| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research Communication |
1 From the Divisions of Cancer Epidemiology and Genetics (AF, NC, JVL, CS, RT, and AS) and of Cancer Control and Population Sciences (AFS and FET), the National Cancer Institute, Bethesda, MD, and the Department of Epidemiology, Michigan State University, East Lansing (EMV).
2 The views expressed are solely those of the authors and do not necessarily reflect the opinions of any state agency.
3 Address reprint requests to A Flood, Division of Cancer Epidemiology and Genetics, the National Cancer Institute, 6120 Executive Boulevard, MSC 7232, Bethesda, MD 20892. E-mail: flooda{at}exchange.nih.gov.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Objective: In a large prospective cohort of women, we examined the association between fruit and vegetable intakes and colorectal cancer.
Design: Between 1987 and 1989, 45490 women with no history of colorectal cancer satisfactorily completed a 62-item Block–National Cancer Institute food-frequency questionnaire. During 386142 person-years of follow-up, 314 women reported incident colorectal cancer, searches of the National Death Index identified an additional 106 colorectal cancers, and a match with state registries identified another 65 colorectal cancers for a total of 485 cases. We used Cox proportional hazards regression analysis to estimate the relative risks (RRs) and 95% CIs in both energy-adjusted and fully adjusted models.
Results: In models using the multivariate nutrient-density model of energy adjustment, RRs for increasing quintile of fruit consumption indicated no significant association with colorectal cancer [RR (95% CI)]: 1.00 (reference), 0.94 (0.70, 1.26), 0.85 (0.63, 1.15), 1.07 (0.81, 1.42), and 1.09 (0.82, 1.44). For vegetable consumption, there was also no significant association in the multivariate nutrient-density model with increasing quintiles of consumption: 1.00 (reference), 0.77 (0.58, 1.02), 0.83 (0.63, 1.10), 0.90 (0.69, 1.19), and 0.92 (0.70, 1.22). Additionally, 3 alternative models of energy adjustment showed no significant association between increases in vegetable intake and the risk of colorectal cancer.
Conclusion: Although the limitations of our study design and data merit consideration, this investigation provides little evidence of an association between fruit and vegetable intakes and colorectal cancer.
Key Words: Fruit • vegetables • colorectal cancer • prospective study • women • energy adjustment • Breast Cancer Detection Demonstration Project
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The most recent comprehensive review of the literature on the association between diet and colorectal cancer concluded after an examination of results from 22 case-control and 4 prospective cohort studies that "convincing" evidence exists that vegetable, but not fruit, consumption decreases risk (5).
Four recent studies, however, have cast some doubt on these conclusions. Results from the Nurse's Health Study and the Health Professionals Follow-up Study—both large, well-designed prospective cohort studies—showed no association between consumption of either fruit or vegetables and colorectal cancer (6). The Polyp Prevention Trial, an intervention study designed to test whether a diet low in fat and high in fruit and vegetables could reduce the recurrence of adenoma, showed no difference in recurrence rates between the intervention and control groups (7). The investigators in the Polyp Prevention Trial provided several possible explanations for this result—insufficient follow-up time, improper timing of the intervention, and no consideration of red meat as a potentially important dietary risk factor. However, these investigators could not rule out the possibility that these dietary factors simply do not influence colorectal neoplasia. Finally, the investigators in the Netherlands Cohort Study on Diet and Cancer (using traditional standards of statistical significance) failed to find any overall associations between fruit or vegetable intake and colon or rectal cancer (8). This study presents evidence from a large prospective cohort of women that allows further evaluation of the relation between fruit and vegetable consumption and the incidence of colorectal cancer.
|
SUBJECTS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Breast Cancer Detection Demonstration Project follow-up cohort
In 1979, the NCI established the BCDDP follow-up study cohort of 64182 participants from a subset of the women enrolled in the original BCDDP screening study. All 4275 women in the screening study with a diagnosis of primary breast cancer, all 25114 women who had undergone a breast biopsy that indicated a benign condition, and all 9628 women who had been recommended to have a biopsy or breast surgery performed but did not have a surgical procedure were included in the follow-up cohort. An additional 25165 women with no history of breast disease were matched to the subjects with breast cancer and to the subjects with benign breast disease for age, time of entry in the BCDDP study, ethnicity, screening center, and duration of participation in the BCDDP. The exclusion of women with a history of breast cancer made no significant difference in the results of our analyses.
The follow-up study proceeded in several phases, beginning with baseline interviews between 1979 and 1981. Of the initial follow-up cohort, 61433 women completed the baseline questionnaire and were therefore eligible for further participation in the study. Participants completed mailed questionnaires during 3 separate follow-up periods: 1987–1989, 1993–1995, and 1995–1998. Each follow-up questionnaire captured information about additional risk factors, updated existing information, and provided self-reports of any cancer diagnoses since the previous phase of the study.
Analytic cohort
We excluded from the study, in sequence, women who did not complete a 1987–1989 follow-up questionnaire (the time of dietary assessment) (n = 9740), women who reported a history of colorectal cancer on the 1987–1989 follow-up questionnaire or earlier (n = 479), women whose reported entry into the cohort occurred on or after their exit date (defined below) (n = 6), and women who did not complete >30 items on their food-frequency questionnaires (FFQs) or who had a reported energy intake >15884 or <1672 kJ/d (>3800 and <400 kcal/d, respectively) (n = 5647). Of the 9740 women who did not complete the 1987–1989 questionnaire, 3066 had died, 505 did not complete the questionnaire because of illness, 1459 refused to complete the questionnaire, and 4710 were either nonresponsive or unable to be contacted. After these exclusions, the analytic cohort contained 45561 women.
We also excluded an additional 71 women who either had missing information on fruit and vegetable intakes or reported an unrealistically high consumption (ie, >16 times/d) on the FFQ. Thus, the final analytic cohort consisted of 45490 individuals.
The maximum follow-up period for each subject extended until the date of completion of the 1995–1998 follow-up questionnaire, the last contact in the 1995–1998 follow-up period if no questionnaire was completed, or the phase 4 anniversary date for those not contacted in the 1995–1998 follow-up period. The phase 4 anniversary date is the estimated date on which subjects would have completed the 1995–1998 questionnaire, with use of mean time intervals from the rest of the cohort, if they had actually completed one.
In the final cohort analyzed, 90.8% (41323) of the women had complete follow-up data through phase 4, meaning that their end-of-study date corresponded to either the date of their first colorectal cancer diagnosis, the date they filled out the phase 4 questionnaire, or their date of death from a cause other than colorectal cancer. The study was approved by the Institutional Review Board of the National Cancer Institute.
Dietary assessment
In the 1987–1989 questionnaire, respondents completed a 62-item Block-NCI FFQ to assess usual dietary intakes over the previous year. Detailed descriptions of this FFQ and its validity have appeared elsewhere (9–11). Software designed for this FFQ yielded estimates of daily intakes of total energy and micronutrients (11).
We expressed intakes of fruit and vegetables in terms of standardized, daily recommended servings based on dietary guidance from the US Department of Health and Human Services, US Department of Agriculture (USDA), as specified in The Food Guide Pyramid (12). A serving of fruit is defined as one medium-sized fresh fruit, 0.5 cup (119 mL) cut fruit, or 6 oz (178 mL) juice. A serving of vegetables is defined as 1 cup (237 mL) leafy vegetables, 0.5 cup (119 mL) other vegetables, or 6 oz (178 mL) juice. We calculated the servings of each fruit or vegetable item listed in the FFQ by converting a medium-sized serving as listed in the Block-NCI FFQ (eg, 0.5 cup, or 119 mL) into the equivalent number of servings as defined by the USDA pyramid. We computed servings for small and large portion sizes on the basis of the instructions for filling out the FFQ (ie, a small-sized serving is one-half as large as a medium-sized serving, and a large-sized serving is 1.5 times the medium-sized serving).
Foods that contributed to the vegetable and fruit food groups appear in Table 1
. For mixtures on the vegetable list, we estimated from USDA recipe information (13) the proportion of the food that was vegetable and applied this value to calculate usual intakes of vegetables from that food for each individual. Intakes of red meat and grains were expressed in terms of the estimated daily frequency of consumption per 1000 kJ. Similarly, standard units of nutrients (eg, mg) per 1000 kJ were used for alcohol, folate, calcium, and vitamin D in all analyses.
|
Case ascertainment
We defined cases to be all-invasive carcinomas of the colon or rectum, International Classification of Diseases site codes 153.0–153.4, 153.6–153.9, and 154.0–154.1 (15). Case ascertainment came first through self-reports of colorectal cancer from the 1993–1995 and 1995–1998 questionnaires. Nonresponders to these questionnaires were vigorously followed up via phone calls and repeated mailings. Of 311 cases identified through questionnaires, we obtained 245 medical records; the diagnosis of colorectal cancer on the self-report was confirmed by 231 (94%) of these records. Given this high confirmation rate, we concluded that self-reports were sufficiently accurate to justify inclusion of all self-reported colorectal cancer cases without supporting medical records (n = 66). Removal of these 66 cases from the analysis did not significantly affect the results related to the main exposures of interest (data not shown).
Persons with pathology reports contradicting self-reported colorectal cancers were designated as noncases in these analyses. Pathology reports obtained for confirmation of other conditions reported in the questionnaire identified an additional 17 cases of colorectal cancer. In addition to the self-reported cases, we also included as cases subjects identified in reports from the National Death Index (through 1997) as having death certificates indicating colorectal cancer (n = 106). Finally, we matched all subjects residing in states with cancer registries (73.5% of the analytic cohort) against those registries. Subjects matched against state registries did not differ significantly with respect to the distribution of risk factors from those who were not matched against state registries. This procedure resulted in the identification of an additional 65 cases of colorectal cancer. Thus, the final analytic cohort comprised 485 cases of invasive colorectal cancer.
Statistical analysis
We used Cox proportional hazards regression analysis (PROC PHREG, version 6.12; SAS Institute Inc, Cary, NC), with age as the underlying time metric to generate energy-adjusted and fully adjusted relative risks (RRs) and 95% CIs for fruit and vegetable intakes separately.
In analyzing continuous covariates (eg, fiber or physical activity), if a subject had an unrealistically extreme (always defined above the 99th percentile) or missing value, we imputed the median value from the whole sample for that variable. For categorical covariates (smoking status, education level, use of NSAIDs, and multivitamin use), we used a dichotomous classification system. The use of a broader classification system for these variables (ie, going from dichotomous to multiple categories) did not produce any significant change in the results. The assessment of use of NSAIDs in terms of duration of use, daily frequency of use, and the total quantity consumed (duration x frequency) also did not significantly affect the results. For missing information about the categorical variables, we created a missing-value indicator variable for inclusion in all models.
We used 4 models to adjust for total energy intake. The primary model was the multivariate nutrient-density model, in which servings of fruit or vegetables divided by total energy entered the model along with total energy (as a separate covariate). The remaining 3 models were as follows: 1) the energy-partition model, in which servings of fruit or vegetables entered the model along with all other food groups contributing energy to the diet; 2) the residual model, in which the residuals of servings of fruit and vegetables regressed on total energy entered the model; and 3) the standard model, in which total energy entered the fruit or vegetable model as a covariate. A more detailed description of the energy-adjustment models is described elsewhere (16). Each of these energy-adjustment models generates different estimates of relative risk (RR) for fruit and vegetable intakes. Others have shown how the differences between these estimates can prove instructive in assessing associations between diet and disease (17–19).
To test the covariates as potential confounders, we examined fruit and vegetable intakes separately and compared the RRs in these models with those in the models with each of the other covariates added one at a time. Body mass index [wt (kg)/ht2 (m)], height, physical activity, and intakes of grain, red meat, alcohol, folate, calcium, and vitamin D entered these models as continuous variables. To test for nonlinear associations between the covariates and colorectal cancer, we also tested models with each classified into quintiles based on the whole sample, but there were no significant differences between these models and the models using continuous terms. Multivitamin use (yes or no in the previous year), use of NSAIDs (ever or never a regular user), smoking status (ever or never a smoker), and education level (high school graduate or less or some college or more) entered the analyses as dichotomous variables. We also tested fruit as a potential confounder in the vegetable model and vegetables as a potential confounder in the fruit model. In no case, however, did the inclusion of any of the covariates result in a change of 
10% in the RR for fruit or vegetable intake (data not shown), suggesting that none of the covariates were important confounders. We did, however, include all of these covariates in a single model for both fruit and vegetables to test the full effect of their combined inclusion.
We tested for interactions between fruit and vegetables and each of the covariates listed above by comparing the -2 log-likelihood statistic from models with and without interaction terms. We considered P values <0.05 as evidence of interaction.
To examine the data on a continuous scale rather than with a linear trend test, we used a spline-model approach for both fruit and vegetable intakes. This approach allowed for greater flexibility in the observation of associations on a continuous scale than would have been possible assuming a linear relation or by using a categorical method.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The distribution of the sample by quintiles of daily fruit and vegetable consumption per 1000 kJ is shown in Table 2. The median daily intake for the quintiles of fruit ranged from 0.05 to 0.50 servings, a >10-fold difference across quintiles. For vegetables, the median for the high quintile (0.98 servings/d) was just under 4 times as great as that for the low quintile (0.25 servings/d).
|
. Compared with the women in the lowest quintile of fruit intake, those in the highest quintile were slightly older; reported somewhat higher consumptions of folate, calcium, and fiber; reported a noticeably lower consumption of total energy; reported somewhat lower intakes of red meat and alcohol; were more likely to have attended at least some college; and were less likely to have used tobacco. The baseline characteristics of the women by quintile of vegetable intake were similar to those for women by quintile of fruit intake, with few exceptions. Compared with the women in the lowest quintile of vegetable intake, those in the highest quintile were not older, consumed less vitamin D, and consumed similar amounts of red meat. In addition, we observed a modest positive correlation between the lowest and highest quintiles of fruit and vegetable intakes (Spearman's r = 0.24).
Results from the energy-adjusted and fully adjusted Cox proportional hazards regression analysis for both fruit and vegetable intakes (multivariate nutrient-density approach for energy adjustment) are shown in Table 3. For fruit intake, neither the energy-adjusted nor the fully adjusted model (accounting for all the potential confounders simultaneously) provided any indication of an association between servings of fruit and risk of colorectal cancer in this cohort. The RR for quintile 5 compared with quintile 1 was 1.09 (95% CI: 0.82, 1.44) in the energy-adjusted model and was 1.15 (95% CI: 0.86, 1.53) in the fully adjusted model. Similarly, for vegetable intake, the RR for quintile 5 compared with quintile 1 indicated no association between servings of vegetables and the risk of colorectal cancer in either the energy-adjusted (RR: 0.92; 95% CI: 0.70, 1.22) or the fully adjusted (RR: 0.95; 95% CI: 0.71, 1.26) model. We observed no evidence of a linear trend or of a dose-response relation for either fruit or vegetable intake in either the energy-adjusted or the fully adjusted model.
|
). Only for broccoli, potatoes, sweet potatoes, and vegetables from chili did we observe any appreciable reduction in the risk of colorectal cancer associated with increased consumption, and in no case was the reduction in risk for quintile 5 significant. Null associations were observed for all other individual vegetable intakes, except for vegetables from spaghetti and beef stew, which showed positive associations with colorectal cancer. Despite this positive association, removal of vegetables from spaghetti and beef stew from the vegetable group did not significantly change the estimated RR for total vegetable intake in any of the models.
|
The results of the spline-model approach showed no nonlinear association between fruit or vegetable intake and colorectal cancer that differed in any qualitative sense from the results of the quintile-based analyses (data not shown). The energy-partition, standard, and residual models showed clear null associations with the risk of colorectal cancer for fruit and vegetable intakes in both the energy-adjusted and the fully adjusted models. Models with no energy adjustment also showed no associations between fruit or vegetable intake and the risk of colorectal cancer. The RRs for quintile 5 compared with quintile 1 for each of these models are shown in Table 5. In no case were the RRs significantly different from 1.0.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Our findings for vegetables were similar to those for fruit. We began the analyses by focusing on the multivariate nutrient-density model for energy adjustment to show the potential benefit associated with increasing the richness of a diet (servings per 1000 kJ) in fruit or vegetables rather than merely increasing the absolute intake. However, the highest quintile of vegetable intake was associated with no significant reduction in the risk of colorectal cancer compared with the lowest quintile. This result did not change when cases not confirmed by either a pathology report or a death certificate were removed from the analysis or when the covariates were added individually or as a group to the model. In addition, we identified no interactions between the covariates, except for the interaction between fruit and vegetable intake (negative interaction) and between vegetable intake and education level (negative interaction). These are potentially interesting findings (especially the interaction between fruit and vegetable intakes), but we must caution against placing too much weight on them because the P values determined from the interaction tests indicated only marginal significance. Thus, we cannot rule out the possibility of chance findings given the multiple-comparisons issues involved in performing a large number of interaction tests. The finding that our results did not change significantly after exclusion of cases identified in the first 2 y of follow-up (data not shown) suggests that the observed effects were not the result of dietary changes related to preclinical disease.
We observed no significant association between vegetable intake and colorectal cancer with the nutrient-density model or with the 3 alternative models of energy adjustment. Given the lack of association in the nutrient-density model and the decidedly null relations observed in the other models, we conclude that there was little overall association between vegetable intake and colorectal cancer in this data set.
The overall results for vegetable intake agree with those from the Nurse's Health Study, the Health Professionals Follow-up Study, and the Alpha-Tocopherol Beta-Carotene Cancer Prevention Study cohorts (6,20), which also showed no evidence of an association between vegetable intake and colorectal cancer. Of the remaining prospective studies that did show an inverse association, the magnitude of risk reduction was always modest and in 3 of the 4 studies was not significant (8,21–23). Our results fit with a growing body of evidence that fails to confirm the hypothesis of an inverse association between vegetable intake and colorectal cancer.
Furthermore, the results by quintiles of individual vegetable intakes did not provide substantial support for an inverse association between the intake of any single vegetable and colorectal cancer. The intake of only 4 vegetables showed any indication at all of a reduction in risk; of these 4 vegetables, none showed a risk substantially greater than that for total vegetable intake, and 2 foods (vegetables from spaghetti and beef stew) even showed an increased risk. Note, however, that the servings per day of vegetables from spaghetti and beef stew were very low in this cohort: the cutoffs for the highest quintile were only 0.020 and 0.030 servings per 1000 kJ, respectively. The servings per day of many individual vegetables were similarly low: the cutoff for the highest quintile was <0.035 servings per 1000 kJ for coleslaw, collards and other greens, spinach, sweet potatoes, beef stew, chili, and spaghetti. However, the results did indicate that none of the vegetables individually and no single food or subset of foods appeared to be inversely associated with colorectal cancer.
However, there are several issues to consider when evaluating the association between fruit and vegetable intakes and colorectal cancer in this and other epidemiologic studies. First, the relative intake of fruit and vegetables in the women in the highest compared with the lowest quintiles of intake indicated a wide range of consumption. However, despite a 10-fold increase in intake from quintiles 1 to 5 for fruit, the median servings per day in the highest quintile was only 0.50 servings per 1000 kJ. The servings per day of vegetables in quintile 5 was somewhat higher, 0.98 servings per 1000 kJ; however, the range of intakes was lower—only 4 times that in quintile 1. Furthermore, the crude consumption (ie, not per 1000 kJ) of vegetables in quintile 5 of the BCDDP cohort was still only 4.9 servings/d, which is low compared with international ranges. According to 1996 data on national food supplies from the Food and Agriculture Organization, vegetable consumption in the United States (5.44 servings/d) was substantially less than that in, for example, China (7.03 servings/d) (24). Although ecologic comparisons of this type must be made with caution, it remains possible that we and other investigators (6,8,20–23,25) did not observe an inverse association between vegetable intake and colorectal cancer because our study population did not consume these foods in sufficient quantity to yield a noticeable reduction in risk.
The possibility of measurement error introduces another obstacle to observing true associations in studies such as ours. FFQs must, by practical necessity, omit many foods that individuals actually consume. In the BCDDP cohort, the fruit group consisted of only 5 food items, and the vegetable group—although somewhat more comprehensive with 14 foods—was still far from exhaustive in its coverage. Furthermore, a single FFQ-based measurement in adulthood may not represent long-term intake without error and may not assess the diet accurately for times when exposure is most critical in determining disease outcome. It is possible that we misclassified people in terms of their total vegetable and fruit intakes, and we must consider seriously the possibility that the FFQ did not include a food or foods that make important contributions to colorectal carcinogenesis. Both the misclassification of intakes and the omission of important food items from the FFQ could lead to an attenuation of risk estimates.
Ultimately, this study failed to provide substantial evidence that the true relation between fruit and vegetable intakes and colorectal cancer is other than null. The limitations inherent to FFQ-based studies, however, make it difficult to rule out definitively the possibility of a reduction in risk of colorectal cancer with an increased intake of vegetables. Nonetheless, given the recent series of negative findings from studies by Michels et al (6), Schatzkin et al (7), and Voorrips et al (8), as well as the results of the present study, the case in favor of the hypothesized inverse association between fruit and vegetable intakes and risk of colorectal cancer has lost considerable strength. To resolve this uncertainty, studies using more robust and more direct tests of the hypothesis are needed. Extensions of the ongoing work to improve FFQ data (25) and the identification of study populations with wider ranges of exposure will make important contributions in this regard. We also must consider seriously, however, the possibility that the consumption of fruit and vegetables may not have a direct, independent role to play in the etiology of colorectal cancer.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  A. M. Nomura, L. R Wilkens, S. P Murphy, J. H Hankin, B. E Henderson, M. C Pike, and L. N Kolonel Association of vegetable, fruit, and grain intakes with colorectal cancer: the Multiethnic Cohort Study Am. J. Clinical Nutrition, September 1, 2008; 88(3): 730 - 737. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Flood, T. Rastogi, E. Wirfalt, P. N Mitrou, J. Reedy, A. F Subar, V. Kipnis, T. Mouw, A. R Hollenbeck, M. Leitzmann, et al. Dietary patterns as identified by factor analysis and colorectal cancer among middle-aged Americans Am. J. Clinical Nutrition, July 1, 2008; 88(1): 176 - 184. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Takachi, M. Inoue, J. Ishihara, N. Kurahashi, M. Iwasaki, S. Sasazuki, H. Iso, Y. Tsubono, S. Tsugane, and for the JPHC Study Group Fruit and Vegetable Intake and Risk of Total Cancer and Cardiovascular Disease: Japan Public Health Center-based Prospective Study Am. J. Epidemiol., January 1, 2008; 167(1): 59 - 70. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. E Millen, A. F Subar, B. I Graubard, U. Peters, R. B Hayes, J. L Weissfeld, L. A Yokochi, R. G Ziegler, and for the Prostate, Lung, Colorectal, and Ovarian (P Fruit and vegetable intake and prevalence of colorectal adenoma in a cancer screening trial Am. J. Clinical Nutrition, December 1, 2007; 86(6): 1754 - 1764. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. F. Krebs, J. H. Himes, D. Jacobson, T. A. Nicklas, P. Guilday, and D. Styne Assessment of Child and Adolescent Overweight and Obesity Pediatrics, December 1, 2007; 120(Supplement_4): S193 - S228. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Koushik, D. J. Hunter, D. Spiegelman, W. L. Beeson, P. A. van den Brandt, J. E. Buring, E. E. Calle, E. Cho, G. E. Fraser, J. L. Freudenheim, et al. Fruits, Vegetables, and Colon Cancer Risk in a Pooled Analysis of 14 Cohort Studies J Natl Cancer Inst, October 3, 2007; 99(19): 1471 - 1483. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Y. Park, A. F. Subar, V. Kipnis, F. E. Thompson, T. Mouw, A. Hollenbeck, M. F. Leitzmann, and A. Schatzkin Fruit and Vegetable Intakes and Risk of Colorectal Cancer in the NIH-AARP Diet and Health Study Am. J. Epidemiol., July 15, 2007; 166(2): 170 - 180. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. A Ello-Martin, L. S Roe, J. H Ledikwe, A. M Beach, and B. J Rolls Dietary energy density in the treatment of obesity: a year-long trial comparing 2 weight-loss diets Am. J. Clinical Nutrition, June 1, 2007; 85(6): 1465 - 1477. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. L. Austin, L. S. Adair, J. A. Galanko, C. F. Martin, J. A. Satia, and R. S. Sandler A Diet High in Fruits and Low in Meats Reduces the Risk of Colorectal Adenomas J. Nutr., April 1, 2007; 137(4): 999 - 1004. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Mannisto, S.-S. Yaun, D. J. Hunter, D. Spiegelman, H.-O. Adami, D. Albanes, P. A. van den Brandt, J. E. Buring, J. R. Cerhan, G. A. Colditz, et al. Dietary Carotenoids and Risk of Colorectal Cancer in a Pooled Analysis of 11 Cohort Studies Am. J. Epidemiol., February 1, 2007; 165(3): 246 - 255. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. A. Smith-Warner, D. Spiegelman, J. Ritz, D. Albanes, W. L. Beeson, L. Bernstein, F. Berrino, P. A. van den Brandt, J. E. Buring, E. Cho, et al. Methods for Pooling Results of Epidemiologic Studies: The Pooling Project of Prospective Studies of Diet and Cancer Am. J. Epidemiol., June 1, 2006; 163(11): 1053 - 1064. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S.-Y. Park, S. P. Murphy, L. R. Wilkens, J. F. Yamamoto, and L. N. Kolonel Allowing for Variations in Multivitamin Supplement Composition Improves Nutrient Intake Estimates for Epidemiologic Studies J. Nutr., May 1, 2006; 136(5): 1359 - 1364. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H.-C. Hung, K. J. Joshipura, R. Jiang, F. B. Hu, D. Hunter, S. A. Smith-Warner, G. A. Colditz, B. Rosner, D. Spiegelman, and W. C. Willett Fruit and Vegetable Intake and Risk of Major Chronic Disease J Natl Cancer Inst, November 3, 2004; 96(21): 1577 - 1584. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Sharp and J. Little Polymorphisms in Genes Involved in Folate Metabolism and Colorectal Neoplasia: A HuGE Review Am. J. Epidemiol., March 1, 2004; 159(5): 423 - 443. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K.J. Dyer, K C H Fearon, K. Buckner, and R.A. Richardson Diet and colorectal cancer risk: Baseline dietary knowledge of colorectal patients Health Education Journal, January 1, 2004; 63(3): 242 - 253. [Abstract] [PDF]
|
|
|
|
 |
|
 M Ponz de Leon, P Benatti, F Borghi, M Pedroni, A Scarselli, C Di Gregorio, L Losi, A Viel, M Genuardi, G Abbati, et al. Aetiology of colorectal cancer and relevance of monogenic inheritance Gut, January 1, 2004; 53(1): 115 - 122. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Little, L. Sharp, S. Duthie, and S. Narayanan Colon Cancer and Genetic Variation in Folate Metabolism: The Clinical Bottom Line J. Nutr., November 1, 2003; 133(11): 3758S - 3766. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. K. Weir, M. J. Thun, B. F. Hankey, L. A. G. Ries, H. L. Howe, P. A. Wingo, A. Jemal, E. Ward, R. N. Anderson, and B. K. Edwards Annual Report to the Nation on the Status of Cancer, 1975-2000, Featuring the Uses of Surveillance Data for Cancer Prevention and Control J Natl Cancer Inst, September 3, 2003; 95(17): 1276 - 1299. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. Riboli and T. Norat Epidemiologic evidence of the protective effect of fruit and vegetables on cancer risk Am. J. Clinical Nutrition, September 1, 2003; 78(3): 559S - 569. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Flood, E. M. Velie, R. Sinha, N. Chaterjee, J. V. Lacey Jr., C. Schairer, and A. Schatzkin Meat, Fat, and Their Subtypes as Risk Factors for Colorectal Cancer in a Prospective Cohort of Women Am. J. Epidemiol., July 1, 2003; 158(1): 59 - 68. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Little, M. J. Khoury, L. Bradley, M. Clyne, M. Gwinn, B. Lin, M.-L. Lindegren, and P. Yoon The Human Genome Project Is Complete. How Do We Develop a Handle for the Pump? Am. J. Epidemiol., April 15, 2003; 157(8): 667 - 673. [Full Text] [PDF]
|
|
|
|
 |
|
 D. A Lawlor and A. R Ness Commentary: The rough world of nutritional epidemiology: Does dietary fibre prevent large bowel cancer? Int. J. Epidemiol., April 1, 2003; 32(2): 239 - 243. [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||
