were then conducted to obtain medical data from participants, physicians, and/or hospitals. All information was reviewed by an independent physician expert committee, and cases were validated with the pathologic reports and classified using the International Chronic Diseases Classification, 10th Revision, Clinical Modification (41). Cancers were validated by histologic reports, and all prostate cancer cases were documented by a pathology report (32,39). All first-incident primary prostate cancers were considered as cases in this study. Statistical analyses. From the 5141 male participants in the SU.VI. MAX study, we excluded 41 men who reported a cancer diagnosis before the start of the follow-up. Among the remaining participants, 3313 provided at least 3 valid 24-h dietary records within the first 2 years of follow-up and thus remained available for analysis. Participants contributed person-time until the date of diagnosis of prostate cancer, the date of last completed questionnaire, the date of death, or September 2007, whichever occurred first. Participants who reported a cancer other than prostate cancer during the study period (n = 163) were included and censored at the date of diagnosis (except those with basal cell skin carcinoma, which was not considered to be cancer). Nutrient intakes were estimated by the mean intake calculated from all dietary records for each man. Baseline characteristics of participants were compared between quartiles of total dietary fiber intake using x2 tests or Fisher tests (ANOVA models). HRs and 95% CIs obtained from Cox proportional hazards models, with age as the primary time variable, were used to characterize the association between quartiles of dietary fiber intake and incident of prostate cancer. We confirmed that the assumptions of proportionality were satisfied through examination of the log–log (survival) vs. log–time plots. Different categories of dietary fiber were tested according to their chemical properties (soluble and insoluble fibers) and according to their food sources (cereal, vegetable, fruit, and legume fibers). Tests for linear trend were performed using the ordinal score on quartiles of fiber intake. Correlations between the different types of dietary fiber intakes were analyzed using Spearman correlation coefficients. Multivariate models were adjusted for factors constitutive Dietary fiber intake and prostate cancer risk
505
Downloaded from jn.nutrition.org at UNIVERSITY OF PITTSBURGH FALK LIBRARY on January 17, 2015
Investigation into Cancer and Nutrition (EPIC) cohort study for fruit fiber among men aged $65 y (26). However, these studies (25,26) used FFQs to assess dietary fiber intakes, which may weaken associations between food intakes and health outcomes (27,28). Thus, new prospective studies are needed to further investigate the relation between dietary fiber intake and prostate cancer risk. In addition, ‘‘dietary fiber’’ refers to a large category of molecules with various biochemical properties (such as solubility) and fermentation products, probably resulting in differential physiologic effects on the management of inflammation, insulin resistance, or circulating steroid hormones and, thus, prostate carcinogenesis (29,30). Nonetheless, only 1 prospective study on prostate cancer risk has investigated separately the different sources of dietary fiber (26), and none has distinguished soluble and insoluble fibers. Furthermore, to our knowledge, previous prospective studies on dietary fiber and prostate cancer risk did not take into account IGF concentrations or stratify for alcohol intake, and only 1 considered stratification by BMI with null result (25). Because overweight and obesity are linked to a low-grade chronic inflammation status (2,6), IGF-1 and IGF-2 concentrations are related to prostate cancer risk (7–10), and alcohol intake is associated with higher concentrations of circulating estrogens (31), we hypothesized that excess weight, IGF-1 and IGF-2 concentrations, and alcohol intake may modulate the association between dietary fiber intake and prostate cancer risk. Hence, our objective was to prospectively investigate the association between different types of dietary fiber (overall, insoluble, soluble, and from different food sources, such as cereals, vegetables, fruits, and legumes) and prostate cancer risk using repeated dietary records, and explore its potential modulation by overweight/obesity status, IGF-1 and IGF-2 plasma concentrations, and alcohol intake.
type of dietary fiber were tested by introducing the product of the 2 variables in the models. Interaction with the SU.VI.MAX trial antioxidant supplementation group was also tested. All tests were 2 sided, and P < 0.05 was considered statistically significant. SAS version 9.3 (SAS Institute) was used for analyses.
Results During a median follow-up of 12.6 y (37,877 person-years), 139 men developed a first primary prostate cancer with a mean age at diagnosis of 63 y. A total of 5.2% of the participants were lost to follow-up. Table 1 presents the characteristics of the participants according to quartiles of total dietary fiber intake. Participants in the upper quartile tended to be older, taller, nonsmokers, more active physically, and had lower BMIs. The mean dietary fiber intake was 21 6 7.3 g/d. Only 25% of the study population reached the 25 g/d minimal French recommendation (43). Main contributors to dietary fiber intake were cereals
TABLE 1 Baseline characteristics of participants (n = 3313) according to quartiles of total dietary fiber intake, SU.VI.MAX Cohort, France, 1994–20071
Age, y BMI, kg/m2 $25 kg/m2, n (%) Height, cm Intervention group, n (%) yes Smoking status Never Former Current Physical activity, n (%) Irregular ,1 h/d walking or equivalent $1 h/d walking or equivalent Educational level, n (%) Primary Secondary University Family history of prostate cancer,3 n (%) yes PSA, mg/L $3 mg/L, n (%) Energy intake, kcal/d Alcohol intake, g/d Total dietary fiber intake, g/d Insoluble fiber, g/d Soluble fiber, g/d Cereal fiber, g/d Vegetable fiber, g/d Fruit fiber, g/d Legume fiber,4 g/d Calcium intake, mg/d Vitamin E intake, mg/d Processed meat intake, g/d Tomato products intake, g/d Legume intake, g/d Blood Selenium, mmol/L
Quartile 1 (n = 828)
Quartile 2 (n = 828)
Quartile 3 (n = 829)
Quartile 4 (n = 828)
P2
51.2 6 4.6 25.9 6 3.2 501 (60.5) 173 6 6.2 424 (51.2)
51.7 6 4.6 25.6 6 3.2 441 (53.3) 173 6 6.3 438 (52.9)
51.8 6 4.8 25.2 6 2.9 407 (49.1) 173 6 6.3 415 (50.1)
52.5 6 4.9 24.8 6 3.0 360 (43.5) 174 6 6.4 413 (49.9)
,0.0001 ,0.0001 ,0.0001 ,0.0001 0.6 ,0.0001
226 (27.3) 423 (51.1) 179 (21.6)
265 (32.0) 460 (55.6) 103 (12.4)
293 (35.3) 435 (52.5) 101 (12.2)
331 (40.0) 428 (51.7) 69 (8.3)
243 (29.4) 181 (21.9) 404 (48.8)
196 (23.7) 208 (25.1) 424 (51.2)
195 (23.5) 212 (25.6) 422 (50.9)
149 (18.0) 181 (21.9) 498 (60.1)
217 (26.2) 305 (36.8) 306 (37.0) 45 (5.4) 1.2 6 1.4 39 (4.7) 2020 6 453 33.5 6 28.0 13.0 6 2.3 10.3 6 1.9 2.7 6 0.7 5.4 6 1.8 2.9 6 1.4 2.0 6 1.4 0.8 6 1.0 818 6 286 10.6 6 3.7 44.5 6 30.7 26.2 6 26.8 13.2 6 16.3 1.1 6 0.2
187 (22.6) 278 (33.6) 363 (43.8) 41 (5.0) 1.3 6 1.4 49 (5.9) 2350 6 428 30.3 6 24.4 18.1 6 1.2 14.4 6 1.1 3.7 6 0.6 7.2 6 2.1 3.9 6 1.6 3.4 6 2.0 1.1 6 1.3 959 6 284 12.9 6 3.8 46.6 6 30.6 32.4 6 27.7 18.8 6 20.6 1.2 6 0.2
180 (21.7) 303 (36.6) 346 (41.7) 49 (5.9) 1.3 6 1.7 54 (6.5) 2593 6 455 28.8 6 22.3 22.4 6 1.4 18.0 6 1.3 4.4 6 0.7 8.7 6 2.5 4.7 6 1.8 4.5 6 2.2 1.5 6 1.5 1064 6 301 14.7 6 4.2 47.0 6 29.1 36.7 6 29.1 24.7 6 22.4 1.1 6 0.2
180 (21.7) 289 (34.9) 359 (43.4) 36 (4.4) 1.3 6 1.7 53 (6.4) 2899 6 560 23.8 6 22.3 30.7 6 5.8 24.8 6 4.9 5.9 6 1.3 12.1 6 4.2 6.0 6 2.3 6.7 6 3.4 2.1 6 2.2 1216 6 343 17.6 6 5.5 45.3 6 30.0 42.2 6 37.0 34.2 6 31.3 1.1 6 0.2
,0.0001
0.05
0.5 0.07 0.4 ,0.0001 ,0.0001 ,0.0001 ,0.0001 ,0.0001 ,0.0001 ,0.0001 ,0.0001 ,0.0001 ,0.0001 ,0.0001 0.9 ,0.0001 ,0.0001 0.009
´ ´ Values are means 6 SDs or n (%). PSA, prostate-specific antigen; SU.VI.MAX, Supplementation en Vitamines et Mineraux Antioxydants. P value for the comparison between quartiles of dietary fiber intake using x2 tests or Fisher tests (P-trend) as appropriate. Data for dietary and biomarker variables were logtransformed to improve normality. All statistical tests were 2 sided. 3 Among first-degree relatives. 4 Including fiber from soy and soy products. 1 2
506
Deschasaux et al.
Downloaded from jn.nutrition.org at UNIVERSITY OF PITTSBURGH FALK LIBRARY on January 17, 2015
to the study design [initial SU.VI.MAX trial intervention group (yes/no), number of dietary records (continuous)], major sociodemographic variables [age (timescale) and educational level (primary, secondary, or university)], lifestyle factors [smoking status (never, former, or current), physical activity (irregular,
