Click for next page ( 431


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 430
IS Assessment of Risk to Human Heady The literature on risk assessment has expanded rapidly in recent years. Some of the most comprehensive reports have been prepared by committees of the National Academy of Sciences (1977, 1980a,b). This chapter is only intended to supplement that literature with primary ref- erence to risk analysis of the effects of diet, particularly nutrients, on the process of carcinogenesis. In this context, risk is defined as the probability that an individual will develop cancer within a given time. Such risk can be estimated by examining retrospective data. To determine dietary and nutritional effects on cancer, it is useful to relate risk analysis to the basic concepts of carcinogenesis. That is, chemically induced carcinogenesis may include (a) an initiation phase, when chemicals or other agents possessing genotoxic activity interact with the genome of somatic cells, and (b) a modification phase, when a variety of events modify that process either simultaneously with, or subsequent to, the initiation phase. Initiators of carcinogenesis are generally mutagenic in one or more systems (McCann et al., 1975; Purchase et al., 1978; Sugimura et al., 1976) and form covalent adducts with DNA and other cellular macromole- cules, usually after enzymatic metabolism (Boyland and Levi, 1935) to chemically reactive metabolites (Magee et al., 1975; Miller, 1970~. If such adducts or other interactions between DNA sites are not removed or repaired, and if the cell, through replication, transmits this molecular aberration to future cell generations, then the initiating event is con- sidered to be irreversible. In contrast to compounds that act as initiators, many other compounds may modify either the initiating events per se, or the subsequent multi- staged events responsible for the progression of the initiated cell to the fully developed neoplastic cell (Slaga, 1980~. A wide variety of chemicals can modify later stages of carcinogenesis. Many of them act exclusively as modifiers; others are also initiators (complete carcino- gens). The process of carcinogenesis may be modified through a great many diverse mechanisms, which can involve the intervention of several physio- logical/biochemical systems, including changes in the function of the enzymes, hormones, immune response, membrane transport and communication activities, etc. Some modifiers have positive effects (i.e., they are promoters or cocarcinogens), whereas others have negative effects (i.e., 430 18-1

OCR for page 430
Assessment of Risk to Human Health 431 they are inhibitors of initiation, antipromoters, or anticocarcinogens) on the progression of the initiated cell to the tumorigenic state. In contrast to initiation, the progress of positive modification (promotion) appears to be more reversible (Doll and Hill, 1964; Roe and Clack, 1963~. This distinction between initiation and modification is necessary for an understanding of the literature on risk analysis. Most such litera- ture assumes that carcinogenic substances act as initiators. Models developed for initiation, e.g., those for threshold response, additivity vs. nonadditivity of toxic response, dose-response kinetics, and extent of exposure ("hitness"), do not necessarily apply to assessment of risk due to tumor modifiers. For example, one of the most straightforward differences is the direct correlation between dose and tumor response for initiators versus the inverse correlation between the tumor response and the dose of certain modifiers such as several nutrients and certain antioxidants. Food may contain both initiators and modifiers of carcinogenesis (see Chapters 12 and 14~. The great diversity of initiators to which humans are exposed is suggested by the wide variety of mutagens with a broad range of potencies present in the food we eat and the excrete produced (Bruce et al., 1977; Sugimura, 1979~. There will always be an ample supply of modifiers in food, including a variety of nonnutritive sub- stances and most, if not all, nutrients (when consumed in amounts that either exceed or are less than those required for optimum nutrition). These modifiers may affect carcinogenesis by influencing hormone status (Albert), 1980), immune response (Axelrod, 1980; Gross and Newberne, 1980), and the activity of carcinogen-metabolizing enzymes (Campbell, 1979; Conney, 1967~. INITIATORS OF CARCINOGENESIS In general, two procedures may be used to estimate acceptable levels of risk for toxic compounds. One method estimates the acceptable daily intake (ADI); the second is the "risk estimate" approach (National Academy of Sciences, 1980a). The ADI is an arbitrary estimate that "is not an estimate of risk nor a guarantee of absolute safety" (National Academy of Sciences, 1980a). It has been widely used for noncarcinogenic toxic chemicals and is based on empirical data pertaining to acute toxicity. This procedure was modi- fied by the Safe Drinking Water Committee to estimate acceptable daily intake for contaminants in drinking water because of the paucity of data for these compounds (National Academy of Sciences, 1977~. To establish an ADI, the highest experimental dose that produces no observable effects is decreased by an "uncertainty factor" or "safety factor" (ranging from 18-2

OCR for page 430
432 DIET, NUTRITION, AND CANCER 10 to 5,000) in order to minimize the probability of harm to the more sensitive members of the general population. A commonly used safety factor for chronic toxicity tests is 100, which assumes a factor of 10 to allow for variability of individual responses within the test species and a second factor of 10 for the assumed differences in response between the test species and humans (Lehman and Fitzhugh, 1954~. Although the 100- fold safety factor has been widely used to evaluate food additives and other chemicals (World Health Organization, 1958, 1972), there are no empirical data to support this specific factor. In general, the greater the uncertainty in the experimental data, the larger the safety factor (National Academy of Sciences, 1977~. The ADI method is unacceptable for estimating risk for initiators because there is a great variation among species in susceptibility to carcinogens, and because of the serious consequences for regulatory action if the estimates are misleading (National Academy of Sciences, 1980a). The more appropriate approach for estimating risk for initiators of carcinogenesis is to estimate the response of humans to low doses based on data derived from exposure of laboratory animals to high doses. This approach is comprised of two quantitative estimations: first, interpola- tion from responses obtained at high doses to estimate the response at low doses within the test species and, second, extrapolation of the data from the test species to estimate the response in humans. Several mathematical models have been proposed to interpolate from high to low doses within the test species (Scientific Committee of the Food Safety Council, 1978; National Academy of Sciences, 1980a). The greatest uncertainty in these models is whether or not the kinetic char- acteristics of the responses at high doses are similar to those at low doses. This is a particularly important consideration when determining whether there is a threshold. For example, is the risk directly propor- tional to dosage level or is there a threshold dose below which the re- sponse is negligible? Even though mechanistic arguments may be developed in favor of a threshold effect (Gehring and Blau, 1977; see dose-response curves in the report of Scientific Committee of the Food Safety Council, 1978), most mathematical models that have been used in the regulatory process assume the nonexistence of a threshold for the general population (National Academy of Sciences, 1980a). At the very least, were thresh- olds to exist for individuals, there would undoubtedly be a distribution of threshold values for which population parameters would need to be estimated. That is considered impractical. Assuming thresholds to be either nonexistent or unmeasurable, the simplest mathematical model for high dose to low dose interpolation is the linear dose-response model, wherein the response is directly propor- tional to the dose at low levels of exposure. Mathematical models using dose-response curves other than linear include the log-linear, log-normal, and log-logistic models. Models based on 'target theory," which are used for radiation-induced carcinogenesis, assume that the site of action has 18-3

OCR for page 430
Assessment of Risk to Human Health 433 a finite number of target sites that require some finite number of "hits" to elicit a response (see review by Turner, 1975~. A more recent exten- sion of this model was proposed by Cornf ield ( 1977 ~ . According to the Scientific Committee of the Food Safety Council (1978), this model, in which multiple hits are assumed and which has the strongest biological foundation, provides a better description of dose-response data than the single-hit model. This "gamma hit model" was found to fit dose-response data for various types of toxic responses better than did some of the simpler models (Scientific Committee of the Food Safety Council, 1978~. There is little support for and use of other models that mathemati- cally describe a tolerance distribution, because they do not take into consideration the multistep nature of chemical carcinogenesis (Finney, 1952~. Some models (Druckrey, 1967; Peto et al., 1972; Pike, 1966) include additional parameters such as the time from initiation of expo- sure to the development of tumor or, where there is no response, the total period of observation. This period of observation would include the time during which chemical modifiers would be expected to exert their effects. As mentioned earlier, the greatest difficulty with all of these models is the interpolation of the response to low doses. Traditional bioassays in animals are limited to the use of levels high enough to produce an observable response. Virtually none of the data resulting from such bioassays can be used to determine the most appropriate model for interpolation to low doses. For example, even though similar risk estimates were obtained from three common dose-response models (the log- normal, log-logistic, and single-hit models) over a 256-fold experimental dose range at relatively high doses, the projected risks at lower doses were increasingly divergent (Food and Drug Administration Committee on Protocols for Safety Evaluation, 1971~. At a dose of 0.01%, which gave a 50% response, the single-hit model yielded a risk estimate 70,000 times higher than that obtained with the log-normal model. Seriously compounding this uncertainty even further is the selection of the most appropriate basis for expressing the dose. When more mean- ingful comparative pharmacokinetic data are unavailable, it becomes nec- essary to select a common unit to express the dose received by various small, short-lived rodent species in order to compare that dose to the dose received by humans. There is no unanimous agreement on whether to express the dose as quantity per unit of body weight, body surface area, or over a lifetime. To calculate the risk to humans from saccharin, the Committee for a Study on Saccharin and Food Safety Policy (National Academy of Sciences, 1978) compared results from four models for high dose to low dose interpolation (the single-hit, multistage, multihit, and probit models) and expressed them as dose received from all three base units (body surface area, body weight, and lifetime). The estimate of risk derived from a Canadian study (Arnold et al., 1980) ranged from 0.001 x 10-6 lifetime cases to 5,200 x 10-6 lifetime cases for ex- posed individuals, a range of 6 orders of magnitude. Most of this var- iance was attributable to the differing risks predicted by linear and probit models. 18-4

OCR for page 430
434 DIET, NUTRITION, AND CANCER The Office of Technology Assessment (OTA) of the U.S. Congress re- cently issued a report on the assessment of technologies for determining cancer risks from the environment (Office of Technology Assessment, 1981~. Although this group accepted the nonthreshold extrapolation from high dose measurements in animals to low dose estimates for humans, it acknowledged that there can be as much as a fortyfold variation in the risk estimate for low doses in humans, depending upon the scaling factor used for determining the relationships between laboratory animals and humans with respect to body size and rate of metabolism. Generally, toxicologists adjust exposures for the differences in scale between spe- cies on the basis of milligram per unit of body weight. This provides the lowest estimate of risk to humans. When the experimental dose is measured in parts per million (e.g., mg/kg diet, mg/cm3 air, or ma/ liter water) and humans are exposed through ingestion, the dose of the chemical can be expressed as parts per million. This method is generally used by the Food and Drug Administration (FDA) and, in some cases by the Environmental Protection Agency (EPA), but it produces an estimate of risk in humans 6 times greater than that estimated by the former method if the mouse is the laboratory animal or 3 times greater if the rat is used. Another method used by EPA takes the scale differences between species into account and adjusts exposure on the basis of the relative body surface areas of the test animal and humans. This gives a pro- jected risk for humans 6 to 14 times higher than that of the first method. The fourth approach is to adjust exposures on the basis of relative body weight over a lifetime. This gives a projected risk for humans 40 times higher than that of the first method. An additional factor to be considered is the shape of the dose- response curve. In general, linear interpolation provides the most conservative estimate. This approach is used by most regulatory agencies and is supported in the OTA report (Office of Technology Assessment, 1981). The ultimate choice of a model for high to low dose interpolation is, therefore, arbitrary. Not only is there great uncertainty in the mathe- matical modeling procedures, but also there is no sound biological basis for any of them (Scientific Committee of the Food Safety Council, 1978~. The Safe Drinking Water Committee concluded that the most suitable model may be the multistage model because of the multistep nature of carcino- genesis and because of the model's relatively conservative estimate of risk (National Academy of Sciences, 1980a). The committee sees no reason to modify that conclusion except to suggest that it should be reserved for initiators of carcinogenesis and that there should be awareness of the large variability among the heterogeneous human population. MODIFIERS OF CARCINOGENESIS For modifiers of carcinogenesis a different approach may have to be considered. There have been few systematic attempts to assess the risk 18-5

OCR for page 430
Assessment of Risk to Human Health 435 for such compounds. However, the models discussed above may provide some leads to potentially useful approaches, such as time to tumor response, originally defined as the latent period by Armitage and Doll (1961~. For example, experiments using doses of the initiator producing a high tumor response would be useful in studying dose-dependent effects of negative modifiers; conversely, experiments that produce barely perceptible re- sponses could be used to examine dose-dependent responses of enhancers of carcinogenesis. Although more data are needed to develop this concept, several investigators have proposed specific mathematical models that incorporate time to tumor response (Druckrey, 1967; Peto et al., 1972; Pike, 1966~. Moreover, continuous exposure to a carcinogen is con- sidered in some models and time to tumor response has been incorporated as a function of dose in order to yield the age-specific incidence rate (An~itage and Doll, 1961; Crump, 1978; Hartley and Sielken, 1977~. Although the time to tumor response may have considerable hypothetical and experimental utility for modifiers of carcinogenesis, such models are severely limited for use in population-based studies because of the unknown contribution and variable distribution of initiators. Furthermore, withdrawal of an initiator may intercept progression of carcinogenesis (Boutwell 1964; Sivak, 1979; Teebor and Becker, 1971~. Halving the dose-dependent progression of carcinogenesis should decrease the final response by that amount, regardless of when such interception occurred (Peso _ al., 1981~. The Safe Drinking Water Committee concluded that conventional risk extrapolation methodology accompanied by sufficient data and reasonable models would predict risk to humans from studies in animals using low doses with a precision varying from 1 to 2 orders of magnitude (National Academy of Sciences, 1980a). Because of the differences in individual susceptibility, the precision may, in fact, be much less. USE OF MUTAGENICITY TESTS As discussed in Chapter 3, the estimation of risk to humans from exposure to mutagens is beset with uncertainties. First, it is difficult to determine the dose of such substances in the diet. For example, the general population may eat small or very small amounts of some extremely strong mutagens (e.g., aflatoxin B1) and large quantities of certain weak mutagens (e.g., flavonoids in vegetables), and the capability of our biological defense mechanisms to protect us from either of these is not well understood. Presumably, the effects of such exposure will be a function of potency and dose, both of which extend over a very wide range. To translate the product of potency and dose for a specific muta- gen into an estimate of absolute risk for a human population, it is nec- essary to know how to convert the values of mutagenicity obtained in 18-6

OCR for page 430
436 DIET, NUTRITION, AND CANCER simple test systems into estimates of carcinogenicity. Unfortunately, no way has been found to predict the carcinogenicity of any specific muta- gen. One of the major difficulties in attempting such a determination is the known interspecies variation in susceptibility to the effect of car- cinogens and mutagens. Until the variables that control susceptibility are better understood, it is impossible to extrapolate from tests for mutagenicity to obtain estimates of carcinogenicity in humans. There- fore, mutagenicity tests, which usually detect initiators, can be used only as qualitative indicators of possible carcinogenicity. In the absence of evidence derived from epidemiological studies, it is not possible to estimate the likelihood that such substances will affect the occurrence of cancer in humans. USE OF EPIDEMIOLOGICAL STUDIES l Chapter 3 describes the approach used in epidemiological studies to assess the importance of exposures to carcinogens or risk factors such as diet or dietary components. The risk that disease will occur in humans during a given time period is measured by the incidence of that disease. Incidence can be accumulated over a lifetime in a population to measure cumulative incidence, which is approximately equivalent to the lifetime risk of disease (Day, 1976~. In cohort studies, the incidence of disease is measured directly: the ratio of incidence in exposed and unexposed cohorts provides a measure of the relative risk. The difference in incidence of disease between exposed and unexposed groups provides a second measure of risk--the attributable risk, which is simply the proportion of disease that can be attributed to the exposure to that particular variable, if it is causal. Such measurements of risk are valid only if the two popu- lations compared differ only by the exposure (variable) being studied and if other factors influencing the risk of disease are controlled in the analysis. The derivation of the attributable risk is dependent on the reasonable assumption that the exposed group would have had an incidence similar to that of the unexposed group if it had not been exposed. In case-control studies, estimates of the relative risk can be de- rived. A large body of statistical and epidemiological literature has accumulated to confirm that the odds ratios derived from case-control studies are approximately equivalent to the relative risk determined in cohort studies. Indeed, the same terminology is normally used in both types of studies. Estimates of population-attributable risk can also be derived from case-control studies. Such estimates are equivalent to those expected in the general population, provided that the cases and the controls are representative of their respective target populations so that estimates of exposure in the population can be derived from the control series. 18-7

OCR for page 430
Assessment of Risk to Human Health 437 These estimates of risk do not take into account the extent to which risk may vary with age. Almost invariably there is a latent period be- tween first exposure and the expression of a risk. For many diseases, risk may appear to remain constant with age, but risk may decrease in older groups because the population may have lost most persons suscepti- ble to the effect, especially if a risk factor or constellation of risk factors is responsible for more than one disease and if these diseases are relatively common in a population or because of a birth cohort effect that is increasing. An additional complexity is that risk may vary in populations because the extent of exposures to various risk factors may vary with time. This may lead to different expectations of lifetime incidence for different birth cohorts. In this respect, incidence among different age groups during the same time period may not accurately reflect the expected lifetime incidence in a population. For example, there has been a decline in risk for tuberculosis in successive cohorts in the United States. The risk for this disease in the present genera- tion is largely restricted to elderly males. Another example is the risk for lung cancer, which has been increasing with successive cohorts, especially among males, so that cross-sectional incidence curves show maximum risk at middle ages and declining risk at older ages. For cervi- cal cancer, there have been complex birth cohort effects. For example, those who were in their twenties during the depression years show a low incidence of cervical cancer, whereas others, particularly those who are now in their twenties, appear to have a higher incidence. Thus, it is difficult to interpret trends in incidence. However, for most of the cancers believed to be influenced by diet and nutrition, the incidence has been relatively stable for many years. Therefore, estimates of the proportion of these cancers attributable to dietary factors are not likely to be severely in error because of differences in risk encountered by different birth cohorts. Because the causation of cancer is often multifactorial, the summa- tion of estimates of the percentage of cancers attributable to individual factors often exceeds 100%. However, this does not invalidate the con- cept. Rather, it indicates that it may be possible to adopt different approaches to preventing a number of cancers. DIET-RELATED CARCINOGENESIS Given the limitations of traditional methods of risk assessment using data from experimental and epidemiological studies, what factors would be the most appropriate to consider in analysis of risk for diet-related carcinogenesis? Because food contains both initiators and modifiers of carcinogene- sis, there is a need for two very different kinds of risk analysis. For the initiators (mostly nonnutritive components of food), the Safe Drink- ing Water Committee's conclusion that a multistage mathematical model 18-8

OCR for page 430
438 DIET, NUTRITION, AND CANCER would be most appropriate (National Academy of Sciences, 1980a) is still considered to be valid. For the modifiers in food (including nutrients and nonnutritive substances), a mathematical model that includes time to tumor response may be preferable. For enhancers, an experimental titra- tion of a dose of the modifier, which has a potency for increasing a near-zero tumorigenic response, may be an acceptable approach. Con- versely, for inhibitors, titration of the dose of the modifier with the potency for eliminating nearly 100% of the response may be preferable. Any experiment to determine the carcinogenic response to a dose of a nutrient will be specific for the initiator, species, diet, and other experimental variables used in the test. Caution will have to be exercised when extrapolating the resultant data to humans. In general, such extrapolation should initially be limited to qualitative rather than to quantitative conclusions. Greater confidence in the initial results may be gained by testing more initia- tors, species, and diets. The necessity for testing additional diets arises from the knowledge that there is an extremely broad array of nu- trient-nutrient interactions. Thus, these reactions must be considered when selecting the new diets to be tested. As new data become available, it may become possible to do more quan- titative extrapolation from the test species to humans with respect to the risk from nutrient intake. The specific dose-response relationship for the effects of nutrients on carcinogenesis will necessarily be limit- ed by the requirement for nutrients for metabolic functions. This sug- gests, therefore, that "risk" from nutrient intake should be defined in terms of the Recommended Dietary Allowances (RDA) (National Academy of Sciences, 1980c). Beginning with the RDA's as the point of reference, the "risk" for a nutrient could be defined in teems of RDA multiples (or fractions there- of). This approach would introduce a broader perspective into the inter- pretation of the "carcinogenic" effects of nutrients. It would acknowl- edge that nutrients are essential and admit the existence of some level of risk, even if that risk were negligible. The dose-response slope constants will undoubtedly differ for various nutrients, and acceptable upper limits of intake, however arbitrary, will vary broadly as functions , , _ , .' , _ ~ ., of the RDA's for different nutrients. Analogous phenomena for noncar- cinogenic toxicity of nutrients have been evaluated elsewhere (Campbell et al., 1980; National Nutrition Consortium, 1978~. The Food and Nutrition Board's most recent edition of the Recommended Dietary Allowances (National Academy of Sciences, 1980c) provides ranges of intake for three vitamins and six minerals, the upper limits being defined as "safe" on the basis of available information. But, it is clear that epidemiological studies will be needed to confirm or deny whatever risk estimates for nutrient intake may be obtained from experi- ments in animals. 18-9

OCR for page 430
Assessment of Risk to Human Health 439 Doll and Peto setting priorities (1981) suggest that laboratory data may be used for _ for regulation. However, they point out that even tests in animals are unreliable, suffering not only from random errors, but also probably from large systematic errors of unknown direction and magnitude. Since there are thousands of chemicals that affect carcino- genesis to some extent or other, in one laboratory test or another, it is difficult to determine what, if any, practicable regulations should be enacted on the basis of laboratory tests. Nevertheless, they suggest ~ ' "- '~ laboratory tests might be to estimate risk for humans by multiplying the potency of each chemical studied by estimates, however crude, of the degree to which humans are exposed to that chemical. This would yield an index of risk for humans. Resultant estimates might provide a basis from which priorities for regu- latory action could be determined. that an aDDroDriate use ot results trom There is one major difficulty with this approach. Although experi- mental or epidemiological studies may have identified risk factors as either potentially hazardous or protective for humans, it is difficult to label many of them as carcinogens or modifiers of carcinogenesis. Furthermore, even if such assessments were possible, the feasibility of modifying the diet of humans would have to be considered. It is impossible to enact regulations pertaining to ties, except for those few substances that clearly qualify as food additives, which are currently regulated according to the provisions in the Delaney Clause (see Section B). Thus, it may not be appropriate to assess priorities for regulatory action at this time, especially since changes in the econ- omy and in the sources of our food supply may well combine to impose dietary changes upon us. The dietary changes now under way appear to be reducing our dependence on foods from animal sources. It is likely that there will be continued reduction in fats from animal sources and an increasing dependence on vegetable and other plant products for protein supplies. Hence, diets may contain increasing amounts of vegetable prod- ucts, some of which may be protective against cancer. However, if it is decided that changes have to be instigated, we should consider reducing exposure of the population to total dietary fat and increasing exposure to protective substances such as those found in fruits and vegetables, while ensuring the maintenance of an ideal body weight for height and well-balanced but varied nutrition. CONTRIBUTION OF DIET TO OVERALL RISK OF CANCER Higginson and Muir (1979) estimated the proportion of cancers related to various aspects of the environment. They believed that precise pro- portions of cancer incidence could not be attributed to diet, but they did include dietary factors among the general heading "Lifestyle." They estimated that possibly 30% of cancers in men and 60X in women in the 18-10

OCR for page 430
440 DIET, NUTRITION, AND CANCER Birmingham and West Midland regions of England and Wales could be attrib- uted to lifestyle. Wynder and Gori (1977) were more specific. On the basis of international and intranational comparisons of cancer incidence, the differences between U.S. mortality rates and the lowest reported worldwide mortality rates for each site, and results of specific case- control studies, they concluded that a little more than 40% of cancers in men and almost 60% of cancers in women in the United States could be attributed to dietary factors. Using a similar approach, Doll and Peto (1981) were somewhat more cautious. They agreed that a substantial proportion of cancers in both sexes in the United States was likely to be attributable to dietary fac- tors, but, by surveying the literature, they provided a rather wide range of estimates (i.e., 10% to 70%) for the proportion of deaths from cancer that could be reduced by practical dietary means. They stated that it might not be possible to achieve a large reduction in the near future, but that dietary modifications might eventually result in a 35% reduction of deaths from cancer in the United States. This reduction was estimated to include a 90% reduction in deaths from cancers of the stomach and large bowel; a 50% reduction in deaths from cancers of the endometrium, gallbladder, pancreas, and breast; a 20% reduction in deaths from cancers of the lung, larynx, bladder, cervix, mouth, pharynx, and esophagus; and a 10% reduction in deaths from other sites. These investigators placed a greater degree of confidence in the projected 35% reduction in overall mortality than in the estimated contribution from specific groups of cancer sites. Only two case-control studies of dietary factors and cancer provided estimates of the proportion of cancers at specific sites that are attrib- utable to dietary factors. On the basis of a case-control study of breast cancer in Canada, Miller (1978) estimated that 27% of the risk of breast cancer for women was attributable to total dietary fat intake. For colorectal cancer, Jain et al. (1980) estimated that 41Z of the risk for males and 44% of the risk for females was attributable to saturated fat intake. Both of these estimates are probably too low, because arti- facts in the diet tend to lead to low estimates of relative risk (Marshall _ al., 1981~. This is particularly true for breast cancer, since estimated effects of dietary factors based on current intake are likely to be substantially below the true effect for a factor that is operational earlier in life, possibly during adolescence. The evidence reviewed by the committee suggests that cancers of most major sites are influenced by dietary patterns. However, the committee concluded that the data are not sufficient to quantitate the contribution of diet to the overall cancer risk or to determine the percent reduction in risk that might be achieved by dietary modifications. 18-11

OCR for page 430
Assessment of Risk to Human Health 441 REFERENCES Alberti, K. G. M. M. 1980. Metabolic pathways--Hormone-metabolite interrelations. Pp. 5-19 in S. J. Karran and K. G. M. M. Alberti, eds. Practical Nutritional Support. Pitman Medical Publishing Co., Tunbridge Wells, Kent, England. Armitage, P., and R. Doll. 1961. Stochastic models for carcinogenesis. Pp. 19-38 in Proceedings of the Fourth Berkeley Symposium on Mathe- matical Statistics and Probability. Volume 4, Contributions to Biol- ogy and Problems of Medicine. University of California Press, Berkeley and Los Angeles. Arnold, D. L., C. A. Moodie, H. C. Grice, S. M. Charbonneau, B. Stavric, B. T. Collins, P. F. McGuire, Z. Z. Zawidzka, and I. C. Munro. 1980. Long-term toxicity of ortho-toluenesulfonamide and sodium saccharin in the rat. Toxicol. Appl. Pharmacol. 52:113-152. Axelrod, A. E. 1980. Nutrition in relation to immunity. Pp. 578-591 in R. S. Goodhart and M. E. Shils, eds. Modern Nutrition in Health and Disease. Lea and Febiger, Philadelphia. Boutwell, R. K. 1964. Some biological aspects of skin carcinogenesis. Prog. Exp. Tumor Res. 4 :207-250. Boyland, E., and A. A. Levi. 1935. Metabolism of polycyclic compounds. I. Production of dihydroxydihydroanthracene from anthracene. Bio- chem. J. 29:2679-2683. Bruce, W. R., A. J. Varghese, R. Furrer, and P. C. Land. 1977. A mutagen in the feces of normal humans. Pp. 1641-1646 in H. H. Hiatt, J. E. Watson, and J. D. Winsten, eds. Origins of Human Cancer, Book C. Human Risk Assessment. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y. Campbell, T. C. 1979. Influence of nutrition on metabolism of carcino- gens. Adv. Nutr. Res. 2:29-55. Campbell, T. C., R. G. Allison, and C. J. Carr. 1980. ~easibility of Identifying Adverse Health Effects of Vitamins and Essential Minerals in Man. Prepared for Bureau of Foods, Food and Drug Administration, Washington, D. C. under Contract Number FDA 223-75-2090. Life Sciences Research Office, Federation of American Societies for Ex- perimental Biology, Bethesda, Md. 76 pp. 18-12

OCR for page 430
442 DIET, NUTRITION, AND CANCER Conney, A. H. 1967. Pharmacological implications of microsomal enzyme induction. Pharmaco1. Rev. 19:317-366. Cornfield, J. 1977. Carcinogenic risk assessment. Science 198:693-699. Crump, K. S. 1978. Low dose extrapolation of animal carcinogenicity data. Biometrics 34:155. Abstract 2568. Day, N. E. 1976. A new measure of age standardized incidence, the cumu- lative rate. Pp. 443-445 in J. Waterhouse, C. Muir, P. Correa, and J. Powell, eds. Cancer Incidence in Five Continents, Volume 3. IARC Scientific Publications No. 15, International Agency for Research on Cancer, Lyon, Prance. Doll, R., and A. B. Hill. 1964. Mortality in relation to smoking: Ten years' observations of British doctors. Br. Med. J. 1:1399-1410; 1460-1467. Doll, R., and R. Peto. 1981. The causes of cancer: Quantitative esti- mates of avoidable risks of cancer in the United States today. J. Natl. Cancer Inst. 66:1192-1308. Druckrey, H. F. 1967. Quantitative aspects in chemical carcinogenesis. Pp. 60-77 in R. Truhaut, ed. Potential Carcinogenic Hazards from Drugs: Evaluation of Risks. UICC Monograph Series, Volume 7, Springer-Verlag, Berlin, Heidelberg, and New York. Finney, D. J. 1952. Probit Analysis. 2nd Edition. Cambridge University Press, London and New York. 318 pp. Food and Drug Administration Committee on Protocols for Safety Evaluation. 1971. Panel on Carcinogenesis Report on Cancer Testing in the Safety Evaluation of Food Additives and Pesticides. Toxicol. Appl. Pharmacol. 20:419-438. Gehring, P. J., and G. E. Blau. 1977. Mechanisms of carcinogenesis: Dose response. J. Environ. Pathol. Toxicol. 1~1~:163-179. Gross, R. L., and P. M. Newberne. 1980. Role of nutrition in immuno- logic function. Physiol. Rev. 60:188-302. Hartley, H. 0., and R. L. Sielken, Jr. 1977. Estimation of safe doses in carcinogenic experiments. Biometrics 33:1-30. Higginson, J., and C. S. Muir. 1979. Environmental carcinogenesis: Misconceptions and limitations to cancer control. J. Natl. Cancer Inst. 63:1291-1298. 18-13

OCR for page 430
Assessment of Risk to Human Health 443 Jain, M., G. M. Cook, F. G. Davis, M. G. Grace, G. R. Howe, and A. B. Miller. 1980. A case-control study of diet and colo-rectal cancer. Int. J. Cancer 26:757-768. Lehman, A. J., and 0. G. Fitzhugh. 1954. 100-fold margin of safety. Q. Bull. Assoc. Food Drug Off. U. S. 18:33-35. Magee, P. N., A. E. Pegg, and P. F. Swann. 1975. Molecular mechanisms of chemical carcinogenesis. Pp. 329-419 in E. Grundmann, ed. Handbuch der Allgemeinen Pathologie, Bd. 6, T. 6., Geschwlilste- Tumors II. Springer-Verlag, Berlin, Heidelberg, and New York. Marshall, J. R., R. Priore, S. Graham, and J. Brasure. 1981. On the distortion of risk estimates in multiple exposure level case-control studies. Am. J. Epidemiol. 113:464-473. McCann, J., E. Choi, E. Yamasaki, and B. N. Ames. 1975. Detection of carcinogens as mutagens in the Salmonella/microsome test: Assay of 300 chemicals. Proc. Natl. Acad. Sci. U.S.A. 72:5135-5139. Miller, A. B. 1978. An overview of hormone-associated cancers. Cancer Res. 38:3985-3990. Miller, J. A. 1970. Carcinogenesis by chemicals: An overview--G. H. A. Clowes memorial lecture. Cancer Res. 30:559-576. National Academy of Sciences. 1977. Chemical contaminants: Safety and risk assessment. Pp. 19-62 in Drinking Water and Health. A report prepared by the Safe Drinking Water Committee. National Academy of Sciences, Washington, D.C. National Academy of Sciences. 1978. Saccharin: Technical Assessment of Risks and Benefits. Part 1 of a 2-part study of the Committee for a Study on Saccharin and Food Safety Policy. National Academy of Sciences, Washington, D.C. t240] pp. National Academy of Sciences. 1980a. Problems of risk estimation. Pp. 25-65 in Drinking Water and Health. Volume 3. A report of the Safe Drinking Water Committee. National Academy Press, Washington, D.C. National Academy of Sciences. 198Ob. Regulating Pesticides. A report prepared by the Committee on Prototype Explicit Analyses for Pesticides. National Academy of Sciences, Washington, D.C. 288 pp. National Academy of Sciences. 1980c. Recommended Dietary Allowances, 9th edition. A report of the Con~rnittee on Dietary Allowances, Food and Nutrition Board. National Academy of Sciences, Washington, D.C. 187 pp. 18-14

OCR for page 430
444 DIET, NUTRITION, AND CANCER National Nutrition Consortium. 1978. Vitamin-Mineral Safety, Toxicity, and Misuse. Report of the Committee on Safety, Toxicity, and Misuse of Vitamins and Trace Minerals. American Dietetic Association, Chicago, Ill. 36 pp. Office of Technology Assessment. 1981. Assessment of Technologies for Determining Cancer Risks from the Environment. Office of Technology Assessment, Congress of the United States, Washington, D.C. 242 pp. Peto, R., P. N. Lee, and W. S. Paige. 1972. Statistical analysis of the bioassay of continuous carcinogens. Br. J. Cancer 26:258-261. Peto, R., R. Doll, J. D. Buckley, and M. B. Sporn. 1981. Can dietary beta-carotene materially reduce human cancer rates? Nature 290:201-208. Pike, M. C. 1966. A method of analysis of a certain class of experiments in carcinogenesis. Biometrics 22:142-161. Purchase, I. F. H., E. Longstaff, J. Ashby, J. A. Styles, D. Anderson, P. A. Lefevre, and F. R. Westwood. 1978. An evaluation of 6 short- term tests for detecting organic chemical carcinogens. Br. J. Cancer 37:873-959. Roe, F. J. C., and J. Clack. 1963. Two-stage carcinogenesis: Effect of length of promoting treatment on the yield of benign and malignant tumours. Br. J. Cancer 17:596-604. Scientific Committee, Food Safety Council. 1978. Proposed system for food safety assessment. Food Cosmet. Toxicol. 16(Suppl. 2):1-136. Sivak, A. 1979. Cocarcinogenesis. Biochim. Biophys. Acta 560:67-89. Slaga, T. J. 1980. Cancer: Etiology, mechanisms, and prevention--A summary. Pp. 24 3-262 in T. J. Slaga, ed. Carcinogenesis--A Com- prehensive Survey. Volume 5: Modifiers of Chemical Carcinogenesis. Raven Press, New York. Sugimura, T., S. Sato, M. Nagao, T. Yahagi, T. Matsushima, Y. Seino, M. Takeuchi, and T. Kawachi. 1976. Overlapping of carcinogens and mutagens. Pp. 191-213 in P. N. Magee, S. Takayama, T. Sugimura, and T. Matsushima, eds. Fundamentals in Cancer Prevention. University Park Press, Baltimore, London, and Tokyo. Teebor, G. W., and F. F. Becker. 1971. Regression and persistence of hyperplastic hepatic nodules induced by N-2-fluorenylacetamide and their relationship to hepatocarcinogenesis. Cancer Res. 31:1-3. Turner, M. E., Jr. 1975. Some classes of hit-theory models. Math. Bioscience 23:219-235. 18-15

OCR for page 430
Assessment of Risk to Human Health 445 World Health Organization. 1958. Procedures for the Testing of Inten- tional Food Additives to Establish Their Safety for Use. Second Report of the Joint FAD/WHO Expert Committee on Food Additives. W.H.O. Tech. Rep. Ser. 144:1-19. World Health Organization. 1972. Evaluation of Certain Food Additives and the Contaminants Mercury, Lead, and Cadmium. Sixteenth Report of the Joint FAD/WHO Expert Committee on Food Additives. W.H.O. Tech. Rep. Ser. 505:1-32. Wynder, E. L., and G. B. Gori. 1977. Contribution of the environment to cancer incidence: An epidemiologic exercise. J. Natl. Cancer Inst. 58:825-832. 18-16

OCR for page 430