Background: Fat and total calories are closely correlated in the diet. Because of this correlation, studies of dietary fat and breast and colorectal cancer incidence have been peculiarly difficult to interpret.
Methods: This article reviews the information from rodent models of breast and colorectal cancer and epidemiological studies addressing the separate effects of calories and fat in the diet. Various statistical approaches described for separating the effects of calories and fat are outlined, with a review of epidemiological case-control, cohort, and prospective controlled clinical studies of diet and breast and colorectal neoplasia.
Results: Animal studies provide convincing evidence for a dominant effect of calorie restriction over fat modification in tumorigenesis. Individual case-control studies show no effect or are inconsistent in distinguishing calories from fat in colorectal carcinogenesis, and are largely null in breast cancer. However, a meta-analysis of these studies in breast cancer defines a relative risk (P < 0.05) of 1.40 for calories "solely" due to fat (1.48; P < 0.05). Cohort studies for colorectal cancer have reported variable results, but a recent large study of women implicates animal fats, particularly from red meat, as a risk for colon cancer, independently of calories. For breast cancer, a marginally significant effect of fat independent of calories was found in one recent cohort study, but earlier large cohort studies could not identify calories or fat as risk factors.
Conclusion: Caloric restriction is likely to be important in inhibiting tumorigenesis based on animal studies. However, the influence of animal fats is seen both in animal studies and epidemiological studies of breast and bowel cancer. The differential effect of fats from different dietary sources in both colon and breast cancer points to a qualitative effect separate from calories in tumorigenesis. Extrapolation of significant analyses of trend to very low-fat diets should be interpreted cautiously without further information on concurrent nutrient deficiencies and competing illnesses. Randomized controlled trials of dietary intervention to prevent neoplasia are feasible and will enhance the quality of information available about diet and tumorigenesis.