The expression “colorectal cancer” refers nonspecifically to three cancers with distinctive epidemiologic and etiologic features. Failure to distinguish among these is the source of much confusion in the literature. Weisburger reviewed the epidemiology, causes, mechanisms, and prevention of cancers of the proximal and distal colon and rectum (38). He cited findings from animal studies, geographical epidemiology, studies of migrants and subpopulations that adhere to distinctive traditions, and time trend. The roles f activity, iatrogenic factors, associated diseases, sex, and genetic elements were considered. Genotoxic and nongenotoxic mechanisms of carcinogenesis were explained and protective factors were discussed.
The evolution of colorectal cancer from normal mucosa is driven multifactorially (40). In heritance and diet are critical; physical activity, alcohol consumption, and body composition are also involved. Dietary influences come into play at each stage of the multistage carcinogenic process, but their effect may depend on genetic factors. Definitive understanding of the relationships between diet, inheritance, and colorectal cancer await the results of large prospective, randomized intervention studies such as have recently been initiated in Europe and the USA’ Giovannucci et at.Reported one of the few studies of the relationship of diet to risk of colorectal adenomas, which are early precursors of cancer (41). Using data from food-frequency questionnaires completed by 7284 male health professionals who had had a colonoscopy or sigmoidoscopy, they determined relative risk of adenoma by quintile if nutrient intake. After adjustment for total energy, risk of colorectal adenoma was positively associated with intake of saturated fat and with the ratio of red meat to fish and chicken intakes, and inversely associated with intake of all sources of dietary fiber. Because adenomatous polyps frequently appear 10 years or more before the clinical diagnosis of cancer, this study links dietary factors with events very early in the transition from normal mucosa to malignancy.
Disagreements over the relative roles of different dietary factors in development of colon cancer frequently originate in the high correlations between nutrients and the ways in which researchers try to compensate for these correlations. (42). This paper described and applied as set of diagnostic tools for assessing the magnitude of collinearity in epidemiologic data, reviewed four methods for addressingcollinearity problems in multivariate egression models, and evaluated the results from application of each method data from a case control study of diet and colon cancer. Although the study addressed only the potential collinearity problems between total fat and total caloric intake, similar difficulties would arise for total protein and total caloric intake. The fundamental problem with multicollinear nutrient data is that they contain insufficient information to discriminate the in fluence of highly correlated nutrients. Therefore the epidemiologist must impose constraints on the data to address analytical problems associated with collineariy; how to do this is a critical element of the procedures described in this paper.
Another problem in comparing case-control studies of colorectal cancer relates to the type of dietary information used-retrospective versus original (43). To evaluate the seriousness of this, participants in a cohort study who had dilled out a dietary questionnaire in 1967 (original information) were asked to fill out a similar questionnaire in 1987, with reference to their dietary habits in 1967 (retrospective information). All 50 surviving subjects who had had a diagnosis of colorectal cancer completed the questionnaire in 1987, as did 150 controls selected at random from original participants who did not have colorectal cancer. Case- subjects and controls had a similar tendency to misestimate their previous food intake, and this tendency was closely related to changes in consumption during the intervening years. Subjects whose dietary habits had not changed appreciably provided retrospective information that agreed well with the original. However, compared with controls, case-subjects tended to underestimate their previous consumption of coffee and sweets and overestimate their consumption of eggs, fruits, and vegetables when their original estimate had been “high” When the original estimate was “low,” case-subjects more frequently overestimated previous consumption of fruits and vegetables, whereas controls overestimated previous consumption of potatoes and milk. When changes in dietary habits were analyzed, case-subjects and controls sowed similar changes.
A case-control study of the relationship of diet to occurrence of colorectal cancer in Athens, Greece, characterized a high-risk diet as high in protein, saturated fat, and cholesterol, and low in vegetables- the opposite of a low-risk diet (44). However, case-subjects had substantially and highly significantly lower serum levels of total cholesterol and high-density lipoprotein (HDL) cholesterol than controls. Thus, again, diet could not explain serum cholesterol values (see 32). Alternative explanations for this well known paradox are discussed in terms of the data from this study.
A case- control study in Sweden found a positive association between colorectal cancer and usual meat consumption during the previous five years (45). High total meat intake, frequent consumption of brown gravy, and preference for a heavily browned meat surface each independently increased the risk; and the risk was greater for rectal than for colon cancer. Surprisingly, consumption of boiled meat was associated with greater risk than consumption of meat fried lightly. Although it has been known for some time that carcinogens are formed when meat I cooked at high temperatures, this is one of the few epidemiologic studies to consider the method of cooking.
In a prospective study of nearly 90,000 women, Willett et al. found an association between intake of red meat and colon cancer (see Chapter 2, Food Safety 1991, reference 27). Numerous readers have commented on this study (46-53). It was noted that the authors failed to distinguish between cancers of the proximal and distal colon, which are thought to have different pathogenetic mechanisms and whose relative incidence in women varies with age (46). Nor was the possibility of hereditary nonpolyposis colorectal cancer considered in the data analysis. In lumping beef, pork, lamb, and processed meat products such as bacon and sausage into the category “red meat” the authors failed to account for the huge range of fat content and possible role of other constituents, especially in processed meats (47). They also did not discuss other aspects of “healthy” and “unhealthy” lifestyles that tend to be associated with healthy and unhealthy diets (48). A practitioner of family and sports medicine was concerned that active women and girls already at risk for iron deficiency might turn away from the best source of iron available to them, when the only significant risk was associated with consumption of more than a quarter of a pound of red meat daily (49). Estimates of dietary fiber used in this study may have been less than ideal and could have had a major effect on conclusions from the study (50). The possibility of a detection bias was suggested (51). High meat consumption would cause more positive tests for occult blood in the stool, and investigation of these could yield some incidental diagnoses of colon cancer. Willett et al. explained why this was unlikely in their study (54). In response to another comment (52), they explained why they did not analyze total mortality in their relatively young population. Finally, a reader looked at the 2:1 risk ration for colon cancer in meat eaters compared with fish eaters from a different perspective (53). Meat eaters had an incidence of colon cancer of 0.113 %, whereas the incidence in fish eaters was 0.059% difference in these rates is seriously worth worrying about. Willett et al. argued that it is (54), considering the young age of the study population and the short duration of follow-up.
A case-control study of diet and colon cancer was conducted in Argentina, where the usual diet is high in red meat (55). One hundred and ten casepatients were matched on age and sex with 220 neighborhood controls. Analysis of information on food consumption during the 5- year period ending 6 months before interview showed a highly protective effect of dietary fiber and some evidence for a protective effect of niacin. Total energy consumption increased relative risk by a factor of 1.84 per 1000 calories.
In a hospital- based study in northeastern Italy, subjects completed a questionnaire giving the frequency of consumption of specific food items before onset of the problem leading to the current hospitalization (56). Analysis of data for 123 cases of colon cancer, 125 cases of rectal cancer, and 699 controls revealed a protective effect for fiber- and vegetable-rich diets. Frequent consumption of refined starchy foods, eggs, cheese, or red meat was a risk factor. Some foods had differential effects on the odds ratios for colon and rectal cancer. For most subjects, a substantial proportion of dietary fiber was supplied by vegetables. Although the study could not determine whether risk was modified by fiber itself, a fiber component, a particular fiber type, or associated micronutrients, failure to find an association with fruit intake seems to argue against a protective role for micrountrients such as vitamin C and β- carotene. Refined carbohydrate products, a risk factor in this study, comprised a larger fraction of the diet than they do in the USA. Somewhat different results were obtained from a hospitalbased case-control study in China, where the usual diet is quite unlike the usual Italian diet (57). Here, increased risk of colorectal cancer was associated with low intake of green vegetables, chives and celery, meat, eggs, grain, and soybean products. Results were similar but not identical for men and women and for colon and rectal cancer. The findings for meat were interesting in this population whose meat consumption is generally low. The greater risk associated with low meat consumption suggests that there is an optimal amount of meat, greater than zero, in a healthy diet.
A dietary study of hospitalized colorectal cancer patients in Majorca, with age-and sex-matched hospital and population controls, found that risk of colorectal cancer increases with increasing intake of total calories and cholesterol and decreases with increasing intake of pulses and folic acid (58). Of the three main components of total energy, protein (mainly animal) and carbohydrate were associated with risk; no significant effect of lipids or saturated fats was found.
Although adenomatous polyps are precursors of colorectal cancer, it is not known what proportion of colorectal cancers originate from such polyps or what proportion of polyps become cancerous. Furthermore, little is known about the etiology of adenomatous polyps beyond the fact that there is a genetic predisposition to them. An Australian case-control study tested the hypothesis is that diet and alcohol consumption play a role in development of adenomatous colorectal polyps and that dietary risk factors are similar to those for colorectal cancer (59). Case and control groups had similar family histories of colorectal cancer. Subjects with adenomatous polyps had a significantly lower intake of fiber and vegetables than controls; men with polyps were also characterized by a higher intake of beef, milk, and beer. Results were less clear in a study of exercise, diet, and adenomatous polyps in Japanese self-defense officials (60). The men, mostly in their early 50s,were given a comprehensive retirement health examination (including a colonoscopy) and a self- administered questionnaire addressing dietary and other habits. Leisure time spent doing strenuous activities was inversely related to the risk of polyps, but dietary relationships were merely suggestive. A nonsignificant decrease in risk was associated with frequent consumption of rice, green tea, and instant coffee. No association was see for fruits, vegetables, meat, fish, or several other food items, In a Swedish study, low physical activity at work and in leisure time was associated with increased risk for colon but not rectal cancer (61). This study also found that frequent intake of meat, brown gravy, heavily browned fried meat, fish, and eggs was associated with increased colorectal cancer risk, as was high intake of fat or protein. A high intake of dietary fiber decreased risk at most colorectal sites.
A study in New York City of diet and the recurrence of colorectal adenomatous polyps found no associations for men (62).For women, however, significantly increased odds ratios were fund for high versus low quartiles of beef, cheese, total fat, and saturated fat consumption. Decreased odds ratios were associated with high fish, carbohydrate, and fiber consumption. Marginal effects were seen for total energy and vitamin A intake. Two ongoing studies are sharpening the focus on fiber’s role on the recurrence of adenomatous polyps (63). In one patients are taught a low-fat, high-fiber, high-fruit, high-vegetable dietary pattern after having polyps removed. In the other, similar subjects are given a high-fiber supplement but otherwise retain their customary dietary pattern. Together these studies should indicate whether the key is fiber or another dietary component or components. The response of men and women to modified diets may differ (64). After polypectomy, 201 subjects in Toronto were randomized to diet-counseling and control groups. After 12 months, fat consumption averaged 25% of energy in the counsel group and 33% in controls; fiber consumption was 35 and 16 g, respectively. Women who received counseling had a reduced rate of polyp recurrence, associated with reduced levels of serum cholesterol and fecal bile acids. Men, however, had an increased rate of recurrence, associated with decreased serum cholesterol but increased fecal bile acids.
Although a relationship between dietary fiber and development of cancers of the gastrointestinal tract seems fairly certain, the mechanisms and inter mediates involved are not well understood. Klurfeld reviewed carcinogenetic mechanisms mediated by dietary fiber (65). The role of fecal bile acids has been studied most extensively, but incongruities in human studies have not been resolved. Bacterial degradation of dietary fiber to short-chain fatty acids seems important, but the most readily fermentable fibers are no more protective than fibers refractive to fermentation. Promising topics for study include physical characteristics of the feces, aqueous-phase bile acids, alterations in mucins, mutagenicity of intestinal contents, changes in mucosal cytokinetics, enzyme activity (e.g., ornithine decarboxylase and aryl hydrocarbon hydroxylase). Neurogenic effects of bulk and short-chain fatty acids, local and systemic gut hormones and growth factors, transit time, pH , and availability of energy.
A large case-control study of dietary fiber and colorectal cancer conducted at SUNY – Buffalo added to the evidence for sex and site differences in response to fiber components (66). High intakes of total vegetable fiber, cellulose, hemicelluloses, and lignin were associated with increased risk of sigmoid colon cancer in women, but there was no association with risk at other subsites in women or at any site in men. The various fiver components appeared to have somewhat different effects in women, but strong correlations between the components made it difficult to analyze these results. This study could not draw any conclusions about cereal fiber, but a Swedish study did (67). A dietary survey of 41 colorectal cancer patients, 41 matched controls, and 801 hospital and population controls found that, per unit of energy, the habitual diet of cancer patients contained less cereal fiber (p<.001) as well as less riboflavin, calcium, and phosphorus.
Overall, cancer patients consumed more fat, protein, carbohydrate, and total energy and less cereal fiber than matched controls, but these differences were not statistically significant. High alcohol consumption was correlated with increased risk (p<.05). High intakes of calcium or cereal fiber, relative to energy, were associated with reduced risk of colon cancer; a similar association was found for intake of total fiber or cereal fiber and risk of rectal cancer.
One factor in the positive correlation between latitude and death rate from colon cancer may be intensity and duration of exposure to sunlight, with consequent effects on vitamin D and calcium metabolism. To investigate this, lavoratory, clinical, and epidemiologic studies were reviewed (68). Patterns of fat, fiber, fruit, and vegetable consumption do not readily explain the geographic pattern of colon- cancer mortality in the USA. However, at high latitudes, serum concentrations of vitamin D metabolites tend to be low and absorption of calcium may be poorer, especially in the elderly. Further, in animal studies, dietary supplementation with calcium and (or) vitamin D partly protects against colon tumors induced by a high-fat diet. All lines of evidence suggest to the authors that substantial reductions in rates of colon cancer could be achieved in many parts of the world by optimizing intake of calcium and vitamin D. The usual cautions regarding vitamin D toxicity are still warranted, and additional precautions are necessary for persons at risk for certain types of renal stones. Safe and effective ranges of calcium and vitamin D consumption for specific populations in the context of geography and culture need to be established worldwide.
A carefully reasoned review of admittedly meager data concluded that intestinal exposure to ingested iron may be an important determinant of colorectal cancer risk in humans (69). Iron has been associated with both initiation and promotion of tumors. Differences in intake of iron and promotion of tumors. Differences in intake of iron and substances such as phytate that affect its bioavailability may explain differences in colon-cancer risk between high – and low-risk countries and within the USA. The review touches on iron’s role in nutrition and metabolism; the complex system of iron absorption, transport, and storage that maintains adequate supplies while protecting against toxicity; the global nonproblem of iron deficiency if the root causes of anemia are controlled; the consequences of iron overload; and the rationale and experimental evidence for iron’s role in colorectal cancer.
The relationship between folate intake and risk of colorectal cancer was assessed in a case –control study at SUNY – Buffalo (70). Again, a differential effect was seen according to site. While folate (controlled for total energy intake) was not consistently associated with colon cancer risk, risk of rectal cancer was reduced in the highest category of intake. For men in the highest quartile, the effect was highly significant (p<.001). The authors propose that the folate content of fruits and vegetables partly explains their protective effect. Folate has not been systematically studied in this regard, and in light of the depressed folate status in some segments of the U.S. population such investigations appear warranted.
Observations that ownership or use of a refrigerator reduces risk of gastric cancer were not initially extended to cancer at other sites. Another study form SUNY – Buffalo measured the magnitude of the “refrigerator effect” on cancers at other gastrointestinal sites (71). Case- patients were matched with age, sex-, and neighborhood-matched controls and interviewed regarding their usual diet, smoking and drinking history, and years of refrigerator ownership. No effects were found for cancer of the mouth or esophagus. For men but not women, refrigerator use was strongly protective against stomach cancer, weakly protective against colon cancer, and possibly weakly protective against real cancer.