Dietary Fat, Calories, and Prostate Cancer Risk

Prostate cancer is a major source of morbidity and mortality. The disease accounts for about a third of all U.S. male cancers, with an estimated 179 300 new cases and 37 000 deaths in 1999 ( 1 ). There is urgent need for a firm scientific foundation from which to launch randomized trials of dietary modification to prevent the incidence and progression of prostate cancer. Ideally, such a foundation should rest on strong and consistent evidence from epidemiologic observations and carefully designed animal experiments. Thus, the article presented by Mukherjee et al. ( 2 ) in this issue of the Journal, evaluating the effects in rodents of restricting intakes of dietary fat and total energy, is of considerable interest. The speculation that cancer risk may be related to dietary intakes of fat or total energy has a long history. Reports of reduced growth in tumors transplanted into energy-restricted rodents date back almost a century [ see Pariza and Boutwell ( 3 ) for a review]. The dietary protocols in these experiments were too crude to distinguish the effects of fat restriction from those of energy restriction and other concomitant nutrient changes. The study designs did not ensure that the experimental groups had equal intakes of other nutrients that might themselves be related to cancer ( 4 ). For example, rats fed ad libitum tend to consume the same amount of energy, regardless of the caloric density of their diets ( 5 ). Thus, compared with rats given a low-fat diet, rats fed a high-fat diet consume less food because fat contributes approximately double the energy contributed by protein or carbohydrate. Consequently, rats fed a high-fat diet consume less vitamins, minerals, protein, and dietary fiber than do rats fed a low-fat diet, unless the high-fat diet is adjusted to ensure equal intakes of these nutrients in the two groups. This adjustment is usually achieved by increasing the amount of these nutrients and reducing the proportion of energy from carbohydrates, so that a high-fat diet contains less carbohydrate than does a low-fat diet. Similarly, adequate studies of the tumorigenic effects of caloric restriction require compensation for the reduced intake of essential nutrients ( 4 , 5 ). In the 1940s, Tannenbaum ( 6 - 8 ) demonstrated that diets high in fat enhanced the development of spontaneous and induced tumors in animal models and that caloric restriction inhibited tumorigenesis. Some of these studies attempted to ensure equal intakes of protein, vitamins, and minerals in the experimental groups. The findings of Tannenbaum have been confirmed for a wide variety of induced and spontaneous tumor models in subsequent work by others ( 4,9,10 ) using adequately adjusted diets. The article by Mukherjee et al. ( 2 ) in this issue of the Journal describes the first carefully designed experiments addressing the relationship between fat and energy intakes and prostate cancer. The investigators transplanted androgen-responsive prostate carcinomas from donor rats to recipient rats and then fed the latter either carbohydrate- or fat-restricted diets. They also injected LNCaP human prostate cancer cells into severe combined immunodeficient (SCID) mice to create LNCaP tumors and then fed the latter different types of diets. The outcomes reported in both sets of experiments were tumor characteristics, such as size, proliferation index, apoptosis (programmed cell death), microvessel density, and vascular endothelial growth factor expression. Mukherjee et al. restricted energy intakes in one of three ways: 1) by reducing energy from fat, keeping intake of all other nutrients essentially equal to that of animals fed ad libitum; 2) by reducing energy from carbohydrate, keeping other nutrients constant; or 3) by reducing total energy from all sources, still keeping other nutrients constant. The investigators found that energy restriction reduced proliferation of both rat and human prostate cancer cells. Importantly, the reduction in tumor growth was similar in all three types of energy-restricted animals. Thus, tumor growth was independent of the proportion of fat in the diet, as long as total energy was restricted. These results are consistent with findings from other animal experiments. Pollard et al. ( 11 ) showed that energy restriction reduced the occurrence of spontaneous tumors of the accessory sex glands in Lobund-Wistar rats. Tumor incidence decreased from 26% in rats fed ad libitum to 6% in restricted rats, and tumors tended to occur later in the restricted rats. Limitations of the study by Pollard et al. include its failure to ensure equal intakes of essential nutrients and uncertainty concerning the origin of the tumors, which may have been the seminal vesicles ( 12 ). Moreover, dietary fat has failed to modify the growth of transplanted rat prostate carcinomas or induction of prostate cancer in rats in all reported experiments that ensured isonutrient intakes ( 13 - 17 ). The results obtained by Mukherjee et al. ( 2 ) indicating a lack of an effect of fat, however, are inconsistent with some other experiments. Pollard and Luckert ( 18 ) reported that adding corn oil to an ad libitum diet without adjustment to ensure equal intakes of other nutrients enhanced tumor induction by carcinogens and androgens. Wang et al. ( 19 ) found that increasing the amount of corn oil at the expense of carbohydrate stimulated the growth of human LNCaP prostate cancer cells in nude mice. However, this experiment was limited by failure to ensure that other nutrients were equivalent in the two groups. Thus, the observed enhancement of fat on prostate tumorigenesis in these two studies may reflect differential nutrient intakes rather than fat intake per se. Alternatively, discrepancies in findings for the effects of fat in the experiments by Mukherjee et al. ( 2 ) and Wang et al. ( 19 ) may reflect other differences between the two studies. In the study by Mukherjee et al., caloric restriction was imposed at the time that the LNCaP cells were transplanted into SCID mice; in contrast, in the study by Wang et al., nude mice were used, the tumors grew more slowly, food was given ad libitum, and dietary intervention was started later. Perhaps sublines of the LNCaP cell line with different growth rates were used. These issues illustrate how small variations in the use of apparently the same model may have a critical influence on the outcome. As with the data from animal experiments, the epidemiologic data concerning dietary fat and total energy intakes are conflicting. Among those studies that gathered dietary data sufficiently comprehensive to allow multivariate assessment of the effects of fat and total energy intakes, two ( 20 , 21 ) found increased risk associated with energy intakes, with little or no relationship between risk and fat intakes, after adjustment for total caloric intake. These studies are supported by the findings of Mukherjee et al. In contrast, two multiethnic case-control studies of prostate cancer ( 22 , 23 ) found that, among men consuming equal fat calories, those with lower total energy intakes (due to lower intakes of carbohydrate or protein) had a cancer risk similar to that of men with higher energy intakes. These observations conflict with the finding of Mukherjee et al. of tumor growth inhibition following energy reduction via carbohydrate restriction. Instead, the data from these studies suggested that prostate cancer risk might be lowered by reducing fat calories, while maintaining total energy intake by increasing calories from other sources. The epidemiologic findings are limited by bias due to measurement error, by uncertainty about the period in life when diet may affect risk, by inability to control the total dietary environment, and by limited ability to adjust for human variation in energy expenditure, body size, and metabolic efficiency. The net energy balance between these latter factors determines weight gain or loss. Thus, weight for height provides an indirect but more reliably measured index of energy intakes. Since long-term energy restriction in individuals with stable metabolic rates and stable energy expenditures leads to reductions in body fat, the findings of Mukherjee et al. of a benefit from energy restriction predict lower prostate cancer risk in lean men than in obese men. However, the epidemiologic literature on the relationship between prostate cancer risk and adiposity is weak and inconsistent. Although epidemiologic data on the relationship between prostate cancer risk and body mass are more extensive than data on energy intake, they are conflicting. Some studies have examined risk in relation to body mass index (BMI), defined as weight in kilograms divided by the square of height in meters; other studies have used different measures of obesity. One case-control study ( 24 ) found a positive association with increased BMI. However, other studies ( 22,25-29 ) have found no association. Several cohort studies have examined prostate cancer risk in relation to BMI, percentage of desirable weight, or relative weight. Some cohort studies ( 30 - 36 ) found positive associations with increasing body mass, whereas others ( 27,37,38 ) did not. In conclusion, the totality of experimental data and some, but not all, epidemiologic data suggest that dietary fat per se does not affect the development or progression of prostate cancer in rodent models but that fat in combination with as yet unknown concomitant dietary changes may enhance carcinogenesis. The current experiments suggest that energy intake may be a major determinant of prostate cancer in animal models, but a complete picture must also include the role of energy expenditure ( 39 , 40 ). What are the implications of these findings for future preventive trials? We need guidance concerning which dietary interventions among men with localized prostate cancer might prevent progression to metastatic disease. Unfortunately, the results obtained by Mukherjee et al. have limited utility for this purpose for two reasons. First, the experiments involved severe energy restriction (30% of total calories), and the feasibility of this Draconian regimen in humans is unclear. Second, the relevance of the results to localized disease is problematic. The findings are based on the Dunning R3327-H rat prostate carcinoma and on LNCaP cells transplanted into SCID mice. Although these are the only androgen-sensitive prostate cancer models generally available at present, they are probably not good models for localized prostate cancer. LNCaP cells were derived from a metastatic lesion and have an uncommon androgen receptor mutation, and the Dunning R3327-H tumor was developed from a large (advanced) prostate tumor that had caused serious disease in its human host ( 41 ). The current experimental findings also are limited by the uncertainties of extrapolating from rodents to humans and by our lack of understanding about the mechanisms underlying the observed relationships between diet and prostate cancer. More work is needed to increase this understanding. Insight into mechanisms will help not only in extrapolating across species but also in designing new experiments and epidemiologic studies that will point the way to intervention trials.