Abstract

BACKGROUND: In 2007, the WCRF published a systematic review with information about a large amount of life style variables in relation to a large variety of cancer types (World Cancer Research Fund). The WCRF concluded that there was a) limited evidence suggesting that non-starchy vegetables, or fruits protect against colorectal cancer, and b) garlic probably protects against colorectal cancer.
OBJECTIVES:

• To review all prospective studies which provided data about total fruits or vegetables, their subgroups, and specific types of vegetables and fruits, in relation with colon-, rectal-, and total colorectal cancer risk, disease progression and mortality/survival.
• To define the amount of consumption found to be related with the described effects on cancer.
• To define possible effect modification by confounders.

DATA SOURCE: The Pubmed database was searched (No start date - Sep 11, 2009) for relevant articles using the keywords "cancer, neoplasm, or neoplasms" combined with "vegetables, vegetable, fruit, or fruits" and a fair amount of other keywords. The exact search term is described in the methods.
Prospective studies published in the English language were included. Reference lists were searched for additional articles.
RESULTS: 51 articles were found which provided information about 32 different cohorts. Of these, 0 articles were excluded.
Results are described when any evidence for an association - as defined in the Methods - was found. In addition, data about total vegetables or fruits is described.

Total colon cancer risk.

• Data about the relation with total vegetables was provided by a pooled analysis of 14 cohorts and 6 additional cohorts. No evidence was found for an association between total vegetables and total colon cancer risk (RR = 0.94). Except for suggestive evidence for a protective effect among men at an intermediate level of consumption (mean 336 intake g/day).
• Data about the relation with total fruit was provided by a pooled analysis of 14 cohorts and 6 additional cohorts. Total fruit probably protects against total colon cancer risk. This evidence was restricted to women (RR = 0.84). Protective effects were found at the level of consumption of 100- < 300 g/day and ≥ 400 g/day.
• Bananas possibly protect against total colon cancer risk at the level of consumption of ≥ 1 banana/week. The effect size was strongest at 1 banana/wk to < 0.5 banana/day (RR = 0.84). Suggestive evidence for a protective effect was found with lower consumption.
• Spinach possibly protects against total colon cancer risk at the level of consumption of ≥ 1 serving (= 1/2 cup)/week (RR = 0.89). Suggestive evidence for a protective effect was found with lower consumption.
• Prunes possibly increase total colon cancer risk (RR = 1.57). No level of consumption could be defined for the association.
• The following vegetable items possibly protect against total colon cancer risk at intermediate (but not high) levels of consumption:
  -peas/lima beans at > 0 to < 1 serving (= 1/2 cup)/wk (RR = 0.89).
  -broccoli at > 0 to < 1 serving (= 1/2 cup)/wk (RR = 0.91).
  -tomatoes/tomato juice at 1 tomato week to < 0.5 tomato/day (RR = 0.88).
  -Brussels sprouts at > 0 to < 1 serving (= 1/2 cup)/wk (RR = 0.91).
• Suggestive evidence was found for a protective effect of total soy foods (but not tofu, or soybean paste soup) when all cancer end points were combined (RR = 0.92).

Distal colon cancer risk.

• High vs low vegetable consumption (≥ 200 g/day) possibly protects against distal colon cancer risk among both sexes (RR = 0.83). Probable evidence for a protective effect was found at the level of consumption of 200- < 300 g/day among men.
• Total fruit possibly protects against distal colon cancer risk among women. This effect was found at the level of consumption of ≥ 400 g/day.
• Items from the umbelliferae family possibly protect against distal colon cancer risk (RR = 0.88), but no evidence was found for an association with total colon cancer.

Proximal colon cancer risk.

• No evidence was found for an association between total vegetables and proximal cancer risk (RR = 1.06).
• Total fruit possibly protects against proximal colon cancer risk among women at the level of consumption of 200- < 300 g/day.

Rectal cancer risk.

• Data about the relation with total vegetables was provided by 9 cohort studies. No evidence was found for an association between total vegetables and rectal cancer risk (RR = 1.01).
• Data about the relation with total fruits was provided by 9 cohort studies. No evidence was found for an association between total fruit and rectal cancer risk (RR = 0.96).

Advanced stage/metastatic disease risk or disease progression.
No data was found about the relation with advanced/metastatic disease or disease progression.
Mortality risk.

• Data about the relation between total vegetables and colon cancer mortality was provided by 2 cohorts. No evidence was found for an association.
• Data about the relation between total fruits and colon, rectal, and colorectal cancer mortality was provided by 6 cohorts. No evidence was found for an association.

CONCLUSION: Among women, total fruit possibly protects against distal, proximal and total colon cancer risk at different levels of consumption. For total colon cancer, this effect was found at the level of consumption of both 100- < 300 g/day, and ≥ 400 g/day. Among men, suggestive evidence was found for a protective effect of total vegetables against total colon cancer risk at an intermediate level of consumption (median of ranges: 336 g/day). For both sexes combined, high vs low vegetables consumption (≥ 200 g/day) possibly protects against distal colon cancer risk. But probable evidence for a protective effect was found at the level of consumption of 200- < 300 g/day among men. No other associations of total fruits or vegetables were found with any cancer end point.
High vs low consumption of bananas (≥ 1 banana/wk) and spinach (≥ 1 serving/wk) possibly protect against total colon cancer risk. Suggestive evidence of a protective effect was found for lower levels of consumption of both items. Suggestive evidence was found of a protective effect of total soy foods with cancer risk, but results were confined to research among Asian populations. Prunes possibly increase total colon cancer risk. The following vegetable items possibly protect against total colon cancer risk at intermediate levels of consumption: peas/lima beans, broccoli, and Brussels sprouts (> 0 to < 1 serving/wk), and tomatoes/tomato juice (1 tomato week to < 0.5 tomato/day).
No evidence was found for any association with rectal cancer, and no evidence was found for effect modification - other than gender - by any of the possible confounders.
LIMITATIONS: None of the studies provided information about advanced stage/metastatic disease or disease progression. Few studies examined the relation between vegetables or fruits and colorectal cancer mortality, and findings were inconsistent. Moreover, no data is available about dietary consumption following the diagnosis of colon or rectal cancer.
REFERENCE:
World Cancer Research Fund / American Institute for Cancer Research. Food, Nutrition, Physical Activity, and the Prevention of Cancer: a Global Perspective. Washington DC: AICR, 2007. Full text

Total vegetables and total colon cancer risk.

Data about the relation between total vegetables and total colon cancer risk was provided by a pooled analysis of 14 cohorts and 6 additional cohorts, including a total of > 9,847 cases (the amount of cases was not defined in one cohort of very large size: Park Y [27]).
A significant protective effect among men, but not women, was found in one cohort of very large size (Nomura AM [30]). No other (non)significant associations were found. The average RR = 0.94 (excluding incomplete data from Park Y [27]).

Inclusion of intermediate levels of consumption:
Among men, (non)significant protective effects were found in the 2nd-5th quintile of consumption (median 212 - median 539 g/d) in one cohort of very large size (Park Y). And a significant protective effect was found in the 3rd quintile of consumption (median 336 g/d) in another cohort of very large size (Nomura AM).
Among women, no (non)significant effects were found at any level of consumption in any cohort.

Effect modification:

• No effect modification was found by age at diagnosis (29), follow-up period (29), multivitamin supplement use (25, 29), having had a sigmoidoscopy or colonoscopy, or a polyp detected (6), HRT use (7), alcohol consumption (15, 25), meat consumption (18, 25), and physical activity (25).
• Most (6, 7, 25), but not all (18) cohorts showed no effect modification by smoking status.
• Most (6), but not all (8) cohorts showed no effect modification by family history of colorectal cancer.
• Most (15, 25), but not all (18) cohorts showed no effect modification by BMI.

Conclusion: Few effects were seen. No evidence was found for an association between total vegetables and total colon cancer risk. But among men, a protective effect at an intermediate levels of consumption is suggested (median intake 336 g/day). Lack of specific data about amounts of cases prevents indication of the level of evidence, but this evidence probably is suggestive. No evidence was found for effect modification.

Total vegetables and distal colon cancer risk.

Data about the relation between total vegetables and distal colon cancer risk was provided by a pooled analysis of 14 cohorts, and two additional cohorts, which were both of very large size.
A significant protective effect was found in the pooled analysis (Koushik A [29]) including 756,217 subjects: RR = 0.82 (0.68-0.98; P = 0.01). Stratified by sex, the strength of the associaton did not change, but the trend became nonsignificant among both men and women. A significant protective effect among men, but not women was found in another cohort (Park Y [27]). No association was found in the remaining cohort (van Duijnhoven FJ [18]).
The average RR = 0.83 (excluding incomplete data from Park Y [27]).

Inclusion of intermediate levels of consumption:
Among men, a strong significant protective effect was found at the level of 200- < 300 g/day in the pooled analysis (Koushik A [29]). In addition, significant protective effects were found at 2 different levels of consumption (median 270, and median 540 g/day) in a cohort of very large size (Park Y).
Among women, also a significant protective effect was found at the consumption level of 200- < 300 g/day in the pooled analysis, but in the remainig cohort, no associations were found at any level of consumption.

Conclusion: A significant protective effect of high vs low consumption (≥ 200 g/day) was found in a pooled analysis, but the trend became nonsignificant when stratified by sex. In addition, a significant protective effect of high vs low consumption was also found in cohort of very large size, but this effect was restricted to men only. High vs low vegetable consumption possibly protects against distal colon cancer risk among both sexes.
A significant protective effect of an intermediate level of consumption (200-< 300 g/day) was found among both men and women, which remained significant when stratified by sex. Among men, this effect was strengthened by results from one cohort of very large size, in which a significant protective effect was found at this level of consumption. Total vegetables probably protect against distal colon cancer risk among men at the level of 200-< 300 g/day.

Total vegetables and proximal colon cancer risk.

Data about the relation between total vegetables and proximal colon cancer risk was provided by a pooled analysis of 14 cohorts, and two additional cohorts, which were both of very large size. No associations were found.
The average RR = 1.06 (excluding incomplete data from Park Y [27]).

Inclusion of intermediate levels of consumption:
Among men, a significant protective effect was found in the 3rd quintile of consumption (median intake 270 g/day) in one cohort of very large size (Park Y). No other associations were found. Clearly, no consistent level of consumption was found for any effect in both sexes:

Conclusion: No associations were found. No evidence was found for an association between total vegetables and proximal cancer risk.

Total vegetables and rectal cancer risk.

Data about the relation between total vegetables and rectal cancer risk was provided by 9 cohort studies, including 2,524 + X cases (no data about the amount of cases was provided by one cohort (Park Y [27]).
A nonsignificantly increased risk among women (trend), but not men was found in one cohort (Voorrips LE [9]). No other associations were found. The average RR = 1.01 (excluding incomplete data from Park Y [27]).

Inclusion of intermediate levels of consumption:
Though in one cohort, the trend increased nonsignificantly, no (non)significant effects were found at any level of consumption among both men and women:

Effect modification:

• No effect modification was found by alcohol consumption (15, 18, 25), meat consumption (18, 25), physical activity (25), and vitamin supplement use (25).
• One (18), but not the other (25) cohort showed an increased risk among current smokers only.
• Most (15, 25), but not all (18) cohorts showed no effect modification by BMI.

Conclusion: Few associations were found. No evidence was found for an association between total vegetables and rectal cancer risk. No evidence was found for effect modification.

Total vegetables and total colorectal cancer risk.

Primary variables of interest were colon cancer, or rectal cancer. But results from articles about some cohorts were not stratified by cancer site (colon + rectal), and only show associations with total colorectal cancer risk.
The association with total colorectal cancer risk is not discussed, but findings are presented in the following table: Click.

Total fruit and total colon cancer risk.

Data about the relation between total fruit and total colon cancer risk was provided by a pooled analysis of 14 cohorts (Koushik A [29]) and 6 additional cohorts, including 9,794 + X cases (the amount of cases was not defined in one cohort of very large size: Park Y [27]).

A significant protective effect was found in the pooled analysis (RR = 0.87 [0.77-0.97; P = 0.04] for men & women combined), in one additional cohort of small size among women only (Shibata A [1]), and in one cohort of very large size among men only (Nomura AM [30]), though the trend was not significant in the latter cohort. These analysis included 6.631 cases (< 68% of all cases). In addition, a nonsignificant protective effect was found in one cohort of very large size (van Duijnhoven FJ [18]) including a total of 1,828 cases (< 19% of all cases).
The average RR = 0.87 (excluding incomplete data from Park Y [27]). Stratified by gender, the RR's were 0.90 for men, and 0.84 for women; but one cohort could not be included in this analysis, because no amount of cases was provided, stratified by gender (Sato Y [24]).
Stratified by gender, protective effects from the pooled analysis were restricted to women, and the trend became nonsignificant.

Inclusion of intermediate levels of consumption:
Among women, significant protective effects were found at different levels of consumption (100- < 300 g/day, and ≥ 400 g/dag) in the pooled analysis (Koushik A [29]). In addition, a significant protective effect was found at the level of consumption of > 246 g/day in one additional cohort of small size (Shibata A [1]).
Among men, a significant protective effect was found in one cohort at 740 g/day (Nomura AM [30]):

Effect modification:

• No effect modification was found by age at diagnosis (29), follow-up period (29), BMI (15, 18, 25), multivitamin supplement use (25, 29), family history of colorectal cancer (6, 8), having had a sigmoidoscopy or colonoscopy, or a polyp detected (6), HRT therapy (7), alcohol consumption (15, 18, 25), and physical activity (25).
• Most (7, 18, 25), but not all (6) cohorts showed no effect modification by smoking status.
• One (25), but not the other (18) cohort showed no effect modification by meat consumption.

Conclusion: A significant protective effect was found in a pooled analysis of 14 cohorts, and 2 additional cohorts, though effects were restricted to one gender in the latter 2 cohorts. A protective effect was found in one cohort of very large size, but this effect was nonsignificant. Total fruit possibly protects against total colon cancer risk (- 13%). The evidence was restricted to women, though the average effect size does not exclude the possibility of an effect among men. Protective effects among women were found at the level of consumption of 100- < 300 g/day and ≥ 400 g/day. No evidence was found for effect modification.

Total fruit and distal colon cancer risk.

Data about the relation between total fruit and distal colon cancer risk was provided by a pooled analysis of 14 cohorts (Koushik A [29]), and two additional cohorts, which were both of very large size.
A significant protective effect among women, but not men, was found in the pooled analysis (12 cohorts for women). In addition, a nonsignificant protective effect was found in one cohort, but results were not stratified by gender (van Duijnhoven FJ [18]). No association was found in the remaining cohort (Park Y [27]).
The average RR = 0.80, but since data from one cohort (Park Y [27]) with RRs of 0.97, and 1.14 could not be included - because no amounts of cases were provided - the true effect size will be smaller.

Inclusion of intermediate levels of consumption:
A significant protective effect was found in the pooled analysis at ≥ 400 g/day among women only (Koushik A [29]). No other (non)significant effects were found at any level of consumption.

Conclusion: A significant protective effect among women was found in a pooled analysis of 12 cohorts, but no significant effects were found in any of the cohorts of very large size. Total fruit possibly protects against distal colon cancer risk among women. This effect was found at the level of consumption of ≥ 400 g/day.

Total fruit and proximal colon cancer risk.

Data about the relation between total fruit and proximal colon cancer risk was provided by a pooled analysis of 14 cohorts, and two additional cohorts, which were both of very large size. No (non)significant associations were found in any cohort.
The average RR = 0.88, but since data from one cohort (Park Y [27]) with RRs of 1.04, and 0.99 could not be included - because no amounts of cases were provided - the true effect size will be smaller.

Inclusion of intermediate levels of consumption:
A significant protective effect at an intermediate level of consumption (200- < 300 g/day) was found among women in the pooled analysis (12 cohorts for women). No other associations were found among men or women.

Conclusion: A significant protective effect at an intermediate level of fruit consumption was found among women in a pooled analysis of 12 cohorts. No other associations were found. Total fruit possibly protects against proximal colon cancer risk among women at the level of consumption of 200- < 300 g/day.

Total fruit and rectal cancer risk.

Data about the relation between total fruits and rectal cancer risk was provided by 9 cohort studies, including 2,542 + X cases (no amount of cases was provided by one cohort (Park Y [27]).
The average RR = 0.96 (excluding incomplete data from Park Y [27]).
Women: A significant protective effect was found in one cohort examining women only (Terry P [15]), but a significantly increased risk was found in one other cohort among women, but not men (Park Y [27]). No other associations were found.
Men: No associations were found.

Inclusion of intermediate levels of consumption:
Among men, significant protective effects were found in one cohort (Nomura AM [30]) at the 3rd and 4th quintile of consumption (median intakes 298, and 450 g/day).
Among women, significant protective effects were found in one cohort (Nomura AM [30] at the 2nd quintile of consumption (median intake 217 g/d), and in another cohort at > 160 g/day (Terry P [15]). But a significantly increased risk was found in a third cohort (Park Y [27]) at the 4th and 5th quintile of consumption (median intakes 368, and 560 g/day):

Effect modification:

• No effect modification was found by alcohol consumption (15, 18, 25), BMI (15, 18, 25), smoking status (18, 25), physical activity (25), and vitamin supplement use (25).
• One (25), but not the other (18) cohort showed no effect modification by meat consumption.

Conclusion: Inconsistent findings were done among women only, but no evidence was found for an association between total fruit and rectal cancer risk. No evidence was found for effect modification.

Total fruits and total colorectal cancer risk.

Primary variables of interest were colon cancer, or rectal cancer. But results from articles about some cohorts were not stratified by cancer site (colon + rectal), and only show associations with total colorectal cancer risk.
The association with total colorectal cancer risk is not discussed, but findings are presented in the following table: Click.

Colorectal cancer mortality.

Data about cancer mortality is reviewed. Total fruits or total vegetables consumption in relation to total-, distal-, or proximal colon, rectal cancer, and total colorectal cancer mortality is discussed. Data about cancer mortality is added to the related tables of cancer risk on this page (extended tables).

1) Total vegetables:
Data about total vegetables in relation to colon cancer mortality was provided by 2 cohorts. A significant protective effect was found among women, but not men, in the largest cohort (7). No significant association with colon cancer + total colorectal cancer mortality was found in the other cohort (12). No other data was found.
2) Total fruits:
Colon cancer: Data about total fruits in relation to colon cancer mortality was provided by 2 cohorts. A significantly increased risk among women, but not men was found in one cohort (22). A significant trend of an increased risk was also found in the other cohort, examining men only (12).
Rectal cancer: Data about total fruits in relation to rectal cancer mortality was provided by 1 cohort (22). No association was found among men or women.
Colorectal cancer: Data about total fruits in relation to total colorectal cancer mortality was provided by 4 cohorts (2, 10, 19, 21). No associations were found.

Conclusion: Few associations were seen among cohorts. No evidence was found for an association between total fruits or total vegetables and colorectal cancer mortality.

Advanced stage/metastatic disease risk.

No data about the relation between fruits or vegetables and advanced stage and/or metastatic disease risk is provided by any article about any of the cohorts.