点击显示 收起
1 From the University at Buffalo, School of Public Health and Health Professions, Department of Social and Preventive Medicine, Buffalo, NY (AEM); the National Cancer Institute, Division of Cancer Control and Population Sciences, Applied Research Program, Risk Factor Monitoring and Methods Branch, Bethesda, MD (AFS); the National Cancer Institute, Division of Cancer Epidemiology and Genetics, Epidemiology and Biostatistics Program, Biostatistics Branch, Bethesda, MD (BIG); the Fred Hutchinson Cancer Research Center, Public Health Sciences, Cancer Prevention, Seattle, WA (UP); the National Cancer Institute, Division of Cancer Epidemiology and Genetics, Epidemiology and Biostatistics Program, Occupational and Environmental Epidemiology Branch, Bethesda, MD (RBH); the University of Pittsburgh, Department of Epidemiology, Pittsburgh, PA (JLW); the Pacific Health Research Institute, Honolulu, HI (LAY); and the National Cancer Institute, Division of Cancer Epidemiology and Genetics, Epidemiology and Biostatistics Program, Office of the Director, Bethesda, MD (RGZ)
2 The Prostate Lung Colorectal and Ovarian Cancer Screening Trial is fully funded by the National Cancer Institute, National Institutes for Health, and the Department of Health and Human Services of the US government. 3 Address reprint requests and correspondence to AE Millen, University at Buffalo, School of Public Health and Health Professions, Department of Social and Preventive Medicine, Farber Hall, Room 270, Buffalo, NY, 14214-8001. E-mail: aemillen{at}buffalo.edu.
ABSTRACT
Background: Research on the association between fruit and vegetable intake and risk of colorectal adenoma is inconclusive.
Objective: We studied whether intake of fruit, vegetables, or their subgroups is associated with a lower risk of prevalent colorectal adenoma.
Design: In men and women (aged 55–74 y) who were screened for colorectal cancer in the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial (PLCO) (1993–2001), we compared 3057 cases with at least one prevalent histologically verified adenoma of the distal large bowel with 29 413 control subjects. Using a food-frequency questionnaire, we quantified intake of fruit and vegetables in the 12 mo before screening as energy-adjusted pyramid servings/d (ps/d). Adjusted odds ratios (ORs) and 95% CIs were estimated by logistic regression.
Results: Risk of distal adenoma was significantly lower among subjects in high (5.7 ps/d) versus low (1.2 ps/d) quintiles of total fruit intake (OR: 0.75; 95% CI: 0.66, 0.86, P for trend <0.001), which was not completely explained by dietary folate or fiber intake. Inverse associations between adenoma and total fruit intake were observed regardless of adenoma histopathology and multiplicity. However, the protective effect was seen only for colon and not rectal adenoma. Total vegetable intake was not significantly associated with reduced risk of adenoma. ORs for colorectal adenoma among persons with high versus low intakes of deep-yellow vegetables, dark-green vegetables, and onions and garlic were significantly related to lower risk of adenoma, although the P for trend for dark-green vegetables was not significant.
Conclusion: Diets rich in fruit and deep-yellow vegetables, dark-green vegetables, and onions and garlic are modestly associated with reduced risk of colorectal adenoma, a precursor of colorectal cancer.
Key Words: Fruit • vegetables • epidemiology • colorectal neoplasms • adenoma • diet
INTRODUCTION
Since the mid 1980s, numerous studies have investigated the relation between colorectal cancer and consumption of fruit and vegetables. The hypotheses as to how fruit and vegetable intake may reduce the risk of colon or rectal cancer are numerous and involve independently or additively the many potential anticarcinogenic compounds found in fruit and vegetables (eg, fiber, carotenoids, vitamin C, folate, glucosinolates, and allium compounds) (1-3).
In 1997 the American Institute for Cancer Research reviewed the literature (4 prospective and 22 case-control studies) on diet and risk of all types of cancer (4), and concluded that the "evidence that diets rich in vegetables protect against cancers of the colon and rectum is convincing. The data on fruits are more limited and inconsistent; no judgment is possible." (4). Since the American Institute for Cancer Research published its review, at least 14 additional prospective studies, which assessed diet before development of disease and thus were not subject to recall bias in diet reports, investigated the role of fruit and vegetables in colorectal cancer prevention (5-18). Among the largest of these prospective studies, 2 (11, 13) observed no reduction in risk of colorectal cancer with intake of fruit and vegetables, and 1 (18) observed a reduction in risk for colorectal cancer with high compared with low intake of total vegetables in men but not women. Additionally, the International Agency for Research on Cancer's 2003 review of the literature suggested no association between fruit and vegetable consumption and risk of colorectal cancer (19), and recent results from the Women's Health Initiative Randomized Controlled Dietary Modification Trial (20) did not support a protective role of fruit and vegetable consumption on colorectal cancer risk in postmenopausal women.
Concurrently, the literature on the role of fruit and vegetable intake with respect to colorectal adenoma, a precursor to colorectal cancer (21), includes 23 studies (22-44); however, only 2 of these were cohort studies with prospective designs in which diet was assessed before the development of disease (32, 40). The prospective cohort studies suggest that intake of fruit and specific vegetables may also reduce the risk of colorectal adenoma, suggesting protective effects early in the carcinogenic process. However, in 2000 a randomized clinical trial of a high-fiber, low-fat diet enriched with fruit and vegetables did not show any influence of these dietary components on the risk of colorectal adenoma recurrence (35).
Because of the remaining uncertainty about the influences of fruit and vegetable intake on colorectal adenoma, we evaluated these risks in the Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial. The present study is substantially larger than any previously published cohort study of colorectal adenoma (n = 1720 cases) (40). Among the 58 000 men and women randomly assigned to the PLCO screening arm by 2001, a histologically verified distal adenoma was diagnosed at baseline in >3000 participants by means of endoscopy. We analyzed associations between intake of fruit and vegetables, which were determined from a baseline food-frequency questionnaire (FFQ), and risk of adenoma in the distal large bowl and by adenoma subtype (histologic type, number, and location in the bowel). We hypothesized that intake of fruit and vegetables would be associated with decreased risk of colorectal adenoma.
SUBJECTS AND METHODS
Study design and population
This study was conducted as part of the PLCO Cancer Screening Trial (45, 46), a multisite investigation (Birmingham, AL; Denver, CO; Detroit, MI; Honolulu, HI; Marshfield, WI; Minneapolis, MN; Pittsburgh, PA: Salt Lake City, UT; St. Louis, MO; and Washington, DC) of the effectiveness of early detection of prostate, lung, colorectal, and ovarian cancers. After approval by the institutional review boards of the US National Cancer Institute and the participating centers, each eligible participant provided informed consent.
PLCO trial participants randomly assigned to the screening arm of the trial were offered a 60-cm flexible sigmoidoscopic exam of the distal colon (including the descending colon, sigmoid colon, and the rectum) at trial entry. Those identified with lesions suspect for colorectal neoplasia (ie, sigmoidoscopically visualized polypoid lesions or masses) were referred to their medical care providers for further diagnostic work-up. All available medical and pathological reports on all lesions removed during the diagnostic and related surgical procedures were obtained and coded by trained medical abstractors.
Between November 1993 and December 2001, 57 560 men and women aged 55–74 y had an initial successful sigmoidoscopic screening exam, which was described in detail elsewhere (47). Of these, 52 102 participants (90.5%) completed the baseline risk factor questionnaire and FFQ. After excluding a total of 7570 participants for self-reported history of cancer other than nonmelanoma skin cancer (n = 2360); self-reported history of colorectal polyps, ulcerative colitis, Crohn disease, familial polyposis, or Gardner syndrome (n = 4796); extremely high or low energy intake (highest and lowest 1% in sex-specific energy intake; n = 998); or >7 items missing on the FFQ (n = 440), 44 532 participants remained in the study. The participants (n = 5972) were ineligible for this study if they had hyperplastic polyps only (n = 1545), benign lesions not further specified (n = 297), colorectal lesions (polyps or cancer) of unknown location (n = 277), polyps of uncertain histology or cancer (n = 1241), indeterminate screening results (n = 13), or positive screening results but no follow-up endoscopy (n = 2599, of which 1887, or 72.6%, had a polyp on the screening exam <5 mm in the distal colon). Some participants were excluded for more than one reason.
The FFQ was completed before (n = 10 838; 28%), on (n = 21 632; 56%), or after (n = 6079; 16%) the day of the sigmoidoscopy among the remaining 38 560 participants (excluding those missing data on the timing of FFQ completion, n = 11). The protocols given to the study centers instructed that, on the day of screening, the participants should complete the risk factor and dietary questionnaire before the screening procedures began. Because of the potential for dietary recall bias, individuals who filled out their FFQs after the screening, and those with missing data, were excluded from these analyses; this left 32 470 participants.
All cases included in this study had a positive result on the screening exam followed by histologic verification of colorectal adenoma of the distal colon (n = 3057). Comparison controls screened negative for polyps of the distal colon (n = 27 966) or screened positive but were found on follow-up diagnostic exam not to have polyps of the distal colon (n = 1447). Of the 3057 cases, 1164 (38.1%) were considered advanced on the basis of the presence of at least one adenoma 1 cm in size or with high-grade dysplasia (including cancer in situ) or villous elements (including tubulovillous adenomas). Adenomas were also categorized according to number, with 735 cases having 2 adenomas. The data for this analysis were last updated in April 2004.
Baseline questionnaires
At the time of random assignment, all study subjects were asked to complete a self-administered baseline questionnaire that included questions on demographic factors, medical history, and health-related behaviors. All participants randomly assigned to the PLCO screening arm were given a 137-item FFQ, which was designed to be self-administered and to characterize usual dietary intake over the past 12 mo (Internet: http://www.cancer.gov/prevention/plco/DQX.pdf). This FFQ was modeled after 3 well-established and validated questionnaires: the National Cancer Institute's Diet History Questionnaire (48), the Block FFQ (49), and the Willett FFQ (50), and incorporated elements of both cognitive (48, 51, 52) and database (53) research. Descriptive data for calculating nutrients and food groups were derived from the two 24-h recalls administered in the 1994–1996 US Department of Agriculture's Continuing Survey of Food Intake by Individuals (CSFII) (53), a nationally representative survey conducted during the period when the FFQ was being used.
The FFQ also queried for use of supplements taken since 25 y of age, including multivitamins [one-a-day type (100% RDA; Bayer Corp, Pittsburgh, PA), therapeutic or high-dose type (>100% RDA, such as Theragran; Bristol-Myers Squibb, New York, NY), Stresstabs (B-complex + vitamin C; Inverness Medical Inc, Waltham, MA), B-complex, and other] and single-nutrient supplements (vitamin A, β-carotene, vitamin C, vitamin E, calcium, and vitamin D). Data were obtained about current and past supplement use (use 2 and 5 y ago), duration of use (y), and daily or weekly dose. Nutrient intake from supplements was estimated from intake of both multiple and single supplements, except for folic acid, which was estimated from intake of multivitamins only (54).
Fruit and vegetable frequency measures and pyramid servings
A methodologic component of these analyses was to compare the study results when fruit and vegetable intake was measured by using frequencies compared with pyramid servings. Food frequencies are the reported amount of times a food is eaten in a specific time period. The fruit and vegetable intake responses on the PLCO FFQ were converted into frequencies/d (f/d) of consumption of specific food groups by summing the daily frequency reports of FFQ line items included in the food group. Frequencies do not incorporate portion size measurements or food intake from hidden sources, such as mixed dishes.
Pyramid servings/d (ps/d) of fruit and vegetable food groups were estimated from the fruit and vegetable intake responses on the FFQ by using a Pyramid Servings Database (55) developed from the 1994–1996 CSFII data (28). The Pyramid Servings Database uses a recipe file to disaggregate food mixtures into their component ingredients and assigns the components to appropriate food groups and also calculates standardized quantitative estimates of food group intake, called pyramid servings. Pyramid servings reflect the full, continuous range of reported food intake, including small amounts of foods from all sources, and are standardized to portion sizes from the 1992 Food Guide Pyramid (56). The Diet History Questionnaire, one of the FFQs that the PLCO FFQ was modeled after, was previously validated for pyramid servings (57), and a more detailed description of pyramid servings is provided elsewhere (57). The fruit and vegetable food groups for which ps/d were created are listed in Appendix A.
View this table:
APPENDIX A. Fruit and vegetables contained in the different food groups
Statistical analyses
Fruit and vegetable intakes were energy-adjusted by using the residual method (58). Log-transformed fruit and vegetable intakes were regressed on log-transformed total energy intake to compute residuals. The mean of the log-transformed fruit and vegetable intake was added to each residual, and the antilogarithm of this sum was taken. Further addition of total calories to the models by using fruit and vegetable intake adjusted by the residual method or by adjusting for energy by using the nutrient density method did not significantly alter the results.
The intake of fruit and vegetable food groups, as ps/d and f/d, was categorized by sex-specific quintiles, with quintile 1 being the lowest intake and quintile 5 the highest. Odds ratios (ORs) and 95% CIs (95% CIs) for colorectal adenoma were calculated by using logistic regression models for quintiles of fruit and vegetable intake, with the lowest quintile level as the reference category. Linear trends (P for trend) were assessed by assigning the quintiles the values 1, 2, 3, 4, or 5.
Two models are presented in the tables. The study design model was adjusted for age at screening, sex, and study center. The multivariate model was additionally adjusted for a priori nondietary risk factors for colorectal adenoma and colorectal cancer: ethnic origin (non-Hispanic white, non-Hispanic black, Hispanic, Asian, Pacific Islander, and American Indian or Alaskan native), educational attainment (<8 y, 8–11 y, 12 y or high school equivalent, post-high school other than college, some college, college graduate, or postgraduate), family history of colon cancer [yes (sibling, parent, or child with colon cancer) or no], smoking (never smoked, only smoked cigars or pipe, quit cigarette smoking 20 y ago and 1 pack cigarettes/d, quit 20 y ago and >1 pack/d, quit <20 y ago and 1 pack/d, or quit <20 y ago and >1 pack/d), alcohol use (<1 g/d, 1–15 g/d, >15–30 g/d, or >30 g/d), use of aspirin and ibuprofen (no regular use, <2 times/mo, 2–3 times/mo, 1 time/wk, 2 times/wk, 3–4 times/wk, 1 time/d, or 2 times/d), use of hormone replacement therapy (never used, former user 5 y, former user 6 y, current user 5 y, current user 6 y), physical activity (no, <1 h, 1 h, 2 h, 3 h, 4 h of vigorous activity/wk), and current body mass index (BMI; in kg/m2). Because substantial confounding, primarily due to smoking, was observed for fruit and vegetable subgroups, we chose to focus on the estimate of risk obtained from the multivariate model.
Relations of colorectal adenoma with fruit and vegetable intake were examined by levels of potential effect modifiers, both dietary (animal fat intake, calcium intake, or red meat consumption) and nondietary (age, sex, BMI, ever having smoked, family history of colon cancer, regular ibuprofen use, regular aspirin use, or multivitamin use), by including interaction terms between intake (as a 5-level, ordinal variable) and the effect modifier. A Wald chi-square test with a P value of < 0.05 for the overall interaction term was considered statistically significant.
All analyses were conducted by using SAS version 8.2 software (SAS Institute Inc, Cary, NC). All statistical tests were two-sided with a significance level of 0.05.
RESULTS
Men and women in the highest quintile of fruit or vegetables reported consuming on average 5 ps/d more than did those in the lowest quintile of intake, and more vegetables were reported consumed than fruit (Table 1). For both fruit and vegetables, high consumers were more likely than low consumers to be older, more highly educated, leaner, more physically active, never-smokers, and infrequent drinkers.
View this table:
TABLE 1. Characteristics of the study participants in quintiles 1 and 5 of fruit and vegetable intake: the PLCO Cancer Screening Trial1
In the overall population of men and women combined, there was a statistically significantly decreased risk of colorectal adenoma among participants who consumed high versus low ps/d of total fruit (OR comparing extreme quintiles = 0.58; 95% CI: 0.52, 0.66; P for trend < 0.001) and total vegetables (OR comparing extreme quintiles = 0.81; 95% CI = 0.72, 0.92; P for trend <0.001), after adjustment for age at screening, sex, and study center (Table 2, study design model). After further adjustment for a priori, nondietary risk factors, the statistically significant association remained for total fruit (OR comparing extreme quintiles = 0.75; 95% CI = 0.66, 0.86; P for trend <0.001), but not for total vegetables (OR comparing extreme quintiles = 0.94; 95% CI = 0.83, 1.06,;P for trend = 0.24) (Table 2, multivariate model). Additionally, in the multivariate model, a reduction in risk of adenoma was observed when extreme quintiles of total fruit and vegetable intake combined (ps/d) were compared (OR comparing extreme quintiles = 0.82; 95% CI = 0.72, 0.93; P for trend <0.001).
View this table:
TABLE 2. Odds ratios (ORs) and 95% CIs for prevalent colorectal adenoma by fruit and vegetable intake, in energy-adjusted pyramid servings/d in the overall population of men and women: the PLCO Cancer Screening Trial1
When potatoes were removed from the total vegetable food group, a modest but not statistically significant relation between total vegetable intake (minus potatoes) and adenoma was observed in the multivariate model (OR comparing extreme quintiles = 0.90; 95% CI = 0.80, 1.02; P for trend = 0.08; Table 2). Also observed in the multivariate model was a decreased risk of adenoma among participants with high versus low intake, in ps/d, of fruit (without juice); fruit juice; citrus, melon, and berry fruit; deep-yellow vegetables; dark-green vegetables; and onions and garlic. However, the P for trend for dark-green vegetables was >0.05.
When these analyses were repeated using f/d rather than ps/d, the results were comparable. However, no relation was observed between colorectal adenoma and intake of dark-green vegetables or onions and garlic (data not shown). For total fruit intake and total vegetable intake, measured in f/d, the multivariate ORs and 95% CIs for adenoma in quintile 5 versus 1 were 0.75 (95% CI = 0.66, 0.86; P for trend <0.001) and 0.91 (95% CI = 0.80, 1.02; P for trend = 0.10), respectively.
The relation between risk of adenoma and total fruit intake also did not vary significantly by age at screening, BMI, family history of colon cancer, regular ibuprofen use, regular aspirin use, multivitamin use, animal fat intake, calcium intake, or red meat consumption, but did differ statistically significantly by ever having smoked (P for interaction = 0.001) and sex (P for interaction < 0.001). The relation between risk of adenoma and total vegetable intake did not vary significantly by any of these factors.
The risk of adenoma among participants in quintile 5 compared with 1 for intake of total fruit was stronger among those who had ever smoked (OR = 0.68; 95% CI = 0.57, 0.80; P for trend < 0.001) than among those who had never smoked (OR = 0.92; 95% CI = 0.74, 1.14; P for trend = 0.29) and in men (OR = 0.62; 95% CI = 0.53, 0.73; P for trend < 0.001) than in women (OR = 0.86; 95% CI = 0.70, 1.06; P for trend = 0.27).
It is possible that the observed associations in Table 2 were confounded by intake of specific nutrients or foods hypothesized to be related to adenoma risk. Further adjustment of the multivariate model for current intake of folate from supplements, calcium from diet and supplements combined, vitamin D from diet and supplements combined, and red meat modestly weakened the risk reduction for all fruit and vegetables listed in Table 3. The inverse associations, ORs comparing extreme quintiles, and the P for trend for total fruit, fruit without juice, and deep-yellow vegetables remained significant (P < 0.05). The P for trend but not the ORs comparing extreme quintiles remained statistically significant (P < 0.05) for citrus, melon, and berry fruit and onions and garlic. The statistically significant protective effect of fruit juice and the statistically significant OR comparing extreme quintiles for dark-green vegetables was removed.
View this table:
TABLE 3. Odds ratios (ORs) and 95% CIs for prevalent colorectal adenoma among participants in quintile 5, relative to quintile 1, of fruit and vegetable intake in energy-adjusted pyramid servings/d: PLCO Cancer Screening Trial
Further analyses were conducted to see whether the observed associations between risk of adenoma and intake of fruit and vegetables could be explained by specific nutrients concentrated in fruit and vegetables, such as dietary folate and fiber (Table 3). Adjustment of the multivariate model for either nutrient could explain only part of the decreased risk associated with fruit consumption (total and without juice). Adjustment of the multivariate model for dietary folate explained practically all of the reduction in risk associated with the fruit juice; citrus, melon, and berry fruit; and all vegetable subgroups. Adjustment of the multivariate model for dietary fiber explained practically all the reduction in risk associated with deep-yellow vegetables and dark-green vegetables, but not fruit juice; citrus, melon, and berry fruit; and onions and garlic. After adjustment of the multivariate model for dietary fiber, the risk of adenoma among participants in quintile 5 versus 1 for intake of total vegetables was increased. After removal of potatoes from the total vegetable group, the risk of adenoma, when comparing extreme quintiles of intake, was no longer significant, although the P for trend remained <0.05.
Additional adjustment of the multivariate model independently for multivitamin use (yes, no), duration of use of multivitamin supplements, calcium supplements, or vitamin D supplements did not remove the statistically significant association between risk for adenoma and total fruit intake (data not shown).
Similar patterns as the overall analyses were observed for nonadvanced and advanced colorectal adenomas and for single and multiple colorectal adenomas (Table 4). Increasing intake of total fruit was statistically significantly related to decreasing risk of nonadvanced, advanced, single, and multiple adenomas. No statistically significant relations were observed for total vegetable intake, although the P for trend for total vegetables without potatoes was statistically significant for advanced adenoma (P for trend = 0.03) and multiple adenomas (P for trend = 0.04). The observed decreased risk of adenoma with intake of fruit and vegetables was more apparent for adenomas in the colon than in the rectum. Total fruit intake was associated with a significant decrease in prevalence of colon adenomas and a nonsignificant decrease in prevalence of rectal adenomas, although the OR was <1.0. Total vegetable intake, after potatoes were excluded, was associated with a significant decrease in the prevalence of colon adenomas when comparing extreme quintiles (P for trend = 0.10), but was unrelated to the prevalence of rectal adenomas.
View this table:
TABLE 4. Adjusted odds ratios (ORs) and 95% CIs for prevalent adenoma by histopathology (nonadvanced or advanced), number (single or multiple), or location (colon or rectum) among participants in quintiles 2–5 compared with quintile 1 for total fruit and vegetable intake measured as pyramid servings/d: the PLCO Cancer Screening Trial1
DISCUSSION
Among the participants of the PLCO Cancer Screening Trial at baseline, we observed a statistically significant decreased risk of prevalent colorectal adenoma among participants who consumed high (5.7 ps/d) compared with low (1.2 ps/d) intakes of total fruit but not total vegetables. The statistically significant associations observed between adenoma and intake of total fruit were consistent regardless of adenoma type or number and were not explained by intake of dietary folate or fiber. However, intake of total fruit was statistically significantly protective only for adenomas in the colon and not the rectum. No statistically significant associations were observed between total vegetable consumption and advanced, nonadvanced, single, multiple, colon, or rectal adenoma. Our results for total fruit and total vegetable intake were similar whether we measured fruit and vegetable intake as frequencies or as pyramid servings. However, using pyramid servings may be beneficial from a public health perspective because they are easily translated into the Food Guide Pyramid's (Internet: http://www.mypyramid.gov/) dietary guidance.
Colorectal adenomas, precursors of most colorectal carcinomas, probably occur a decade or more before they evolve into colorectal cancer (59, 60). If diet is critical in the early stages of colorectal carcinogenesis, then assessing diet closer to the appearance of adenomas, which are often asymptomatic, may provide relevant information. To date, research on fruit and vegetable intake and risk of colorectal adenoma has been inconclusive.
Only one randomized dietary interventional trial evaluated the importance of vegetable and fruit intake in colorectal carcinogenesis. In this trial (n = 1905), a high-fiber, low-fat diet enriched with fruit and vegetables (3.5 servings/1000 kcal) had no effect on adenoma recurrence among persons previously diagnosed with at least one colorectal adenoma (35). However, the 4-y intervention period may have been too short to affect tumor development, and the increase in vegetable and fruit intake in the intervention arm compared with the control arm was modest (1.1 servings/1000 kcal).
Two cohort studies (32, 40), with prospectively collected dietary data and physician-confirmed self-reports of adenomas, have investigated the relation between risk of colorectal adenomas and intake of fruit and vegetables. In one of these cohorts, which was conducted in men (32), a statistically significant trend (P = 0.03) for decreased risk of prevalent colorectal adenoma with intake of fruit (minus fruit juice), but not vegetables, was observed. In the other cohort study, which was conducted in women (40), a statistically significant decreased risk of prevalent adenoma in high versus low consumers of total fruit, but not vegetables, was observed. However, no statistically significant decreased risk for incident colorectal adenoma was observed for high versus low intake of fruit or vegetables. However, those authors did observe a statistically significant decreased risk (12%) of incident colorectal adenoma with each additional serving of fruit per day.
In our study. we observed a decreased risk of adenoma with intake of total fruit in men but not women. The sex difference could not be explained by a narrower distribution of pyramid servings of total fruit or vegetables in women, because the opposite was true. The difference could also not be explained by the smaller number of women than of men with adenoma, because we had 80% power to detect an OR of 0.75 among women and 0.80 among men. However, we may not have observed a statistically significant risk estimate in women because of the measurement error characteristics of the FFQ, which are shown to be greater in women (61). It is also possible that the difference is explained by a yet unknown sex-specific biological mechanism.
We also observed a stronger protective effect for adenoma with intake of total fruit among persons who had ever smoked than among never smokers. This suggests that consumption of fruit may protect against adenoma development more so in persons who had ever smoked than in never smokers. This is the opposite of what Michels et al (40) found in the Nurses' Health Study. More research on how smoking might modify the relation between fruit and vegetable intake and risk of colorectal adenoma is needed.
We also observed a decreased risk of adenoma with high versus low intake of citrus, melon, and berry fruit; deep-yellow vegetables; dark-green vegetables; and onions and garlic. Other studies have analyzed associations between adenoma risk and intake of citrus fruit or fruit high in vitamin C (31, 38–40), green and yellow or high-carotenoid vegetables (24, 31, 32, 38–40), and onions or garlic (31, 38). Statistically significant decreased risks of adenoma were previously observed when comparing high with low consumers of citrus fruit (40), high–vitamin C fruit (31), folate-rich vegetables (32), green and yellow vegetables (24, 40), high-carotenoid vegetables (31), and garlic (31). However, unlike previous studies, we found no statistically significant decreased risks with consumption of cruciferous vegetables (23, 31) and legumes (24, 31, 40, 42, 43).
Our results were strongest for intake of total fruit, and previous research on fiber and colorectal adenoma showed that intake of fruit fiber, in particular, may be protective for colorectal adenoma. A previous cohort study found no association between total fiber intake and colorectal cancer, but did observe a borderline statistically significant decreased risk of colorectal cancer with increasing intake of fruit fiber (62). A statistically significant effect of high versus low intake of fruit fiber on colorectal cancer risk was also observed in a different cohort study (63). In the PLCO trial (64), a statistically significant inverse association was observed between colorectal adenoma and fiber from fruit and fiber from grains and cereals but not fiber from legumes or vegetables.
The association we observed between fruit intake and adenoma could not be fully explained by intake of dietary folate or fiber, both of which are thought to protect against colorectal cancer (65, 66). Perhaps, a combination of the nonfiber phytochemicals and soluble fiber like pectin found in fruit (1) may be protective against colorectal cancer by slowing glucose absorption and thus promoting better blood glucose control and reducing hyperinsulinemia (1, 67–70) or via production of short-chain fatty acids that have been show to reduce cancer cell growth in vitro (1, 71).
Our study had limitations. Our results are not generalizable to all cases of colorectal adenoma because our cases and controls were defined on the basis of pathology in the distal colon and rectum; participants with adenomatous polyps in the proximal colon only would have been considered controls. Additionally, our cases with distal adenomas include subjects with advanced, nonadvanced, single, and multiple adenomas, but further analysis showed that our results did not vary greatly when they were stratified by histologic adenoma type, number, or location. Additionally, FFQs are subject to measurement error (61), which can attenuate the risk estimates (61, 72, 73), and dietary patterns change over time, and diet was assessed at only one time point. It is also possible, as in all observational studies, that residual confounding exists, and our multivariate models did not account for all confounding.
The strengths of our study include its large number of cases (n = 3707) of distal colorectal adenoma identified by a standardized flexible sigmoidoscopy screening protocol, characterized by clinical follow-up with histologic confirmation. Comparison controls for this study were also endoscopically determined to by polyp-free. Sigmoidoscopy also confirmed that the controls also had no distal adenomatous or villous polyps. Participants were from diverse regions in the United States and reported a wide range of fruit and vegetable intake, which included both the national average of fruit and vegetable consumption (4.8 servings/d for persons ages 2 y and older in 2001; 47) and the amount recommended in the 2005 Dietary Guidelines for the average combined intake of fruit and vegetables (10 servings/d; 48).
Overall, this study showed that diets high in fruit, dark-green vegetables, deep-yellow vegetables, and onions and garlic are associated with decreased risk of colorectal adenoma. In this population, the observed decreased risk of adenoma with increasing intake of fruit was not completely explained by the folate or fiber content of the fruit. We conclude, from the results of this screening trial, that to reduce the risk of development of colorectal adenoma, persons should continue to consume diets rich in fruit and vegetables, which contain many potential anticarcinogenic compounds.
ACKNOWLEDGMENTS
We thank Matthew Moore of Information Management Services Inc, Silver Spring, MD, for his support of analysis runs.
The contributions of the authors were as follows—AEM: conducted the analyses, interpreted the work, and was the primary writer of this manuscript; AFS: was involved in the primary design, analyses, writing, and interpretation of this work; BIG: contributed to the statistical analysis, writing, and interpretation of this work; UP: contributed to the analysis, writing, and interpretation of this work; RBH: was involved in the conception, design, analysis, writing, and interpretation of this work; JLW: was involved in the conception, design, writing, and interpretation of this work; LAY: was involved in the conception, design, writing, and interpretation of this work; RGZ: contributed to the conception, design, analysis, writing, and interpretation of this work. None of the authors had a conflict of interest to report.
REFERENCES