AJCN. First published ahead of print May 20, 2015 as doi: 10.3945/ajcn.114.098418.
Prospective association between alcohol intake and hormonedependent cancer risk: modulation by dietary fiber intake1,2 Anne-Sophie Chhim,3,4 Philippine Fassier,3,4 Paule Latino-Martel,3,4 Nathalie Druesne-Pecollo,3,4 Laurent Zelek,3–5 Lucie Duverger,5 Serge Hercberg,3,4,6 Pilar Galan,3,4 Mélanie Deschasaux,3,4 and Mathilde Touvier3,4 3
Sorbonne Paris Cité Epidemiology and Biostatistics Research Center (CRESS), Nutritional Epidemiology Research Team (EREN), Inserm U1153, Inra U1125, Cnam, Paris, France; 4Universities, Bobigny, France; and 5Oncology and 6Public Health Departments, Avicenne Hospital, Bobigny, France.
ABSTRACT Background: Alcohol intake is associated with increased circulating concentrations of sex hormones, which in turn may increase hormonedependent cancer risk. This association may be modulated by dietary fiber intake, which has been shown to decrease steroid hormone bioavailability (decreased blood concentration and increased sex hormone–binding globulin concentration). However, this potential modulation has not been investigated in any prospective cohort. Objectives: Our objectives were to study the relation between alcohol intake and the risk of hormone-dependent cancers (breast, prostate, ovarian, endometrial, and testicular) and to investigate whether dietary fiber intake modulated these associations. Design: This prospective observational analysis included 3771 women and 2771 men who participated in the Supplémentation en Vitamines et Minéraux Antioxydants study (1994–2007) and completed at least 6 valid 24-h dietary records during the first 2 y of follow-up. After a median follow-up of 12.1 y, 297 incident hormone-dependent cancer cases, including 158 breast and 123 prostate cancers, were diagnosed. Associations were tested via multivariate Cox proportional hazards models. Results: Overall, alcohol intake was directly associated with the risk of hormone-dependent cancers (tertile 3 vs. tertile 1: HR: 1.36; 95% CI: 1.00, 1.84; P-trend = 0.02) and breast cancer (HR: 1.70; 95% CI: 1.11, 2.61; P-trend = 0.04) but not prostate cancer (P-trend = 0.3). In stratified analyses (by sex-specific median of dietary fiber intake), alcohol intake was directly associated with hormone-dependent cancer (tertile 3 vs. tertile 1: HR: 1.76; 95% CI: 1.10, 2.82; P-trend = 0.002), breast cancer (HR: 2.53; 95% CI: 1.30, 4.95; P-trend = 0.02), and prostate cancer (HR: 1.37; 95% CI: 0.65, 2.89; P-trend = 0.02) risk among individuals with low dietary fiber intake but not among their counterparts with higher dietary fiber intake (P-trend = 0.9, 0.8, and 0.6, respectively). The P-interaction between alcohol and dietary fiber intake was statistically significant for prostate cancer (P = 0.01) but not for overall hormone-dependent (P = 0.2) or breast (P = 0.9) cancer. Conclusion: In line with mechanistic hypotheses and experimental data, this prospective study suggested that dietary fiber intake might modulate the association between alcohol intake and risk of hormonedependent cancer. This trial was registered at clinicaltrials.gov as NCT00272428. Am J Clin Nutr doi: 10.3945/ajcn.114.098418. Keywords alcohol, breast cancer, dietary fiber, hormone-dependent cancer, prospective study, prostate cancer
INTRODUCTION
Hormone-dependent cancers include cancer of the prostate and testis in men and breast, ovarian, and endometrial cancer in women. The sex steroid hormones, androgen, estrogen, and progestin, are the primary modulators of both normal development and maintenance of these organs, as well as malignant growths. Increased concentrations of circulating estrogens and androgens have been related to a greater risk of hormonedependent cancers (1–5). Among other potentially procarcinogenic mechanisms, alcohol intake has been shown to increase circulating concentrations of steroid hormones and act on their receptors (6–13). The positive association between alcohol intake and breast cancer risk is now well established (13–15). In contrast, this association is less clear for other hormone-dependent cancers (16–21) and deserves further prospective investigation. Indeed, the level of proof regarding the association between prostate (18), ovarian (20), and endometrial (19) cancer and alcohol intake was qualified by the World Cancer Research Fund and the American Institute for Cancer Research as “limited—no conclusion.” The simultaneous intake of nutritional bioactive compounds may modulate the association between alcohol and cancer risk. Dietary fibers represent a complex group of molecules that includes nonstarch polysaccharides (cellulose, hemicelluloses, pectins, chitins, and hydrocolloids such as gums, mucilages, and glucans), oligosaccharides (inulin, a-galactosides, fructooligosaccharides, and galacto-oligosaccharides), resistant starch, resistant dextrins, and other compounds associated with dietary fiber polysaccharides such as lignin and waxes (22). Dietary fibers are good candidates for this potential modulatory role (9), because epidemiologic and experimental data showed that they may reduce concentrations of circulating estrogens and androgens through various mechanisms (9, 23–28). This is one of the 1 Supported by a grant from the French Ministry of Research and Higher Education (A-SC) and a PhD grant from the Cancéropoˆle Ile-de-France (public funding; MD). 2 Address correspondence to M Touvier, Nutritional Epidemiology Research Team (EREN), SMBH Paris 13, 74 rue Marcel Cachin, F-93017, Bobigny, France. E-mail: [email protected]. Received August 25, 2014. Accepted for publication April 29, 2015. doi: 10.3945/ajcn.114.098418.
Am J Clin Nutr doi: 10.3945/ajcn.114.098418. Printed in USA. Ó 2015 American Society for Nutrition
Copyright (C) 2015 by the American Society for Nutrition
1 of 8
2 of 8
CHHIM ET AL.
mechanistic hypotheses that support the protective effect of dietary fiber intake on breast and prostate cancer risk observed in several cohorts (29–32). To our knowledge, no prospective study on alcohol consumption and hormone-dependent, breast, or prostate cancer risk has investigated a potential modulation of these relations by dietary fiber intake. Thus, our objectives were to prospectively investigate the association between alcohol intake and hormone-dependent cancer risk (overall, breast, and prostate cancers) and to explore its potential modulation by dietary fiber intake. METHODS
Study population The Supplémentation en Vitamines et Minéraux Antioxydants (SU.VI.MAX) study was a population-based, double-blind, placebo-controlled, randomized trial (registered at clinicaltrials.gov as NCT00272428) initially designed to assess the effect of a daily antioxidant supplementation on the incidence of cardiovascular disease and cancer (33). A total of 13,017 subjects were enrolled in 1994–1995. All participants took a single daily capsule containing a combination of 120 mg ascorbic acid, 30 mg vitamin E, 6 mg b-carotene, 100 mg selenium, and 20 mg zinc or a placebo. The intervention study lasted 8 y, and observational follow-up of health events was maintained until September 2007. Subjects provided written informed consent, and the study was approved by the Ethics Committee for Studies with Human Subjects at the Paris-Cochin Hospital (no. 706) and the Commission Nationale de l’Informatique et des Libertés (no. 334641). Dietary data collection During follow-up, participants were invited to complete a dietary record every 2 mo, in which they reported all foods and beverages consumed during a period of 24 h. These dietary records were randomly distributed between week and weekend days and over all seasons to take into account intraindividual variability. Dietary records from the first 2 y of follow-up were used in the present study to comply with its prospective design. Food record completion was done through the Minitel Telematic Network, a French telephone-based terminal that was an Internet prototype. Portion sizes were assessed via a validated picture booklet (34), and the amounts consumed from composite dishes were estimated by using French recipes validated by food and nutrition professionals. The mean daily energy, alcohol, dietary fiber, and other nutrient intakes were estimated by using a published French food composition table (35). Total dietary fiber content was obtained by using the Association of Official Analytic Chemists method (36). Dietary fiber intakes in the SU.VI. MAX study have been described previously (30, 31, 37). Data collection for covariates At enrollment, self-administered questionnaires were filled by participants, reporting sociodemographic characteristics (sex, date of birth, and educational level); smoking status; medication use (including baseline use of hormonal treatment of menopause); number of live births; family history of overall, breast, and prostate cancer; and menopausal status. Baseline physical
activity was self-reported by asking the participants if they currently practiced a physical activity on a regular basis and, if yes, if it was equivalent to 1 h of walking/d or more/less. Anthropometric measurements were taken during a baseline clinical examination. Weight was measured to the nearest 0.1 kg with an electronic scale (Seca) by study nurses and physicians. Subjects wore light clothing and no shoes. Height was measured to the nearest 1 cm with a wall-mounted stadiometer under the same conditions. A 35-mL fasting venous blood sample was collected in evacuated tubes. These plasma samples were used to determine the baseline concentrations of selenium and total prostate-specific antigen measured by immunoassay (Roche Diagnostics). Case ascertainment Health events occurring during the follow-up were selfreported by participants. Medical data were then requested from participants, physicians, and/or hospitals and reviewed by an independent physician expert committee. Pathologic reports were used to validate the cases and to extract cancer type characteristics. Cases were classified by using the International Chronic Disease Classification, 10th Revision, Clinical Modification (38). All incident first primary hormone-dependent cancers were considered as cases in this study: breast (in women), prostate, ovarian, endometrial, and testicular cancer. Statistical analyses From the 13,017 participants in the SU.VI.MAX study, 161 were excluded because they had reported a cancer diagnosis before the start of the follow-up. Among the remaining participants, 6542 provided at least 6 valid 24-h dietary records within the first 2 y of follow-up and thus were available for analysis. For all covariates, ,5% of values were missing and were replaced by the respective mode value. Baseline characteristics of participants were compared between cases and noncases by using x2 tests or Student’s t tests, as appropriate. HRs and 95% CIs obtained from Cox proportional hazards models with age as the primary time variable were used to characterize the association between tertiles of alcohol intake and risk of incident hormone-dependent (all sites), breast, and prostate cancer. Participants contributed person-time until the date of diagnosis of the studied cancer type (overall hormonedependent, breast, or prostate), the date of the last completed questionnaire, the date of death, or September 2007, whichever occurred first. Participants who reported another cancer than the one studied during the study period were included and censored at the date of that diagnosis (except for basal cell carcinoma, which was not considered cancer). We confirmed that the assumptions of proportionality were satisfied through examination of the loglog (survival) vs. log-time plots. Tests for linear trend were performed by using the ordinal score of tertiles of alcohol intake. Multivariate models were adjusted for age (time scale in the Cox model), sex (for overall hormone-dependent cancers only), active intervention group in the SU.VI.MAX trial (yes/no), number of dietary records (continuous), height (continuous), BMI (in kg/m2; ,25, $25 to ,30, or $30), physical activity (irregular or ,1 or $1 h/d walking or equivalent), smoking status (never, former, or current), educational level (primary, secondary, or university),
ALCOHOL, FIBER, AND HORMONE-DEPENDENT CANCER RISK
mean daily dietary fiber intake (continuous), mean daily energy intake without alcohol (continuous), and first-degree family history of cancer (overall, breast, or prostate) (yes or no). For breast cancer analyses, additional adjustments were performed for baseline menopausal status (yes or no), use of hormonal treatment of menopause (yes or no), and number of live births (continuous). For prostate cancer analyses, additional adjustment was performed for baseline prostate-specific antigen concentration (continuous). Interaction between dietary fiber intake (, vs. $ sex-specific population median) and alcohol intake (tertiles) was tested by introducing an interaction term into the models (product of the 2 variables). Analyses were conducted overall and then stratified by dietary fiber intake. Models were also computed separately for premenopausal and postmenopausal breast cancer (women contributed to the premenopausal model until their age of menopause, and conversely, women contributed to the postmenopausal model from their age of menopause). Sensitivity analyses were performed by excluding heavy drinkers (.100 g alcohol/d) and cancers diagnosed during the first 2 y of follow-up. Further adjustment for a “healthy” food pattern was also tested. This dietary pattern was extracted by principal component analysis, by using the SAS “proc factor” procedure, from the mean intakes of 31 food groups across all 24-h records collected during the first 2 y of the study. Additional adjustments were also tested for dietary intake of total fat for women and for dietary intake of calcium, processed meat, tomato-based products, and vitamin E and plasma selenium concentration for men (continuous). All tests were 2-sided, and P , 0.05 was considered statistically significant. SAS version 9.3 (SAS Institute) was used for analyses. RESULTS
Mean follow-up time was 12.1 y. Of the subjects, 5.2% were lost to follow-up. A total of 297 incident hormone-dependent cancer cases were diagnosed during the follow-up: 158 breast (40 premenopausal and 118 postmenopausal), 123 prostate, and 16 other hormone-dependent cancers (11 ovarian and 5 endometrial). No case of testicular cancer was diagnosed. Mean age at diagnosis was 56.2 y for breast cancer and 63.1 for prostate cancer. Among breast cancer cases, 58.9% were estrogen receptor positive, and 43.0% were progesterone receptor positive. Regarding histologic type of breast cancer, 62.0% were ductal, 14.6% were lobular, and 23.4% were other types. Among prostate cancer cases, 5.1% had a Gleason score between 2 and 4, 87.2% between 5 and 7, and 7.7% between 8 and 10. The exposures of interest were as follows: 81% of alcohol intake came from wine (red, white, or rosé), 9% from beer and cider, and 10% from spirits (data not shown). Sources of dietary fibers were cereals (38.0%), vegetables (22.2%), fruit (21.2%), and legumes (6.1%). Of the dietary fibers, 20.1% were soluble and 79.8% were insoluble. Characteristics of the subjects are presented in Table 1. Compared with noncases, subjects with hormone-dependent cancer tended to be older, were more frequently obese, had more family history of cancer, were more frequently menopausal at baseline, and took more hormonal treatment of menopause (for women).
3 of 8
Associations between tertiles of alcohol intake and overall hormone-dependent, breast, and prostate cancer risks are presented in Table 2. Tertiles of alcohol intake were directly associated with hormone-dependent (tertile 3 vs. tertile 1: HR: 1.36; 95% CI: 1.00, 1.84; P-trend = 0.02) and breast (HR: 1.70; 95% CI: 1.11, 2.61; P-trend = 0.04) cancer risks, whereas no statistically significant association was observed for prostate cancer (P-trend = 0.3). Results for the associations between tertiles of alcohol intake and overall hormone-dependent, breast, and prostate cancer risks stratified by dietary fiber intake are shown in Table 3. Although the P-interaction between alcohol and dietary fiber intake was statistically significant for prostate cancer (P = 0.01) but not for overall hormone-dependent (P = 0.2) or breast (P = 0.9) cancer, differential results were observed across dietary fiber strata. Indeed, alcohol intake was directly associated with hormonedependent (P-trend = 0.002), breast (P-trend = 0.02), and prostate (P-trend = 0.02) cancer risks among individuals with lower dietary fiber intake but not among those with higher dietary fiber consumption (P-trend = 0.9, 0.8, and 0.6, respectively), despite similar statistical power across fiber strata. Similar results were observed when the stratification was performed by the median of soluble and insoluble fiber intake (data not tabulated): P-trend = 0.008 for hormone-dependent cancer, 0.04 for breast cancer, and 0.007 for prostate cancer in subjects with low soluble fiber intake, whereas the corresponding P values were 0.5, 0.3, and 0.4 in subjects with higher soluble fiber intake. Similarly, P-trend = 0.01 for hormonedependent cancer, 0.02 for breast cancer, and 0.01 for prostate cancer in subjects with low insoluble fiber intake, whereas the corresponding P values were 0.5, 0.5, and 0.6 in subjects with higher insoluble fiber intake. Also, similar results were observed regarding breast cancer risk when analyses were restricted to postmenopausal women: alcohol intake was directly associated with breast cancer risk among subjects with lower dietary fiber intake (tertile 3 vs. tertile 1: HR: 2.39; 95% CI: 1.10, 5.18; P-trend = 0.06), whereas no association was observed among those with higher dietary fiber intake (P-trend = 0.3). The statistical power was insufficient to perform such an analysis in premenopausal women. Further adjustments for a healthy dietary pattern (all models), for fat intake (breast cancer models), or for dietary intake of calcium, processed meat, tomato products, vitamin E, and plasma selenium concentration (prostate cancer models) provided similar results (data not shown). Sensitivity analyses excluding cancer cases diagnosed within the first 2 y of follow-up (26 hormonedependent, 21 breast, and 5 prostate cancer cases) did not modify the findings, nor did sensitivity analyses excluding subjects who consumed more than 100 g/d alcohol (heavy drinkers, n = 39). DISCUSSION
This prospective study confirmed the direct association between alcohol intake and breast cancer risk. It also suggested that dietary fiber intake may modulate the association between alcohol intake and hormone-dependent cancers, which, to our knowledge, had never been investigated in any epidemiologic study, despite its mechanistic plausibility. Indeed, although the P-interaction was statistically significant only for prostate cancer, alcohol intake was directly associated with overall
4 of 8
CHHIM ET AL.
TABLE 1 Baseline characteristics of the study population (n = 6542), Supplémentation en Vitamines et Minéraux Antioxydants cohort, France, 1994–2007
Age, y Sex, n (%) Male Female Educational level, n (%) Primary Secondary University Smoking status, n (%) Nonsmoker Former smoker Current smoker Physical activity, n (%) Irregular/none ,1 h/d of walking or equivalent .1 h/d of walking or equivalent Height, cm BMI,3 n (%) Underweight Normal weight Overweight Obese Menopausal status,4 n (%) Nonmenopausal Menopausal Use of hormonal treatment of menopause (yes),4 n (%) Number of live births4 Plasma prostate-specific antigen concentration,6 mg/L First-degree family history of cancer, n (%) All sites (yes) Breast cancer (yes)4 Prostate cancer (yes)6 Intervention group, n (%) Antioxidants Placebo Number of 24-h dietary records Energy intake without alcohol, kcal/d Alcohol intake, g/d Total dietary fiber intake, g/d Dietary fiber intake from cereals, g/d Dietary fiber intake from fruit, g/d Dietary fiber intake from vegetables, g/d Dietary fiber intake from legumes, g/d Dietary insoluble fiber intake, g/d Dietary soluble fiber intake, g/d
Noncases (n = 6245)
Hormone-dependent cancer cases (n = 297)
Breast cancer cases (n = 158)
Prostate cancer cases (n = 123)
49.1 6 6.42
52.0 6 6.3
49.7 6 6.4
54.9 6 4.7
2648 (42.4) 3597 (57.6%)
123 (41.4) 174 (58.6)
158 (100.0)
1274 (20.4) 2381 (38.1) 2590 (41.5)
61 (20.5) 104 (35.0) 132 (44.4)
33 (20.9) 53 (33.5) 72 (45.6)
24 (19.5) 47 (38.2) 52 (42.3)
3078 (49.3) 2345 (37.6) 822 (13.2)
147 (49.5) 109 (36.7) 41 (13.8)
87 (55.1) 42 (26.6) 29 (18.4)
50 (40.7) 61 (49.6) 12 (9.8)
80 (26.9) 91 (30.6) 126 (42.4) 166.9 6 8.3
50 (31.7) 53 (33.5) 55 (34.8) 162.6 6 6.2
29 (23.6) 31 (25.2) 63 (51.2) 173.2 6 6.3
P value1 ,0.0001 0.7
123 (100.0) 0.5
0.9
0.5 1519 1879 2847 166.7
(24.3) (30.1) (45.6) 6 8.4
144 3968 1764 369
(2.3) (63.5) (28.3) (5.9)
2517 1080 1083 1.9 1.2
(70.0) (30.0) (30.1) 6 1.2 6 1.3
97 (55.8) 77 (44.3) 70 (40.2) 2.1 6 1.2 3.4 6 3.7
92 (58.2) 66 (41.8) 61 (38.6) 2.0 6 1.2
2182 (34.9) 302 (8.4) 124 (4.7)
129 (43.4) 26 (14.9) 16 (13.0)
69 (43.7) 24 (15.2)
3113 (49.9) 3132 (50.2) 10.6 6 2.4 1983 6 539 18.8 6 20.7 19.0 6 6.5 7.2 6 3.3 4.0 6 2.6 4.2 6 1.9 1.2 6 1.4 15.2 6 5.4 3.8 6 1.3
140 (47.1) 157 (52.9) 10.6 6 2.4 1977 6 533 19.7 6 19.0 19.1 6 6.8 7.3 6 3.2 4.1 6 2.5 4.2 6 2.0 1.1 6 1.2 15.3 6 5.6 3.9 6 1.3
73 (46.2) 85 (53.8) 10.4 6 2.4 1778 6 414 12.9 6 12.1 17.6 6 5.2 6.6 6 2.4 3.8 6 2.1 4.0 6 1.9 1.1 6 1.3 14.0 6 4.3 3.6 6 1.0
9 189 71 28
(3.0) (63.6) (23.9) (9.4)
7 116 20 15
(4.4) (73.4) (12.7) (9.5)
1 64 45 13
0.7 0.04
(0.8) (52.0) (36.6) (10.6) ,0.00015
3.4 6 3.7 51 (41.5) 16 (13.0)
0.0055 0.075 ,0.00017 0.003 0.0035 ,0.00017 0.4
61 (49.6) 62 (50.4) 11.0 6 2.3 2276 6 544 30.2 6 22.1 21.3 6 8.1 8.4 6 3.9 4.4 6 3.1 4.6 6 2.1 1.1 6 1.2 17.1 6 6.7 4.2 6 1.5
0.9 0.8 0.4 0.7 0.5 0.6 0.8 0.6 0.8 0.6
P value for the comparison between overall hormone-dependent cancer cases and noncases, by Student’s t test or x2 test where appropriate. Mean 6 SD (all such values). 3 Cutoffs for BMI (in kg/m2) were ,18.5 for underweight, 18.5–25 for normal weight, 25–30 for overweight, and $30 for obesity. 4 In women only. 5 P value for the comparison between breast cancer cases and female noncases. 6 In men only. 7 P value for the comparison between prostate cancer cases and male noncases. 1 2
hormone-dependent, breast, and prostate cancer in subjects with low dietary fiber intake but not in subjects with higher dietary fiber intake. The direct association between alcohol intake and increased breast cancer risk is now well established (13, 15, 39). Regarding prostate cancer, a nonlinear dose-response meta-analysis of 22 cohort and 50 case-control studies showed a statistically sig-
nificant increased risk of alcohol intake of 25 and 50 g/d but not of 100 g/d (17). In the recent World Cancer Research Fund/ American Institute for Cancer Research meta-analysis, which included all available prospective studies (21), the summary HR for prostate cancer risk was not significant for total alcohol intake, but a weak, significantly increased risk was observed for wine and spirits. There was statistical evidence of nonlinearity in
5 of 8
ALCOHOL, FIBER, AND HORMONE-DEPENDENT CANCER RISK TABLE 2 Associations between tertiles of alcohol intake and overall hormone-dependent, breast, and prostate cancer risk, SU.VI. MAX cohort, France, 1994–20071 Cases/noncases, n
HR (95% CI)
P-trend
75/2105 112/2069 110/2071
1 1.52 (1.13, 2.04) 1.36 (1.00, 1.84)
0.02
36/1221 55/1202 67/1190
1 1.55 (1.01, 2.38) 1.70 (1.11, 2.61)
0.04
35/888 47/877 41/883
1 1.28 (0.82, 2.00) 0.89 (0.55, 1.45)
0.3
2
Hormone-dependent cancers (n = 297) T1 T2 T3 Breast cancer (n = 158)3 T1 T2 T3 Prostate cancer (n = 123)4 T1 T2 T3
1 From multivariate Cox proportional hazards models. Cutoffs for tertiles of alcohol intake were 3.0/12.1 g/d for women and 15.3/35.4 g/d for men. SU.VI.MAX, Supplémentation en Vitamines et Minéraux Antioxydants; T1, tertile 1; T2, tertile 2; T3, tertile 3. 2 Ajusted for age, sex, active intervention group during SU.VI.MAX trial, number of dietary records, smoking status, educational level, physical activity, height, BMI, family history of cancer, energy intake without alcohol, and dietary fiber intake. 3 Same adjustment + menopausal status at baseline, use of hormonal treatment of menopause at baseline, number of live births, and family history of breast (instead of overall) cancer. 4 Same adjustment + baseline prostate specific antigen concentration and family history of prostate (instead of overall) cancer.
several of these relations, with a small decrease in advanced prostate cancer risk with higher intakes (.7 drinks/d for total alcohol and .3 drinks/d for spirits). In our study, most alcohol intake came from wine; thus, it was not possible to model the effects of the different types of alcohol. We observed a direct association between total alcohol intake and prostate cancer risk in the low dietary fiber strata: this association was statistically significant for moderate drinkers (second tertile) but not for higher drinkers (third tertile). This phenomenon, in line with previous nonlinear data, deserves further investigation. One explanation could pertain to diagnosis bias. Indeed, prostate cancer is often asymptomatic and discovered with symptoms only at advanced stages. Higher drinkers may generally be less careful about their health and may have been less inclined to consult a physician. Another explanation could be related to misclassification of sicker men who had stopped drinking at baseline but drank before. Epidemiologic evidence is still insufficient regarding alcohol intake and the risk of other hormone-dependent cancers (16, 18–20). Statistical power was too limited to perform specific analyses for cancer sites other than breast and prostate in this study. The most original finding of our study was that the main associations tended to be modulated by dietary fiber intake in stratified analyses. Caution is needed when interpreting these results because the interaction was significant for prostate cancer only. This may be a result of the low statistical power of the interaction test (40) and thus requires confirmation by future prospective studies. No previous epidemiologic study has directly investigated this potential modulation. However, one prospective study specifically evaluated the relation of alcohol and dietary fiber intake with circulating sex hormone concentrations among premenopausal women and observed opposing associations: direct for alcohol and inverse for dietary fiber intake (9). Conversely, few prospective studies have investigated the relation between dietary fiber intake and cancer risk stratified by alcohol
consumption (31, 32, 41). In 2 prospective studies on breast cancer risk, no modulation by alcohol intake was detected (32, 41). In a previous investigation within the SU.VI.MAX cohort, dietary fiber intake was prospectively associated with decreased prostate cancer risk in men with higher alcohol intake but not in men with alcohol intake below the population median (31). The potential modulation of the relation between alcohol and hormone-dependent cancer risk by dietary fiber intake is in line with our initial hypothesis and with mechanistic and experimental data. Dietary fibers are metabolized by gut microbiota and may influence hormone-sensitive cancer risk through decreased hormonal activity (42, 43). Several mechanisms may be involved in this phenomenon. First, dietary fibers are thought to modify the enterohepatic circulation of estrogens by decreasing the activity of the colonic b-D-glucuronidase, an enzyme allowing estrogens to reenter the circulation (25, 27). Thus, fewer estrogens reenter the circulation and more estrogens are excreted (27). Second, dietary fiber intake has been associated with increased sex hormone–binding globulin concentrations (26) and thus decreased bioavailability of estrogens (44), which probably results from the beneficial influence of dietary fibers on insulin resistance (45) [insulin resistance being associated with a decrease in sex hormone–binding globulin concentrations (44)]. Our results suggest that although alcohol may exert its carcinogenic effect via several plausible mechanisms [including a genotoxic effect of acetaldehyde, a role as a solvent for tobacco carcinogens, and via the production of reactive oxygen species and nitrogen species (6)], its capacity to increase circulating steroid hormone concentrations (6–13) probably plays a central role in hormone-dependent cancer etiology (1–5). Strengths of our study pertained to its prospective design with long follow-up, its originality with respect to existing literature, and precise assessment of dietary intake based on at least six 24-h dietary records. However, several limitations should be acknowledged. First,
6 of 8
CHHIM ET AL. TABLE 3 Associations between tertiles of alcohol intake and overall hormone-dependent, breast and prostate cancer risk, stratified by median of dietary fiber intake, SU.VI.MAX cohort, France, 1994–20071 Cases/noncases, n Hormone-dependent cancers (n = 297)3 Dietary fiber intake , median T1 T2 T3 Dietary fiber intake $ median T1 T2 T3 Breast cancer (n = 158)4 Dietary fiber intake , median T1 T2 T3 Dietary fiber intake $ median T1 T2 T3 Prostate cancer (n = 123)5 Dietary fiber intake , median T1 T2 T3 Dietary fiber intake $ median T1 T2 T3
HR (95% CI)
P-trend
P-interaction2 0.2
28/1020 58/984 59/1121
1 2.27 (1.43, 3.59) 1.76 (1.10, 2.82)
0.002
47/1085 54/1085 51/950
1 1.09 (0.74, 1.63) 1.12 (0.74, 1.70)
0.9
0.9 13/634 24/550 35/629
1 2.25 (1.13, 4.47) 2.53 (1.30, 4.95)
0.02
23/587 31/652 32/561
1 1.13 (0.65, 1.95) 1.24 (0.70, 2.19)
0.8
0.01 13/388 30/438 23/493
1 2.45 (1.23, 4.87) 1.37 (0.65, 2.89)
0.02
22/500 17/439 18/390
1 0.71 (0.37, 1.37) 0.76 (0.38, 1.51)
0.6
1
From multivariate Cox proportional hazards models. Cutoffs for tertiles of alcohol intake were 3.0 and 12.1 g/d for women and 15.3 and 35.4 g/d for men. Median of mean daily dietary fiber intake was 16.8 g/d for women and 20.5 g/d for men. SU.VI.MAX, Supplémentation en Vitamines et Minéraux Antioxydants; T1, tertile 1; T2, tertile 2; T3, tertile 3. 2 Between alcohol and dietary fiber intakes. 3 Adjusted for age, sex, active intervention group during SU.VI.MAX trial, number of dietary records, smoking status, educational level, physical activity, height, BMI, family history of cancer, energy intake without alcohol, and dietary fiber intake. 4 Same adjustment + menopausal status at baseline, use of hormonal treatment of menopause at baseline, number of live births, and family history of breast (instead of overall) cancer. 5 Same adjustment + baseline prostate specific antigen concentration and family history of prostate (instead of overall) cancer.
the relatively small number of cases may have limited our ability to detect some of the hypothesized associations. The number of cancer cases was not sufficient to perform specific analyses for each breast cancer type (receptor status and histologic type) or according to the Gleason score for prostate cancer. Second, the observed relations could be affected by unmeasured or residual confounding. Some behavioral/environmental factors could not be taken into account in this study because of a lack of data (e.g., exposure to ionizing or ultraviolet radiation). However, most pertinent behavioral risk factors (i.e., other dietary factors, weight status, smoking habits, and physical activity) were taken into account in our study. Thus, it is unlikely that residual confounding entirely explains the observed associations. Next, dietary fiber refers to a complex group of molecules, and all types of fibers may not have the same properties with regard to modulation of cancer risk. For instance, it has been suggested that oligosaccharides and resistant starch are involved in antiinflammatory pathways through interaction with gut microflora, resulting in the production of short-chain fatty acids and reinforcement of the intestinal barrier (46–48). However, the reference
method used to quantify total dietary fiber in our study did not allow the measurement of oligosaccharides and provided only a very limited measure of dietary resistant starch (no measure of native and physically inaccessible resistant starch) (36). We can hypothesize that the modulatory effect of dietary fiber on the alcohol-cancer relations suggested by our findings would be even stronger if all types of dietary fiber were better taken into account by more precise analytic methods. Besides, nutrient composition data did not allow us to distinguish the different types of fiber in the analyses (nonstarch polysaccharides, oligosaccharides, resistant starch, lignin, etc.). We observed similar results for soluble and insoluble fibers. However, beyond this aspect, information on specific types of fiber and/or their fermentation properties would have been more informative but was not available. Finally, subjects were volunteers involved in a longterm nutrition study and thus had more health-conscious behaviors than the general French population, such as lower tobacco use [13% vs. 34% current smokers (49)]. However, their dietary fiber intake (mean: 19 g/d) was similar to the one observed in a French nationally representative survey (19.0 g/d in adults aged 55–79 y) (50).
ALCOHOL, FIBER, AND HORMONE-DEPENDENT CANCER RISK
In conclusion, this prospective study suggested that dietary fiber intake may modulate the associations between alcohol intake and hormone-dependent, breast, and prostate cancer risks, with a statistically significant interaction for prostate cancer. Mechanistic data support this epidemiologic result: dietary fibers may contribute to counteract the procarcinogenic increase in circulating sex hormones induced by alcohol intake. These results provide useful insights to better understand the etiologic involvement of alcohol in carcinogenesis and may contribute to explain the contrasted findings of some prior prospective studies investigating alcohol intake and cancer risk. From a public health standpoint, these findings also suggest that several risk factors (such as alcohol consumption and low dietary fiber intake) may cumulate and act in synergy to increase cancer risk. Further experimental data and large prospective cohorts are needed to confirm these results. The authors thank Dr. Valentina A. Andreeva for English language editing and Younes Esseddik, Paul Flanzy, Mohand Ait Oufella, Yasmina Chelghoum, and Than Duong Van (computer scientists); Nathalie Arnault, Véronique Gourlet, Dr. Fabien Szabo, Laurent Bourhis, and Stephen Besseau (statisticians); and Rachida Mehroug (logistics assistant) for their technical contribution to the SU.VI.MAX study. The author responsibilities were as follows—A-SC, MD, and MT: designed the research; AS-C, SH, PG, and MT: conducted the research; A-SC: performed statistical analysis; A-SC and MT: wrote the manuscript; A-SC, PF, PL-M, ND-P, LZ, LD, SH, PG, MD, and MT: contributed to the data interpretation and revised each draft for important intellectual content; MT: had primary responsibility for the final content; and all authors: read and approved the final manuscript. The authors reported no conflicts of interest related to this study. The funders had no role in the design, implementation, analysis, or interpretation of the data.
REFERENCES 1. Bosland MC. The role of steroid hormones in prostate carcinogenesis. J Natl Cancer Inst Monogr 2000;27:39–66. 2. Danforth KN, Eliassen AH, Tworoger SS, Missmer SA, Barbieri RL, Rosner BA, Colditz GA, Hankinson SE. The association of plasma androgen levels with breast, ovarian and endometrial cancer risk factors among postmenopausal women. Int J Cancer 2010;126:199–207. 3. Key T, Appleby P, Barnes I, Reeves G. Endogenous sex hormones and breast cancer in postmenopausal women: reanalysis of nine prospective studies. J Natl Cancer Inst 2002;94:606–16. 4. Key TJ. Endogenous oestrogens and breast cancer risk in premenopausal and postmenopausal women. Steroids 2011;76:812–5. 5. Walker K, Bratton DJ, Frost C. Premenopausal endogenous oestrogen levels and breast cancer risk: a meta-analysis. Br J Cancer 2011;105:1451–7. 6. Boffetta P, Hashibe M. Alcohol and cancer. Lancet Oncol 2006;7:149–56. 7. García-Closas M, Herbstman J, Schiffman M, Glass A, Dorgan JF. Relationship between serum hormone concentrations, reproductive history, alcohol consumption and genetic polymorphisms in premenopausal women. Int J Cancer 2002;102:172–8. 8. International Agency for Research on Cancer. Consumption of alcoholic beverages. IARC Monogr Eval Carcinog Risks Hum 2012;100: 373–499. 9. Maskarinec G, Morimoto Y, Takata Y, Murphy SP, Stanczyk FZ. Alcohol and dietary fibre intakes affect circulating sex hormones among premenopausal women. Public Health Nutr 2006;9:875–81. 10. Onland-Moret NC, Peeters PH, van der Schouw YT, Grobbee DE, van Gils CH. Alcohol and endogenous sex steroid levels in postmenopausal women: a cross-sectional study. J Clin Endocrinol Metab 2005;90: 1414–9. 11. Rinaldi S, Peeters PH, Bezemer ID, Dossus L, Biessy C, Sacerdote C, Berrino F, Panico S, Palli D, Tumino R, et al. Relationship of alcohol intake and sex steroid concentrations in blood in pre- and postmenopausal women: the European Prospective Investigation into Cancer and Nutrition. Cancer Causes Control 2006;17:1033–43.
7 of 8
12. Seitz HK, Stickel F. Molecular mechanisms of alcohol-mediated carcinogenesis. Nat Rev Cancer 2007;7:599–612. 13. Seitz HK, Pelucchi C, Bagnardi V, La Vecchia C. Epidemiology and pathophysiology of alcohol and breast cancer: update 2012. Alcohol Alcohol 2012;47:204–12. 14. Baan R, Straif K, Grosse Y, Secretan B, El GF, Bouvard V, Altieri A, Cogliano V. Carcinogenicity of alcoholic beverages. Lancet Oncol 2007;8:292–3. 15. World Cancer Research Fund/American Institute for Cancer Research. Systematic literature review—continuous update project report: the associations between food, nutrition and physical activity and the risk of breast cancer. Washington (DC): American Institute for Cancer Research; 2010. 16. Kelemen LE, Bandera EV, Terry KL, Rossing MA, Brinton LA, Doherty JA, Ness RB, Kjaer SK, Chang-Claude J, Kobel M, et al. Recent alcohol consumption and risk of incident ovarian carcinoma: a pooled analysis of 5,342 cases and 10,358 controls from the Ovarian Cancer Association Consortium. BMC Cancer 2013;13:28. 17. Rota M, Scotti L, Turati F, Tramacere I, Islami F, Bellocco R, Negri E, Corrao G, Boffetta P, La VC, et al. Alcohol consumption and prostate cancer risk: a meta-analysis of the dose-risk relation. Eur J Cancer Prev 2012;21:350–9. 18. 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. 19. World Cancer Research Fund/American Institute for Cancer Research. Systematic literature review— continuous update project report: the associations between food, nutrition and physical activity and the risk of endometrial cancer. Washington (DC): American Institute for Cancer Research; 2013. 20. World Cancer Research Fund/American Institute for Cancer Research. Systematic literature review—continuous update project report: the associations between food, nutrition and physical activity and the risk of ovarian cancer. Washington (DC): American Institute for Cancer Research; 2014. 21. World Cancer Research Fund/American Institute for Cancer Research. Systematic literature review—continuous update project report: the associations between food, nutrition and physical activity and the risk of prostate cancer. Washington (DC): American Institute for Cancer Research; 2014. 22. European Food Safety Authority. Scientific opinion on dietary reference values for carbohydrates and dietary fibre. EFSA Panel on Dietetic Products, Nutrition, and Allergies. EFSA J 2010;8:1462. 23. Cohen LA, Zhao Z, Zang EA, Wynn TT, Simi B, Rivenson A. Wheat bran and psyllium diets: effects on N-methylnitrosourea-induced mammary tumorigenesis in F344 rats. J Natl Cancer Inst 1996;88:899–907. 24. Dorgan JF, Judd JT, Longcope C, Brown C, Schatzkin A, Clevidence BA, Campbell WS, Nair PP, Franz C, Kahle L, et al. Effects of dietary fat and fiber on plasma and urine androgens and estrogens in men: a controlled feeding study. Am J Clin Nutr 1996;64:850–5. 25. Gerber M. Fiber and breast cancer: another piece of the puzzle—but still an incomplete picture. J Natl Cancer Inst 1996;88:857–8. 26. Longcope C, Feldman HA, McKinlay JB, Araujo AB. Diet and sex hormone–binding globulin. J Clin Endocrinol Metab 2000;85:293–6. 27. Moore MA, Park CB, Tsuda H. Soluble and insoluble fiber influences on cancer development. Crit Rev Oncol Hematol 1998;27:229–42. 28. Ross JK, Pusateri DJ, Shultz TD. Dietary and hormonal evaluation of men at different risks for prostate cancer: fiber intake, excretion, and composition, with in vitro evidence for an association between steroid hormones and specific fiber components. Am J Clin Nutr 1990;51:365–70. 29. Aune D, Chan DS, Greenwood DC, Vieira AR, Rosenblatt DA, Vieira R, Norat T. Dietary fiber and breast cancer risk: a systematic review and meta-analysis of prospective studies. Ann Oncol 2012;23:1394–402. 30. Deschasaux M, Zelek L, Pouchieu C, His M, Hercberg S, Galan P, Latino-Martel P, Touvier M. Prospective association between dietary fiber intake and breast cancer risk. PLoS ONE 2013;8:e79718. 31. Deschasaux M, Pouchieu C, His M, Hercberg S, Latino-Martel P, Touvier M. Dietary total and insoluble fiber intakes are inversely associated with prostate cancer risk. J Nutr 2014;144:504–10. 32. Ferrari P, Rinaldi S, Jenab M, Lukanova A, Olsen A, Tjonneland A, Overvad K, Clavel-Chapelon F, Fagherazzi G, Touillaud M, et al. Dietary fiber intake and risk of hormonal receptor–defined breast cancer in the European Prospective Investigation into Cancer and Nutrition study. Am J Clin Nutr 2013;97:344–53.
8 of 8
CHHIM ET AL.
33. Hercberg S, Galan P, Preziosi P, Bertrais S, Mennen L, Malvy D, Roussel AM, Favier A, Briancon S. The SU.VI.MAX study: a randomized, placebo-controlled trial of the health effects of antioxidant vitamins and minerals. Arch Intern Med 2004;164:2335–42. 34. Le Moullec N, Deheeger M, Preziosi P, Montero P, Valeix P, RollandCachera MF, Potier de Courcy G, Christides JP, Galan P, Hercberg S. Validation du manuel photos utilisé pour l’enqueˆte alimentaire de l’étude SU.VI.MAX (Validation of the food portion size booklet used in the SU.VI.MAX study). Cah Nutr Diet 1996;31:158–64. 35. Hercberg S. Table de Composition SU.VI.MAX des Aliments. Paris (France): Les éditions INSERM/Economica; 2005. 36. Prosky L, Asp NG, Schweizer TF, DeVries JW, Furda I. Determination of insoluble, soluble, and total dietary fiber in foods and food products: interlaboratory study. J Assoc Off Anal Chem 1988;71:1017–23. 37. Lairon D, Bertrais S, Vincent S, Arnault N, Galan P, Boutron MC, Hercberg S. Dietary fibre intake and clinical indices in the French Supplementation en Vitamines et Mineraux AntioXydants (SU.VI. MAX) adult cohort. Proc Nutr Soc 2003;62:11–5. 38. World Health Organization. International classification of diseases and related health problems. 10th revision. Geneva (Switzerland): World Health Organization; 1993. 39. Park SY, Kolonel LN, Lim U, White KK, Henderson BE, Wilkens LR. Alcohol consumption and breast cancer risk among women from five ethnic groups with light to moderate intakes: the Multiethnic Cohort Study. Int J Cancer 2014;134:1504–10. 40. Marshall SW. Power for tests of interaction: effect of raising the type I error rate. Epidemiol Perspect Innov 2007;4:4. 41. Park Y, Brinton LA, Subar AF, Hollenbeck A, Schatzkin A. Dietary fiber intake and risk of breast cancer in postmenopausal women: the National Institutes of Health–AARP Diet and Health Study. Am J Clin Nutr 2009;90:664–71.
42. Aubertin-Leheudre M, Gorbach S, Woods M, Dwyer JT, Goldin B, Adlercreutz H. Fat/fiber intakes and sex hormones in healthy premenopausal women in USA. J Steroid Biochem Mol Biol 2008;112: 32–9. 43. Gaskins AJ, Mumford SL, Zhang C, Wactawski-Wende J, Hovey KM, Whitcomb BW, Howards PP, Perkins NJ, Yeung E, Schisterman EF. Effect of daily fiber intake on reproductive function: the BioCycle Study. Am J Clin Nutr 2009;90:1061–9. 44. Arcidiacono B, Iiritano S, Nocera A, Possidente K, Nevolo MT, Ventura V, Foti D, Chiefari E, Brunetti A. Insulin resistance and cancer risk: an overview of the pathogenetic mechanisms. Exp Diabetes Res 2012;2012:789174. 45. Robertson MD, Bickerton AS, Dennis AL, Vidal H, Frayn KN. Insulinsensitizing effects of dietary resistant starch and effects on skeletal muscle and adipose tissue metabolism. Am J Clin Nutr 2005;82:559– 67. 46. Fung KY, Ooi CC, Zucker MH, Lockett T, Williams DB, Cosgrove LJ, Topping DL. Colorectal carcinogenesis: a cellular response to sustained risk environment. Int J Mol Sci 2013;14:13525–41. 47. Topping DL. Cereal complex carbohydrates and their contribution to human health. J Cereal Sci 2007;46:220–9. 48. Vinolo MA, Rodrigues HG, Nachbar RT, Curi R. Regulation of inflammation by short chain fatty acids. Nutrients 2011;3: 858–76. 49. Castetbon K, Vernay M, Malon A, Salanave B, Deschamps V, Roudier C, Oleko A, Szego E, Hercberg S. Dietary intake, physical activity and nutritional status in adults: the French nutrition and health survey (ENNS, 2006–2007). Br J Nutr 2009;102:733–43. 50. Dubuisson C, Lioret S, Touvier M, Dufour A, Calamassi-Tran G, Volatier JL, Lafay L. Trends in food and nutritional intakes of French adults from 1999 to 2007: results from the INCA surveys. Br J Nutr 2010;103:1035–48.
Prospective association between alcohol intake and hormonedependent cancer risk: modulation by dietary fiber intake1,2 Anne-Sophie Chhim,3,4 Philippine Fassier,3,4 Paule Latino-Martel,3,4 Nathalie Druesne-Pecollo,3,4 Laurent Zelek,3–5 Lucie Duverger,5 Serge Hercberg,3,4,6 Pilar Galan,3,4 Mélanie Deschasaux,3,4 and Mathilde Touvier3,4 3
Sorbonne Paris Cité Epidemiology and Biostatistics Research Center (CRESS), Nutritional Epidemiology Research Team (EREN), Inserm U1153, Inra U1125, Cnam, Paris, France; 4Universities, Bobigny, France; and 5Oncology and 6Public Health Departments, Avicenne Hospital, Bobigny, France.
ABSTRACT Background: Alcohol intake is associated with increased circulating concentrations of sex hormones, which in turn may increase hormonedependent cancer risk. This association may be modulated by dietary fiber intake, which has been shown to decrease steroid hormone bioavailability (decreased blood concentration and increased sex hormone–binding globulin concentration). However, this potential modulation has not been investigated in any prospective cohort. Objectives: Our objectives were to study the relation between alcohol intake and the risk of hormone-dependent cancers (breast, prostate, ovarian, endometrial, and testicular) and to investigate whether dietary fiber intake modulated these associations. Design: This prospective observational analysis included 3771 women and 2771 men who participated in the Supplémentation en Vitamines et Minéraux Antioxydants study (1994–2007) and completed at least 6 valid 24-h dietary records during the first 2 y of follow-up. After a median follow-up of 12.1 y, 297 incident hormone-dependent cancer cases, including 158 breast and 123 prostate cancers, were diagnosed. Associations were tested via multivariate Cox proportional hazards models. Results: Overall, alcohol intake was directly associated with the risk of hormone-dependent cancers (tertile 3 vs. tertile 1: HR: 1.36; 95% CI: 1.00, 1.84; P-trend = 0.02) and breast cancer (HR: 1.70; 95% CI: 1.11, 2.61; P-trend = 0.04) but not prostate cancer (P-trend = 0.3). In stratified analyses (by sex-specific median of dietary fiber intake), alcohol intake was directly associated with hormone-dependent cancer (tertile 3 vs. tertile 1: HR: 1.76; 95% CI: 1.10, 2.82; P-trend = 0.002), breast cancer (HR: 2.53; 95% CI: 1.30, 4.95; P-trend = 0.02), and prostate cancer (HR: 1.37; 95% CI: 0.65, 2.89; P-trend = 0.02) risk among individuals with low dietary fiber intake but not among their counterparts with higher dietary fiber intake (P-trend = 0.9, 0.8, and 0.6, respectively). The P-interaction between alcohol and dietary fiber intake was statistically significant for prostate cancer (P = 0.01) but not for overall hormone-dependent (P = 0.2) or breast (P = 0.9) cancer. Conclusion: In line with mechanistic hypotheses and experimental data, this prospective study suggested that dietary fiber intake might modulate the association between alcohol intake and risk of hormonedependent cancer. This trial was registered at clinicaltrials.gov as NCT00272428. Am J Clin Nutr doi: 10.3945/ajcn.114.098418. Keywords alcohol, breast cancer, dietary fiber, hormone-dependent cancer, prospective study, prostate cancer
INTRODUCTION
Hormone-dependent cancers include cancer of the prostate and testis in men and breast, ovarian, and endometrial cancer in women. The sex steroid hormones, androgen, estrogen, and progestin, are the primary modulators of both normal development and maintenance of these organs, as well as malignant growths. Increased concentrations of circulating estrogens and androgens have been related to a greater risk of hormonedependent cancers (1–5). Among other potentially procarcinogenic mechanisms, alcohol intake has been shown to increase circulating concentrations of steroid hormones and act on their receptors (6–13). The positive association between alcohol intake and breast cancer risk is now well established (13–15). In contrast, this association is less clear for other hormone-dependent cancers (16–21) and deserves further prospective investigation. Indeed, the level of proof regarding the association between prostate (18), ovarian (20), and endometrial (19) cancer and alcohol intake was qualified by the World Cancer Research Fund and the American Institute for Cancer Research as “limited—no conclusion.” The simultaneous intake of nutritional bioactive compounds may modulate the association between alcohol and cancer risk. Dietary fibers represent a complex group of molecules that includes nonstarch polysaccharides (cellulose, hemicelluloses, pectins, chitins, and hydrocolloids such as gums, mucilages, and glucans), oligosaccharides (inulin, a-galactosides, fructooligosaccharides, and galacto-oligosaccharides), resistant starch, resistant dextrins, and other compounds associated with dietary fiber polysaccharides such as lignin and waxes (22). Dietary fibers are good candidates for this potential modulatory role (9), because epidemiologic and experimental data showed that they may reduce concentrations of circulating estrogens and androgens through various mechanisms (9, 23–28). This is one of the 1 Supported by a grant from the French Ministry of Research and Higher Education (A-SC) and a PhD grant from the Cancéropoˆle Ile-de-France (public funding; MD). 2 Address correspondence to M Touvier, Nutritional Epidemiology Research Team (EREN), SMBH Paris 13, 74 rue Marcel Cachin, F-93017, Bobigny, France. E-mail: [email protected]. Received August 25, 2014. Accepted for publication April 29, 2015. doi: 10.3945/ajcn.114.098418.
Am J Clin Nutr doi: 10.3945/ajcn.114.098418. Printed in USA. Ó 2015 American Society for Nutrition
Copyright (C) 2015 by the American Society for Nutrition
1 of 8
2 of 8
CHHIM ET AL.
mechanistic hypotheses that support the protective effect of dietary fiber intake on breast and prostate cancer risk observed in several cohorts (29–32). To our knowledge, no prospective study on alcohol consumption and hormone-dependent, breast, or prostate cancer risk has investigated a potential modulation of these relations by dietary fiber intake. Thus, our objectives were to prospectively investigate the association between alcohol intake and hormone-dependent cancer risk (overall, breast, and prostate cancers) and to explore its potential modulation by dietary fiber intake. METHODS
Study population The Supplémentation en Vitamines et Minéraux Antioxydants (SU.VI.MAX) study was a population-based, double-blind, placebo-controlled, randomized trial (registered at clinicaltrials.gov as NCT00272428) initially designed to assess the effect of a daily antioxidant supplementation on the incidence of cardiovascular disease and cancer (33). A total of 13,017 subjects were enrolled in 1994–1995. All participants took a single daily capsule containing a combination of 120 mg ascorbic acid, 30 mg vitamin E, 6 mg b-carotene, 100 mg selenium, and 20 mg zinc or a placebo. The intervention study lasted 8 y, and observational follow-up of health events was maintained until September 2007. Subjects provided written informed consent, and the study was approved by the Ethics Committee for Studies with Human Subjects at the Paris-Cochin Hospital (no. 706) and the Commission Nationale de l’Informatique et des Libertés (no. 334641). Dietary data collection During follow-up, participants were invited to complete a dietary record every 2 mo, in which they reported all foods and beverages consumed during a period of 24 h. These dietary records were randomly distributed between week and weekend days and over all seasons to take into account intraindividual variability. Dietary records from the first 2 y of follow-up were used in the present study to comply with its prospective design. Food record completion was done through the Minitel Telematic Network, a French telephone-based terminal that was an Internet prototype. Portion sizes were assessed via a validated picture booklet (34), and the amounts consumed from composite dishes were estimated by using French recipes validated by food and nutrition professionals. The mean daily energy, alcohol, dietary fiber, and other nutrient intakes were estimated by using a published French food composition table (35). Total dietary fiber content was obtained by using the Association of Official Analytic Chemists method (36). Dietary fiber intakes in the SU.VI. MAX study have been described previously (30, 31, 37). Data collection for covariates At enrollment, self-administered questionnaires were filled by participants, reporting sociodemographic characteristics (sex, date of birth, and educational level); smoking status; medication use (including baseline use of hormonal treatment of menopause); number of live births; family history of overall, breast, and prostate cancer; and menopausal status. Baseline physical
activity was self-reported by asking the participants if they currently practiced a physical activity on a regular basis and, if yes, if it was equivalent to 1 h of walking/d or more/less. Anthropometric measurements were taken during a baseline clinical examination. Weight was measured to the nearest 0.1 kg with an electronic scale (Seca) by study nurses and physicians. Subjects wore light clothing and no shoes. Height was measured to the nearest 1 cm with a wall-mounted stadiometer under the same conditions. A 35-mL fasting venous blood sample was collected in evacuated tubes. These plasma samples were used to determine the baseline concentrations of selenium and total prostate-specific antigen measured by immunoassay (Roche Diagnostics). Case ascertainment Health events occurring during the follow-up were selfreported by participants. Medical data were then requested from participants, physicians, and/or hospitals and reviewed by an independent physician expert committee. Pathologic reports were used to validate the cases and to extract cancer type characteristics. Cases were classified by using the International Chronic Disease Classification, 10th Revision, Clinical Modification (38). All incident first primary hormone-dependent cancers were considered as cases in this study: breast (in women), prostate, ovarian, endometrial, and testicular cancer. Statistical analyses From the 13,017 participants in the SU.VI.MAX study, 161 were excluded because they had reported a cancer diagnosis before the start of the follow-up. Among the remaining participants, 6542 provided at least 6 valid 24-h dietary records within the first 2 y of follow-up and thus were available for analysis. For all covariates, ,5% of values were missing and were replaced by the respective mode value. Baseline characteristics of participants were compared between cases and noncases by using x2 tests or Student’s t tests, as appropriate. HRs and 95% CIs obtained from Cox proportional hazards models with age as the primary time variable were used to characterize the association between tertiles of alcohol intake and risk of incident hormone-dependent (all sites), breast, and prostate cancer. Participants contributed person-time until the date of diagnosis of the studied cancer type (overall hormonedependent, breast, or prostate), the date of the last completed questionnaire, the date of death, or September 2007, whichever occurred first. Participants who reported another cancer than the one studied during the study period were included and censored at the date of that diagnosis (except for basal cell carcinoma, which was not considered cancer). We confirmed that the assumptions of proportionality were satisfied through examination of the loglog (survival) vs. log-time plots. Tests for linear trend were performed by using the ordinal score of tertiles of alcohol intake. Multivariate models were adjusted for age (time scale in the Cox model), sex (for overall hormone-dependent cancers only), active intervention group in the SU.VI.MAX trial (yes/no), number of dietary records (continuous), height (continuous), BMI (in kg/m2; ,25, $25 to ,30, or $30), physical activity (irregular or ,1 or $1 h/d walking or equivalent), smoking status (never, former, or current), educational level (primary, secondary, or university),
ALCOHOL, FIBER, AND HORMONE-DEPENDENT CANCER RISK
mean daily dietary fiber intake (continuous), mean daily energy intake without alcohol (continuous), and first-degree family history of cancer (overall, breast, or prostate) (yes or no). For breast cancer analyses, additional adjustments were performed for baseline menopausal status (yes or no), use of hormonal treatment of menopause (yes or no), and number of live births (continuous). For prostate cancer analyses, additional adjustment was performed for baseline prostate-specific antigen concentration (continuous). Interaction between dietary fiber intake (, vs. $ sex-specific population median) and alcohol intake (tertiles) was tested by introducing an interaction term into the models (product of the 2 variables). Analyses were conducted overall and then stratified by dietary fiber intake. Models were also computed separately for premenopausal and postmenopausal breast cancer (women contributed to the premenopausal model until their age of menopause, and conversely, women contributed to the postmenopausal model from their age of menopause). Sensitivity analyses were performed by excluding heavy drinkers (.100 g alcohol/d) and cancers diagnosed during the first 2 y of follow-up. Further adjustment for a “healthy” food pattern was also tested. This dietary pattern was extracted by principal component analysis, by using the SAS “proc factor” procedure, from the mean intakes of 31 food groups across all 24-h records collected during the first 2 y of the study. Additional adjustments were also tested for dietary intake of total fat for women and for dietary intake of calcium, processed meat, tomato-based products, and vitamin E and plasma selenium concentration for men (continuous). All tests were 2-sided, and P , 0.05 was considered statistically significant. SAS version 9.3 (SAS Institute) was used for analyses. RESULTS
Mean follow-up time was 12.1 y. Of the subjects, 5.2% were lost to follow-up. A total of 297 incident hormone-dependent cancer cases were diagnosed during the follow-up: 158 breast (40 premenopausal and 118 postmenopausal), 123 prostate, and 16 other hormone-dependent cancers (11 ovarian and 5 endometrial). No case of testicular cancer was diagnosed. Mean age at diagnosis was 56.2 y for breast cancer and 63.1 for prostate cancer. Among breast cancer cases, 58.9% were estrogen receptor positive, and 43.0% were progesterone receptor positive. Regarding histologic type of breast cancer, 62.0% were ductal, 14.6% were lobular, and 23.4% were other types. Among prostate cancer cases, 5.1% had a Gleason score between 2 and 4, 87.2% between 5 and 7, and 7.7% between 8 and 10. The exposures of interest were as follows: 81% of alcohol intake came from wine (red, white, or rosé), 9% from beer and cider, and 10% from spirits (data not shown). Sources of dietary fibers were cereals (38.0%), vegetables (22.2%), fruit (21.2%), and legumes (6.1%). Of the dietary fibers, 20.1% were soluble and 79.8% were insoluble. Characteristics of the subjects are presented in Table 1. Compared with noncases, subjects with hormone-dependent cancer tended to be older, were more frequently obese, had more family history of cancer, were more frequently menopausal at baseline, and took more hormonal treatment of menopause (for women).
3 of 8
Associations between tertiles of alcohol intake and overall hormone-dependent, breast, and prostate cancer risks are presented in Table 2. Tertiles of alcohol intake were directly associated with hormone-dependent (tertile 3 vs. tertile 1: HR: 1.36; 95% CI: 1.00, 1.84; P-trend = 0.02) and breast (HR: 1.70; 95% CI: 1.11, 2.61; P-trend = 0.04) cancer risks, whereas no statistically significant association was observed for prostate cancer (P-trend = 0.3). Results for the associations between tertiles of alcohol intake and overall hormone-dependent, breast, and prostate cancer risks stratified by dietary fiber intake are shown in Table 3. Although the P-interaction between alcohol and dietary fiber intake was statistically significant for prostate cancer (P = 0.01) but not for overall hormone-dependent (P = 0.2) or breast (P = 0.9) cancer, differential results were observed across dietary fiber strata. Indeed, alcohol intake was directly associated with hormonedependent (P-trend = 0.002), breast (P-trend = 0.02), and prostate (P-trend = 0.02) cancer risks among individuals with lower dietary fiber intake but not among those with higher dietary fiber consumption (P-trend = 0.9, 0.8, and 0.6, respectively), despite similar statistical power across fiber strata. Similar results were observed when the stratification was performed by the median of soluble and insoluble fiber intake (data not tabulated): P-trend = 0.008 for hormone-dependent cancer, 0.04 for breast cancer, and 0.007 for prostate cancer in subjects with low soluble fiber intake, whereas the corresponding P values were 0.5, 0.3, and 0.4 in subjects with higher soluble fiber intake. Similarly, P-trend = 0.01 for hormonedependent cancer, 0.02 for breast cancer, and 0.01 for prostate cancer in subjects with low insoluble fiber intake, whereas the corresponding P values were 0.5, 0.5, and 0.6 in subjects with higher insoluble fiber intake. Also, similar results were observed regarding breast cancer risk when analyses were restricted to postmenopausal women: alcohol intake was directly associated with breast cancer risk among subjects with lower dietary fiber intake (tertile 3 vs. tertile 1: HR: 2.39; 95% CI: 1.10, 5.18; P-trend = 0.06), whereas no association was observed among those with higher dietary fiber intake (P-trend = 0.3). The statistical power was insufficient to perform such an analysis in premenopausal women. Further adjustments for a healthy dietary pattern (all models), for fat intake (breast cancer models), or for dietary intake of calcium, processed meat, tomato products, vitamin E, and plasma selenium concentration (prostate cancer models) provided similar results (data not shown). Sensitivity analyses excluding cancer cases diagnosed within the first 2 y of follow-up (26 hormonedependent, 21 breast, and 5 prostate cancer cases) did not modify the findings, nor did sensitivity analyses excluding subjects who consumed more than 100 g/d alcohol (heavy drinkers, n = 39). DISCUSSION
This prospective study confirmed the direct association between alcohol intake and breast cancer risk. It also suggested that dietary fiber intake may modulate the association between alcohol intake and hormone-dependent cancers, which, to our knowledge, had never been investigated in any epidemiologic study, despite its mechanistic plausibility. Indeed, although the P-interaction was statistically significant only for prostate cancer, alcohol intake was directly associated with overall
4 of 8
CHHIM ET AL.
TABLE 1 Baseline characteristics of the study population (n = 6542), Supplémentation en Vitamines et Minéraux Antioxydants cohort, France, 1994–2007
Age, y Sex, n (%) Male Female Educational level, n (%) Primary Secondary University Smoking status, n (%) Nonsmoker Former smoker Current smoker Physical activity, n (%) Irregular/none ,1 h/d of walking or equivalent .1 h/d of walking or equivalent Height, cm BMI,3 n (%) Underweight Normal weight Overweight Obese Menopausal status,4 n (%) Nonmenopausal Menopausal Use of hormonal treatment of menopause (yes),4 n (%) Number of live births4 Plasma prostate-specific antigen concentration,6 mg/L First-degree family history of cancer, n (%) All sites (yes) Breast cancer (yes)4 Prostate cancer (yes)6 Intervention group, n (%) Antioxidants Placebo Number of 24-h dietary records Energy intake without alcohol, kcal/d Alcohol intake, g/d Total dietary fiber intake, g/d Dietary fiber intake from cereals, g/d Dietary fiber intake from fruit, g/d Dietary fiber intake from vegetables, g/d Dietary fiber intake from legumes, g/d Dietary insoluble fiber intake, g/d Dietary soluble fiber intake, g/d
Noncases (n = 6245)
Hormone-dependent cancer cases (n = 297)
Breast cancer cases (n = 158)
Prostate cancer cases (n = 123)
49.1 6 6.42
52.0 6 6.3
49.7 6 6.4
54.9 6 4.7
2648 (42.4) 3597 (57.6%)
123 (41.4) 174 (58.6)
158 (100.0)
1274 (20.4) 2381 (38.1) 2590 (41.5)
61 (20.5) 104 (35.0) 132 (44.4)
33 (20.9) 53 (33.5) 72 (45.6)
24 (19.5) 47 (38.2) 52 (42.3)
3078 (49.3) 2345 (37.6) 822 (13.2)
147 (49.5) 109 (36.7) 41 (13.8)
87 (55.1) 42 (26.6) 29 (18.4)
50 (40.7) 61 (49.6) 12 (9.8)
80 (26.9) 91 (30.6) 126 (42.4) 166.9 6 8.3
50 (31.7) 53 (33.5) 55 (34.8) 162.6 6 6.2
29 (23.6) 31 (25.2) 63 (51.2) 173.2 6 6.3
P value1 ,0.0001 0.7
123 (100.0) 0.5
0.9
0.5 1519 1879 2847 166.7
(24.3) (30.1) (45.6) 6 8.4
144 3968 1764 369
(2.3) (63.5) (28.3) (5.9)
2517 1080 1083 1.9 1.2
(70.0) (30.0) (30.1) 6 1.2 6 1.3
97 (55.8) 77 (44.3) 70 (40.2) 2.1 6 1.2 3.4 6 3.7
92 (58.2) 66 (41.8) 61 (38.6) 2.0 6 1.2
2182 (34.9) 302 (8.4) 124 (4.7)
129 (43.4) 26 (14.9) 16 (13.0)
69 (43.7) 24 (15.2)
3113 (49.9) 3132 (50.2) 10.6 6 2.4 1983 6 539 18.8 6 20.7 19.0 6 6.5 7.2 6 3.3 4.0 6 2.6 4.2 6 1.9 1.2 6 1.4 15.2 6 5.4 3.8 6 1.3
140 (47.1) 157 (52.9) 10.6 6 2.4 1977 6 533 19.7 6 19.0 19.1 6 6.8 7.3 6 3.2 4.1 6 2.5 4.2 6 2.0 1.1 6 1.2 15.3 6 5.6 3.9 6 1.3
73 (46.2) 85 (53.8) 10.4 6 2.4 1778 6 414 12.9 6 12.1 17.6 6 5.2 6.6 6 2.4 3.8 6 2.1 4.0 6 1.9 1.1 6 1.3 14.0 6 4.3 3.6 6 1.0
9 189 71 28
(3.0) (63.6) (23.9) (9.4)
7 116 20 15
(4.4) (73.4) (12.7) (9.5)
1 64 45 13
0.7 0.04
(0.8) (52.0) (36.6) (10.6) ,0.00015
3.4 6 3.7 51 (41.5) 16 (13.0)
0.0055 0.075 ,0.00017 0.003 0.0035 ,0.00017 0.4
61 (49.6) 62 (50.4) 11.0 6 2.3 2276 6 544 30.2 6 22.1 21.3 6 8.1 8.4 6 3.9 4.4 6 3.1 4.6 6 2.1 1.1 6 1.2 17.1 6 6.7 4.2 6 1.5
0.9 0.8 0.4 0.7 0.5 0.6 0.8 0.6 0.8 0.6
P value for the comparison between overall hormone-dependent cancer cases and noncases, by Student’s t test or x2 test where appropriate. Mean 6 SD (all such values). 3 Cutoffs for BMI (in kg/m2) were ,18.5 for underweight, 18.5–25 for normal weight, 25–30 for overweight, and $30 for obesity. 4 In women only. 5 P value for the comparison between breast cancer cases and female noncases. 6 In men only. 7 P value for the comparison between prostate cancer cases and male noncases. 1 2
hormone-dependent, breast, and prostate cancer in subjects with low dietary fiber intake but not in subjects with higher dietary fiber intake. The direct association between alcohol intake and increased breast cancer risk is now well established (13, 15, 39). Regarding prostate cancer, a nonlinear dose-response meta-analysis of 22 cohort and 50 case-control studies showed a statistically sig-
nificant increased risk of alcohol intake of 25 and 50 g/d but not of 100 g/d (17). In the recent World Cancer Research Fund/ American Institute for Cancer Research meta-analysis, which included all available prospective studies (21), the summary HR for prostate cancer risk was not significant for total alcohol intake, but a weak, significantly increased risk was observed for wine and spirits. There was statistical evidence of nonlinearity in
5 of 8
ALCOHOL, FIBER, AND HORMONE-DEPENDENT CANCER RISK TABLE 2 Associations between tertiles of alcohol intake and overall hormone-dependent, breast, and prostate cancer risk, SU.VI. MAX cohort, France, 1994–20071 Cases/noncases, n
HR (95% CI)
P-trend
75/2105 112/2069 110/2071
1 1.52 (1.13, 2.04) 1.36 (1.00, 1.84)
0.02
36/1221 55/1202 67/1190
1 1.55 (1.01, 2.38) 1.70 (1.11, 2.61)
0.04
35/888 47/877 41/883
1 1.28 (0.82, 2.00) 0.89 (0.55, 1.45)
0.3
2
Hormone-dependent cancers (n = 297) T1 T2 T3 Breast cancer (n = 158)3 T1 T2 T3 Prostate cancer (n = 123)4 T1 T2 T3
1 From multivariate Cox proportional hazards models. Cutoffs for tertiles of alcohol intake were 3.0/12.1 g/d for women and 15.3/35.4 g/d for men. SU.VI.MAX, Supplémentation en Vitamines et Minéraux Antioxydants; T1, tertile 1; T2, tertile 2; T3, tertile 3. 2 Ajusted for age, sex, active intervention group during SU.VI.MAX trial, number of dietary records, smoking status, educational level, physical activity, height, BMI, family history of cancer, energy intake without alcohol, and dietary fiber intake. 3 Same adjustment + menopausal status at baseline, use of hormonal treatment of menopause at baseline, number of live births, and family history of breast (instead of overall) cancer. 4 Same adjustment + baseline prostate specific antigen concentration and family history of prostate (instead of overall) cancer.
several of these relations, with a small decrease in advanced prostate cancer risk with higher intakes (.7 drinks/d for total alcohol and .3 drinks/d for spirits). In our study, most alcohol intake came from wine; thus, it was not possible to model the effects of the different types of alcohol. We observed a direct association between total alcohol intake and prostate cancer risk in the low dietary fiber strata: this association was statistically significant for moderate drinkers (second tertile) but not for higher drinkers (third tertile). This phenomenon, in line with previous nonlinear data, deserves further investigation. One explanation could pertain to diagnosis bias. Indeed, prostate cancer is often asymptomatic and discovered with symptoms only at advanced stages. Higher drinkers may generally be less careful about their health and may have been less inclined to consult a physician. Another explanation could be related to misclassification of sicker men who had stopped drinking at baseline but drank before. Epidemiologic evidence is still insufficient regarding alcohol intake and the risk of other hormone-dependent cancers (16, 18–20). Statistical power was too limited to perform specific analyses for cancer sites other than breast and prostate in this study. The most original finding of our study was that the main associations tended to be modulated by dietary fiber intake in stratified analyses. Caution is needed when interpreting these results because the interaction was significant for prostate cancer only. This may be a result of the low statistical power of the interaction test (40) and thus requires confirmation by future prospective studies. No previous epidemiologic study has directly investigated this potential modulation. However, one prospective study specifically evaluated the relation of alcohol and dietary fiber intake with circulating sex hormone concentrations among premenopausal women and observed opposing associations: direct for alcohol and inverse for dietary fiber intake (9). Conversely, few prospective studies have investigated the relation between dietary fiber intake and cancer risk stratified by alcohol
consumption (31, 32, 41). In 2 prospective studies on breast cancer risk, no modulation by alcohol intake was detected (32, 41). In a previous investigation within the SU.VI.MAX cohort, dietary fiber intake was prospectively associated with decreased prostate cancer risk in men with higher alcohol intake but not in men with alcohol intake below the population median (31). The potential modulation of the relation between alcohol and hormone-dependent cancer risk by dietary fiber intake is in line with our initial hypothesis and with mechanistic and experimental data. Dietary fibers are metabolized by gut microbiota and may influence hormone-sensitive cancer risk through decreased hormonal activity (42, 43). Several mechanisms may be involved in this phenomenon. First, dietary fibers are thought to modify the enterohepatic circulation of estrogens by decreasing the activity of the colonic b-D-glucuronidase, an enzyme allowing estrogens to reenter the circulation (25, 27). Thus, fewer estrogens reenter the circulation and more estrogens are excreted (27). Second, dietary fiber intake has been associated with increased sex hormone–binding globulin concentrations (26) and thus decreased bioavailability of estrogens (44), which probably results from the beneficial influence of dietary fibers on insulin resistance (45) [insulin resistance being associated with a decrease in sex hormone–binding globulin concentrations (44)]. Our results suggest that although alcohol may exert its carcinogenic effect via several plausible mechanisms [including a genotoxic effect of acetaldehyde, a role as a solvent for tobacco carcinogens, and via the production of reactive oxygen species and nitrogen species (6)], its capacity to increase circulating steroid hormone concentrations (6–13) probably plays a central role in hormone-dependent cancer etiology (1–5). Strengths of our study pertained to its prospective design with long follow-up, its originality with respect to existing literature, and precise assessment of dietary intake based on at least six 24-h dietary records. However, several limitations should be acknowledged. First,
6 of 8
CHHIM ET AL. TABLE 3 Associations between tertiles of alcohol intake and overall hormone-dependent, breast and prostate cancer risk, stratified by median of dietary fiber intake, SU.VI.MAX cohort, France, 1994–20071 Cases/noncases, n Hormone-dependent cancers (n = 297)3 Dietary fiber intake , median T1 T2 T3 Dietary fiber intake $ median T1 T2 T3 Breast cancer (n = 158)4 Dietary fiber intake , median T1 T2 T3 Dietary fiber intake $ median T1 T2 T3 Prostate cancer (n = 123)5 Dietary fiber intake , median T1 T2 T3 Dietary fiber intake $ median T1 T2 T3
HR (95% CI)
P-trend
P-interaction2 0.2
28/1020 58/984 59/1121
1 2.27 (1.43, 3.59) 1.76 (1.10, 2.82)
0.002
47/1085 54/1085 51/950
1 1.09 (0.74, 1.63) 1.12 (0.74, 1.70)
0.9
0.9 13/634 24/550 35/629
1 2.25 (1.13, 4.47) 2.53 (1.30, 4.95)
0.02
23/587 31/652 32/561
1 1.13 (0.65, 1.95) 1.24 (0.70, 2.19)
0.8
0.01 13/388 30/438 23/493
1 2.45 (1.23, 4.87) 1.37 (0.65, 2.89)
0.02
22/500 17/439 18/390
1 0.71 (0.37, 1.37) 0.76 (0.38, 1.51)
0.6
1
From multivariate Cox proportional hazards models. Cutoffs for tertiles of alcohol intake were 3.0 and 12.1 g/d for women and 15.3 and 35.4 g/d for men. Median of mean daily dietary fiber intake was 16.8 g/d for women and 20.5 g/d for men. SU.VI.MAX, Supplémentation en Vitamines et Minéraux Antioxydants; T1, tertile 1; T2, tertile 2; T3, tertile 3. 2 Between alcohol and dietary fiber intakes. 3 Adjusted for age, sex, active intervention group during SU.VI.MAX trial, number of dietary records, smoking status, educational level, physical activity, height, BMI, family history of cancer, energy intake without alcohol, and dietary fiber intake. 4 Same adjustment + menopausal status at baseline, use of hormonal treatment of menopause at baseline, number of live births, and family history of breast (instead of overall) cancer. 5 Same adjustment + baseline prostate specific antigen concentration and family history of prostate (instead of overall) cancer.
the relatively small number of cases may have limited our ability to detect some of the hypothesized associations. The number of cancer cases was not sufficient to perform specific analyses for each breast cancer type (receptor status and histologic type) or according to the Gleason score for prostate cancer. Second, the observed relations could be affected by unmeasured or residual confounding. Some behavioral/environmental factors could not be taken into account in this study because of a lack of data (e.g., exposure to ionizing or ultraviolet radiation). However, most pertinent behavioral risk factors (i.e., other dietary factors, weight status, smoking habits, and physical activity) were taken into account in our study. Thus, it is unlikely that residual confounding entirely explains the observed associations. Next, dietary fiber refers to a complex group of molecules, and all types of fibers may not have the same properties with regard to modulation of cancer risk. For instance, it has been suggested that oligosaccharides and resistant starch are involved in antiinflammatory pathways through interaction with gut microflora, resulting in the production of short-chain fatty acids and reinforcement of the intestinal barrier (46–48). However, the reference
method used to quantify total dietary fiber in our study did not allow the measurement of oligosaccharides and provided only a very limited measure of dietary resistant starch (no measure of native and physically inaccessible resistant starch) (36). We can hypothesize that the modulatory effect of dietary fiber on the alcohol-cancer relations suggested by our findings would be even stronger if all types of dietary fiber were better taken into account by more precise analytic methods. Besides, nutrient composition data did not allow us to distinguish the different types of fiber in the analyses (nonstarch polysaccharides, oligosaccharides, resistant starch, lignin, etc.). We observed similar results for soluble and insoluble fibers. However, beyond this aspect, information on specific types of fiber and/or their fermentation properties would have been more informative but was not available. Finally, subjects were volunteers involved in a longterm nutrition study and thus had more health-conscious behaviors than the general French population, such as lower tobacco use [13% vs. 34% current smokers (49)]. However, their dietary fiber intake (mean: 19 g/d) was similar to the one observed in a French nationally representative survey (19.0 g/d in adults aged 55–79 y) (50).
ALCOHOL, FIBER, AND HORMONE-DEPENDENT CANCER RISK
In conclusion, this prospective study suggested that dietary fiber intake may modulate the associations between alcohol intake and hormone-dependent, breast, and prostate cancer risks, with a statistically significant interaction for prostate cancer. Mechanistic data support this epidemiologic result: dietary fibers may contribute to counteract the procarcinogenic increase in circulating sex hormones induced by alcohol intake. These results provide useful insights to better understand the etiologic involvement of alcohol in carcinogenesis and may contribute to explain the contrasted findings of some prior prospective studies investigating alcohol intake and cancer risk. From a public health standpoint, these findings also suggest that several risk factors (such as alcohol consumption and low dietary fiber intake) may cumulate and act in synergy to increase cancer risk. Further experimental data and large prospective cohorts are needed to confirm these results. The authors thank Dr. Valentina A. Andreeva for English language editing and Younes Esseddik, Paul Flanzy, Mohand Ait Oufella, Yasmina Chelghoum, and Than Duong Van (computer scientists); Nathalie Arnault, Véronique Gourlet, Dr. Fabien Szabo, Laurent Bourhis, and Stephen Besseau (statisticians); and Rachida Mehroug (logistics assistant) for their technical contribution to the SU.VI.MAX study. The author responsibilities were as follows—A-SC, MD, and MT: designed the research; AS-C, SH, PG, and MT: conducted the research; A-SC: performed statistical analysis; A-SC and MT: wrote the manuscript; A-SC, PF, PL-M, ND-P, LZ, LD, SH, PG, MD, and MT: contributed to the data interpretation and revised each draft for important intellectual content; MT: had primary responsibility for the final content; and all authors: read and approved the final manuscript. The authors reported no conflicts of interest related to this study. The funders had no role in the design, implementation, analysis, or interpretation of the data.
REFERENCES 1. Bosland MC. The role of steroid hormones in prostate carcinogenesis. J Natl Cancer Inst Monogr 2000;27:39–66. 2. Danforth KN, Eliassen AH, Tworoger SS, Missmer SA, Barbieri RL, Rosner BA, Colditz GA, Hankinson SE. The association of plasma androgen levels with breast, ovarian and endometrial cancer risk factors among postmenopausal women. Int J Cancer 2010;126:199–207. 3. Key T, Appleby P, Barnes I, Reeves G. Endogenous sex hormones and breast cancer in postmenopausal women: reanalysis of nine prospective studies. J Natl Cancer Inst 2002;94:606–16. 4. Key TJ. Endogenous oestrogens and breast cancer risk in premenopausal and postmenopausal women. Steroids 2011;76:812–5. 5. Walker K, Bratton DJ, Frost C. Premenopausal endogenous oestrogen levels and breast cancer risk: a meta-analysis. Br J Cancer 2011;105:1451–7. 6. Boffetta P, Hashibe M. Alcohol and cancer. Lancet Oncol 2006;7:149–56. 7. García-Closas M, Herbstman J, Schiffman M, Glass A, Dorgan JF. Relationship between serum hormone concentrations, reproductive history, alcohol consumption and genetic polymorphisms in premenopausal women. Int J Cancer 2002;102:172–8. 8. International Agency for Research on Cancer. Consumption of alcoholic beverages. IARC Monogr Eval Carcinog Risks Hum 2012;100: 373–499. 9. Maskarinec G, Morimoto Y, Takata Y, Murphy SP, Stanczyk FZ. Alcohol and dietary fibre intakes affect circulating sex hormones among premenopausal women. Public Health Nutr 2006;9:875–81. 10. Onland-Moret NC, Peeters PH, van der Schouw YT, Grobbee DE, van Gils CH. Alcohol and endogenous sex steroid levels in postmenopausal women: a cross-sectional study. J Clin Endocrinol Metab 2005;90: 1414–9. 11. Rinaldi S, Peeters PH, Bezemer ID, Dossus L, Biessy C, Sacerdote C, Berrino F, Panico S, Palli D, Tumino R, et al. Relationship of alcohol intake and sex steroid concentrations in blood in pre- and postmenopausal women: the European Prospective Investigation into Cancer and Nutrition. Cancer Causes Control 2006;17:1033–43.
7 of 8
12. Seitz HK, Stickel F. Molecular mechanisms of alcohol-mediated carcinogenesis. Nat Rev Cancer 2007;7:599–612. 13. Seitz HK, Pelucchi C, Bagnardi V, La Vecchia C. Epidemiology and pathophysiology of alcohol and breast cancer: update 2012. Alcohol Alcohol 2012;47:204–12. 14. Baan R, Straif K, Grosse Y, Secretan B, El GF, Bouvard V, Altieri A, Cogliano V. Carcinogenicity of alcoholic beverages. Lancet Oncol 2007;8:292–3. 15. World Cancer Research Fund/American Institute for Cancer Research. Systematic literature review—continuous update project report: the associations between food, nutrition and physical activity and the risk of breast cancer. Washington (DC): American Institute for Cancer Research; 2010. 16. Kelemen LE, Bandera EV, Terry KL, Rossing MA, Brinton LA, Doherty JA, Ness RB, Kjaer SK, Chang-Claude J, Kobel M, et al. Recent alcohol consumption and risk of incident ovarian carcinoma: a pooled analysis of 5,342 cases and 10,358 controls from the Ovarian Cancer Association Consortium. BMC Cancer 2013;13:28. 17. Rota M, Scotti L, Turati F, Tramacere I, Islami F, Bellocco R, Negri E, Corrao G, Boffetta P, La VC, et al. Alcohol consumption and prostate cancer risk: a meta-analysis of the dose-risk relation. Eur J Cancer Prev 2012;21:350–9. 18. 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. 19. World Cancer Research Fund/American Institute for Cancer Research. Systematic literature review— continuous update project report: the associations between food, nutrition and physical activity and the risk of endometrial cancer. Washington (DC): American Institute for Cancer Research; 2013. 20. World Cancer Research Fund/American Institute for Cancer Research. Systematic literature review—continuous update project report: the associations between food, nutrition and physical activity and the risk of ovarian cancer. Washington (DC): American Institute for Cancer Research; 2014. 21. World Cancer Research Fund/American Institute for Cancer Research. Systematic literature review—continuous update project report: the associations between food, nutrition and physical activity and the risk of prostate cancer. Washington (DC): American Institute for Cancer Research; 2014. 22. European Food Safety Authority. Scientific opinion on dietary reference values for carbohydrates and dietary fibre. EFSA Panel on Dietetic Products, Nutrition, and Allergies. EFSA J 2010;8:1462. 23. Cohen LA, Zhao Z, Zang EA, Wynn TT, Simi B, Rivenson A. Wheat bran and psyllium diets: effects on N-methylnitrosourea-induced mammary tumorigenesis in F344 rats. J Natl Cancer Inst 1996;88:899–907. 24. Dorgan JF, Judd JT, Longcope C, Brown C, Schatzkin A, Clevidence BA, Campbell WS, Nair PP, Franz C, Kahle L, et al. Effects of dietary fat and fiber on plasma and urine androgens and estrogens in men: a controlled feeding study. Am J Clin Nutr 1996;64:850–5. 25. Gerber M. Fiber and breast cancer: another piece of the puzzle—but still an incomplete picture. J Natl Cancer Inst 1996;88:857–8. 26. Longcope C, Feldman HA, McKinlay JB, Araujo AB. Diet and sex hormone–binding globulin. J Clin Endocrinol Metab 2000;85:293–6. 27. Moore MA, Park CB, Tsuda H. Soluble and insoluble fiber influences on cancer development. Crit Rev Oncol Hematol 1998;27:229–42. 28. Ross JK, Pusateri DJ, Shultz TD. Dietary and hormonal evaluation of men at different risks for prostate cancer: fiber intake, excretion, and composition, with in vitro evidence for an association between steroid hormones and specific fiber components. Am J Clin Nutr 1990;51:365–70. 29. Aune D, Chan DS, Greenwood DC, Vieira AR, Rosenblatt DA, Vieira R, Norat T. Dietary fiber and breast cancer risk: a systematic review and meta-analysis of prospective studies. Ann Oncol 2012;23:1394–402. 30. Deschasaux M, Zelek L, Pouchieu C, His M, Hercberg S, Galan P, Latino-Martel P, Touvier M. Prospective association between dietary fiber intake and breast cancer risk. PLoS ONE 2013;8:e79718. 31. Deschasaux M, Pouchieu C, His M, Hercberg S, Latino-Martel P, Touvier M. Dietary total and insoluble fiber intakes are inversely associated with prostate cancer risk. J Nutr 2014;144:504–10. 32. Ferrari P, Rinaldi S, Jenab M, Lukanova A, Olsen A, Tjonneland A, Overvad K, Clavel-Chapelon F, Fagherazzi G, Touillaud M, et al. Dietary fiber intake and risk of hormonal receptor–defined breast cancer in the European Prospective Investigation into Cancer and Nutrition study. Am J Clin Nutr 2013;97:344–53.
8 of 8
CHHIM ET AL.
33. Hercberg S, Galan P, Preziosi P, Bertrais S, Mennen L, Malvy D, Roussel AM, Favier A, Briancon S. The SU.VI.MAX study: a randomized, placebo-controlled trial of the health effects of antioxidant vitamins and minerals. Arch Intern Med 2004;164:2335–42. 34. Le Moullec N, Deheeger M, Preziosi P, Montero P, Valeix P, RollandCachera MF, Potier de Courcy G, Christides JP, Galan P, Hercberg S. Validation du manuel photos utilisé pour l’enqueˆte alimentaire de l’étude SU.VI.MAX (Validation of the food portion size booklet used in the SU.VI.MAX study). Cah Nutr Diet 1996;31:158–64. 35. Hercberg S. Table de Composition SU.VI.MAX des Aliments. Paris (France): Les éditions INSERM/Economica; 2005. 36. Prosky L, Asp NG, Schweizer TF, DeVries JW, Furda I. Determination of insoluble, soluble, and total dietary fiber in foods and food products: interlaboratory study. J Assoc Off Anal Chem 1988;71:1017–23. 37. Lairon D, Bertrais S, Vincent S, Arnault N, Galan P, Boutron MC, Hercberg S. Dietary fibre intake and clinical indices in the French Supplementation en Vitamines et Mineraux AntioXydants (SU.VI. MAX) adult cohort. Proc Nutr Soc 2003;62:11–5. 38. World Health Organization. International classification of diseases and related health problems. 10th revision. Geneva (Switzerland): World Health Organization; 1993. 39. Park SY, Kolonel LN, Lim U, White KK, Henderson BE, Wilkens LR. Alcohol consumption and breast cancer risk among women from five ethnic groups with light to moderate intakes: the Multiethnic Cohort Study. Int J Cancer 2014;134:1504–10. 40. Marshall SW. Power for tests of interaction: effect of raising the type I error rate. Epidemiol Perspect Innov 2007;4:4. 41. Park Y, Brinton LA, Subar AF, Hollenbeck A, Schatzkin A. Dietary fiber intake and risk of breast cancer in postmenopausal women: the National Institutes of Health–AARP Diet and Health Study. Am J Clin Nutr 2009;90:664–71.
42. Aubertin-Leheudre M, Gorbach S, Woods M, Dwyer JT, Goldin B, Adlercreutz H. Fat/fiber intakes and sex hormones in healthy premenopausal women in USA. J Steroid Biochem Mol Biol 2008;112: 32–9. 43. Gaskins AJ, Mumford SL, Zhang C, Wactawski-Wende J, Hovey KM, Whitcomb BW, Howards PP, Perkins NJ, Yeung E, Schisterman EF. Effect of daily fiber intake on reproductive function: the BioCycle Study. Am J Clin Nutr 2009;90:1061–9. 44. Arcidiacono B, Iiritano S, Nocera A, Possidente K, Nevolo MT, Ventura V, Foti D, Chiefari E, Brunetti A. Insulin resistance and cancer risk: an overview of the pathogenetic mechanisms. Exp Diabetes Res 2012;2012:789174. 45. Robertson MD, Bickerton AS, Dennis AL, Vidal H, Frayn KN. Insulinsensitizing effects of dietary resistant starch and effects on skeletal muscle and adipose tissue metabolism. Am J Clin Nutr 2005;82:559– 67. 46. Fung KY, Ooi CC, Zucker MH, Lockett T, Williams DB, Cosgrove LJ, Topping DL. Colorectal carcinogenesis: a cellular response to sustained risk environment. Int J Mol Sci 2013;14:13525–41. 47. Topping DL. Cereal complex carbohydrates and their contribution to human health. J Cereal Sci 2007;46:220–9. 48. Vinolo MA, Rodrigues HG, Nachbar RT, Curi R. Regulation of inflammation by short chain fatty acids. Nutrients 2011;3: 858–76. 49. Castetbon K, Vernay M, Malon A, Salanave B, Deschamps V, Roudier C, Oleko A, Szego E, Hercberg S. Dietary intake, physical activity and nutritional status in adults: the French nutrition and health survey (ENNS, 2006–2007). Br J Nutr 2009;102:733–43. 50. Dubuisson C, Lioret S, Touvier M, Dufour A, Calamassi-Tran G, Volatier JL, Lafay L. Trends in food and nutritional intakes of French adults from 1999 to 2007: results from the INCA surveys. Br J Nutr 2010;103:1035–48.
