Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
l Introduction The 21 compounds reviewed in this volume were evaluated at the request of the EPA to assist that agency in the event that it becomes necessary to regulate these chemicals in the future. Since 13 of the 21 reviews are up- dates of evaluations published in earlier volumes of this series, the discus- sions are limited to new data. Only 8 compounds were evaluated for the first time. Chapter 3 reviews epidemiological data concerning the diseases that may result from exposure to either arsenic or asbestos in drinking wa- ter. Both substances were reviewed comprehensively in the first and third volumes of Dnnking Water and Health (National Research Council, 1977, 1980~. The more limited review in this report puts into perspective the po- tential risk to human health posed by chronic exposure. In general, the format for the toxicity evaluations in Chapter 2 is similar to that developed in the previous volumes. When the data were sufficient, the committee calculated a suggested no-adverse-response level (SNARL) for chronic toxicity. In most cases, whenever chronic SNARL's were esti- mated, data were available from studies lasting a major portion of the life- time of the test animal. For these SNARL's an arbitrary assumption (also used in previous volumes) was made that 20~o of the intake of the chemical of concern was derived from drinking water. This was done to provide a basis for calculation and also because there is virtually no information on the relative contribution of other sources to intake of the compounds re- viewed. In the event that this value becomes known for a particular com- pound, the SNARL value can be easily adjusted. Because of this assump- tion, it would be inappropriate to use these values as though they were 1
2 DRINKING WATER AND HEALTH maximum contaminant intakes. These numbers are not a guarantee of ab- solute safety. Furthermore, SNARL's are based on exposure to a single agent and do not take into account possible interactions with other con- taminants. In all cases, the safety (or uncertainty) factor used in the calcu- lations of the SNARL's reflects the degree of confidence in the data as well as the combined judgment of the committee members. As in Volumes 1, 3, and 4 of Drinking Water arid Health (National Research Council, 1977, 1980, 1982), the following assumptions were used to determine which un- certainty factor should be used in the calculations: · A factor of 10 was used when there were good chronic exposure data on humans and supportive chronic data on other species. · A factor of 100 was used when good chronic toxicity data were avail- able for one or more species. · A factor of 1,000 was used when the chronic toxicity data were limited or incomplete. In previous volumes the committee calculated acute SNARL's in addi- tion to chronic SNARL's when there were adequate data. This was not done in the present volume because it was anticipated that another Na- tional Research Council committee would do this as part of its ongoing activities. Where there was evidence of carcinogenicity in one or more animal spe- cies, the committee calculated a numerical risk estimate using the method- ology described in the first and third volumes of Drinking Water and Health (National Research Council, 1977, 1980~. More details are pro- vided under the "Carcinogenic Risk Estimate" section for each compound where this procedure was used. There is an expanded discussion of risk assessment in Chapter III of Drinking Water and Health, Volume 3 (Na- tional Research Council, 1980~. The committee that prepared Volume 4 (National Research Council, 1982) did not estimate risks for chemicals that were potentially carcino- genic to humans. The present committee decided that the inclusion of esti- mates would be useful, even if some people may disagree with the selection of a particular mathematical model. Presented with each cancer risk esti- mate are the tumor incidence data so that the reader may substitute any other mathematical model. In this volume (as in previous volumes) the risk estimates from data on different species and tumors that resulted from exposure to a particular compound were averaged to yield one composite estimate. This was done because when tumors develop in more than one species it is not possible to know which is more representative of humans and also to reduce the confusion that results when more than one risk es-
Introduction 3 timate number is associated with each compound. I-he individual risk esti- mates for each species and tumor site are presented in tabular format. The committee discussed several different methods for interpreting the conflicting and ambiguous data on the carcinogenicity of several com- pounds. Many ambiguities could be resolved if better metabolic data in humans were available for classes of compounds, so that the metabolic fate of the compound in different animal species and strains could be com- pared to that in humans. Regrettably, little such information is now avail- able. Since uncertainties in the interpretation of bioassay results may be caused by the use of high doses and by differences in response among the species, strain, and sex of the test animals, the committee agreed to adopt the approach of the International Agency for Research on Cancer (1980, pp. 18-19) for interpreting animal carcinogenicity tests: In general, the evidence that a chemical produces tumours in experimental ani- mals is of two degrees: (a) sufficient evidence of carcinogenicity is provided by the production of malignant tumours; and (b) limited evidence of carcinogenicity re- flects qualitative and/or quantitative limitations of the experimental results. Sufficient evidence of carcinogenicity is provided by experimental studies that show an increased incidence of malignant tumours: (i) in multiple species or strains, and/or (ii) in multiple experiments (routes and/or doses), and/or (iii) to an unusual degree (with regard to incidence, site, type and/or precocity of onset). Additional evidence may be provided by data concerning dose-response, muta- genicity or structure. In the present state of knowledge, it would be difficult to define a predictable rela- tionship between the dose (mg/kg low/day) of a particular chemical required to produce cancer in test animals and the dose which would produce a similar inci- dence of cancer in humans. The available data suggest, however, that such a rela- tionship may exist at least for certain classes of carcinogenic chemicals. Data that provide sufficient evidence of carcinogenicity in test animals may therefore be used in an approximate quantitative evaluation of the human risk at some given e~cpo- sure level, provided that the nature of the chemical concerned and the physiologi- cal, pharmacological and toxicological differences between the test animals and the humans are taken into account. However, no acceptable methods are currently available for quantifying the possible errors in such a procedure, whether it is used to generalize between species or to extrapolate from high to low doses. The meth- odology for such quantitative extrapolation to humans requires further develop- ment. Evidence for the carcinogenicity of some chemicals in experimental animals may be limited for two reasons. Firstly, experimental data may be restricted to such a point that it is not possible to determine a causal relationship between administra- tion of a chemical and the development of a particular lesion in the animals. Sec- ondly, there are certain neoplasms, including lung tumours and hepatomas in
4 DRINKING WATER AND HEALTH mice, which have been considered of lesser significance than neoplasms occulting at other sites for the purpose of evaluating the carcinogenicity of chemicals. Such tumours occur spontaneously in high incidence in these animals, and their malig- nancy is often difficult to establish. An evaluation of the significance of these tu- mours following administration of a chemical is the responsibility of particular Working Groups preparing individual monographs, and it has not been possible to set down rigid guidelines; the relevance of these tumours must be dete.l..ined by considerations which include experimental design and completeness of reporting. Hence 'sufficient evidence' of carcinogenicity and 'limited evidence' of carcinoge- nicity do not indicate categories of chemicals: the inherent definitions of those terms indicate varying degrees of experimental evidence, which may change if and when new data on the chemical become available. The main drawback to any rigid classification of chemicals with regard to their carcinogenic capacity is the as yet incomplete knowledge of the mechanism(s) of carcinogenesis. The committee agrees that rigid guidelines for establishing carcinoge- nicity are difficult to develop and that each compound must ultimately be evaluated on the basis of the available evidence. Occasionally a risk esti" mate was performed on data that, based on the above criteria, were judged to be limited. This was done so the relative risk of different chemicals could be compared, but the committee recommends that less weight be given to these estimates than to those based on sufficient evidence. Although current risk assessment procedures are fraught with uncer- tainty, they are the only available means to obtain a comparable perspec- tive on the potential for cancer development from chronic exposure to se- lected compounds. In this volume, the committee uses the process of quantitative risk assessment in estimating the probability of cancer after a lifetime daily consumption of 1 liter of water containing compounds in a concentration of 1 ~g/liter.The committee recognizes that these risk esti- mates are indeed estimates and should not be considered in a sense of cer- titude. It should be realized that the risk estimation process involves a di- rect extrapolation of doses in animals to humans, based on body surface area. Implicit in this process are the assumptions that animal models and human beings respond in a quantitative manner and that it is appropriate to extrapolate to doses as much as six orders of magnitude lower than those given to animals. Within the scientific community, there is considerable debate over the toxicological bases of the above assumptions, which are accepted here as workable hypotheses for the calculation of numerical risks to provide guid- ance for qualitative judgements. However, the committee is concerned that undue credence will be given to the quantitative aspects of such risk values, especially when quantitative risk estimates are fundamentally unverifi-
Introduction 5 able under commonly used research methodology and economic re- sources. The committee recognizes that regulatory agencies must base their deci- sions on the most scientifically sound evaluation of the data. The deficien- cies of such data must be recognized, but they should not preclude care- fully reasoned, cautious judgments. To gain more confidence in current extrapolation models, it is essential that the level of uncertainty be re- duced. This can be accomplished only by vigorous research efforts to over- come the many limitations of using animal models to predict potential re- sponses in humans. The biological differences in response to chemical exposure among spe- cies or strains should be regarded as important information in predicting the potential qualitative and quantitative responses of humans. When an adequate bioassay produces positive findings of carcinogenicity in one spe- cies and negative results in another, the differences in results should be explained in terms of the biological differences between the animals. These differences should then be considered in evaluating potential risk to hu- mans. The committee recognizes that this usually cannot be done, but that there is a need to develop data that will make this possible. It strongly recommends that methods be developed to enhance the utility of labora- tory animals to predict toxic effects in humans through an increased un- derstanding of their biological similarities and differences. Many short-term mutagenicity tests are currently used to aid in the eval- uation of the carcinogenic potential of chemicals. Many of these tests are based on the use of microorganisms or cells from mammals or plants to detect agents that react with and modify deoxyribonucleic acid (DNA). This is an important step because ample evidence suggests that DNA dam- age leads to mutational events in all living organisms. A cellular genetic change is important to humans because it may result in a cellular transfor- mation that, in turn, may produce neoplasia. This process is referred to as the somatic mutation theory of carcinogenesis. Since DNA is a target mole- cule for mutagenesis, a chemical was designated as a mutagen in this ~rol- ume based on its ability to induce mutations in any one short-tenn test. However, it should be recognized that epigenetic mechanisms may also play a role in tumorigenicity in the absence of direct DNA change. There- fore, some carcinogens may not be detectable in mutagenicity screening tests. There are now generally accepted guidelines for interpreting short-term test results as a basis for assessing risk to human health. A series of hierar- chical tests may be used to evaluate the mutagenic effects of chemicals. Prom the findings of such tests, one can predict the qualitative, but not
6 DRINKING WATER AND HEALTH quantitative, effects on human health. Evaluation of potential carcinoge- nicity is currently based on correlations between short-term mutagenicity test results and data obtained from studies of carcinogenicity in laboratory animals or epidemiological studies of humans. However, data on the lim- ited number of chemicals known to be carcinogens in humans are not suffi- cient to provide an adequate basis for validation of the short-term tests. In addition, most validation studies have included chemicals that have been assayed in carcinogenicity studies in animals, but such studies do not al- ways provide unequivocal answers as to the carcinogenic or noncarcino- genic nature of a chemical. Therefore, caution should be exercised in using mutagenicity test results to predict potential carcinogenicity, especially since not all mutagens are carcinogenic, nor are all carcinogens muta- genic. The National Research Council Committee on Chemical Environmental Mutagens (National Research Council, 1983) made the following state- ment concerning the risk of mutation to future human populations: Efficient mutagenicity tests that use experimental organisms and in vitro systems are a product of great advances in basic knowledge and of a substantial research effort. The principal task of the Committee [on Chemical Environmental Muta- gens] is to study hold the results of such tests can be used to assess the risk to future human populations. There are nc, reliable data from direct human experience, so it is necessary to rely on experimental test systems. Some tests are exquisitely sensi- tive to chemical mutagens, but use microorganisms or mammalian-cell cultures of uncertain relevance to human germ cell damage. Although there has been great progress in the development of test systems, it is still not possible to predict the human impact of a mutagen with confidence. There are two reasons for this: first, in the absence of human data, it is not possible to validate the test systems, and second, one must assume that they predict human effects. Even if the damage to human germ cells could be measured precisely, we lack the knowledge to translate the measurements into a total impact on the health and welfare of future genera- tions this situation is not likely to change soon. Assessment of teratogenicity is quite different from assessment of muta- genicity and carcinogenicity. A teratogen is defined as an agent that pro- duces either a major or minor deviation from normal morphology or func- tion during embryonic or fetal development. The molecular and cellular mechanisms that produce teratogenesis are multiple, which has made the search for a prescreen very difficult. There are at least 10 different in vitro systems under development and validation, but none of them have come into common usage. Another unique aspect of teratogenicity testing is that almost any agent administered at a high enough dose during a sensitive time in prenatal development can produce a teratogenic effect. Sucrose and sodium chloride are examples of such agents (Shepard, 1980~. This
Introduction 7 means that in humans, exposure level, time of exposure, individual sensi- tivity, and potency are probably the controlling factors for teratogenicity rather than solely the demonstration that an agent is teratogenic in a labo- ratory animal or in some in vitro test. Another unique feature of teratology tests is that the exposure period in humans is generally limited to 9 months, and the end point is often mea- surable at birth. Examples of delayed teratogenic action have been identi- fied but are not common. This relatively short exposure period and the end point measurement have enabled investigators to make some epidemiolog- ical analyses. To date, such studies have indicated that approximately 25 agents may be teratogenic in humans. The data for most of the compounds reviewed in this report were not sufficient to judge with certainty that a particular compound was or was not teratogenic in animals. Therefore, many of the evaluations are based on what the committee considered to be limited evidence. Because of the limited nature of the data, no judgment can be made about teratogenic potential in humans. Postnatal studies were included in the reviews, but studies in ovo were not considered. The reviews of arsenic and asbestos are limited to evaluations of epide- miological studies. Each of these substances has been reviewed and evalu- ated in previous volumes of Drinking Water and Health, but recent data make another evaluation timely. In the case of arsenic, recent epidemiological studies conducted in the United States are at variance with the results of some older studies done in Taiwan. Furthermore, there is now some evidence that arsenic may be an essential nutrient. These issues are examined in detail in Chapter 3. For asbestos there are now epidemiological data on populations exposed via drinking water in addition to the data on occupational exposures via inhalation. These oral exposure studies are examined in detail, and the risk of cancer is compared with the risk derived from the inhalation stud- ies. Finally, the committee presents a model to estimate the risk of gastric cancer from ingested asbestos fibers. REFERENCES International Agency for Research on Cancer. 1980. Some Metals and Metallic Compounds. IARC Monographs on the Evaluation of the Carcinogenic Risk of Chemicals to Humans. Lyon: International Agency for Research on Cancer. National Research Council. 1977. Drinking Water and Health. Report of the Safe Drinking Water Committee, Advisory Center on Toxicology, Assembly of Life Sciences. Washing- ton, D.C.: National Academy of Sciences. 939 pp. National Research Council. 1980. Drinking Water and Health. Vol. 3. Report of the Safe
DRINKING WATER AND HEALTH Drinking Water Committee, Board on Toxicology and Environmental Health Hazards, Assembly of Life Sciences. Washington, D.C.: National Academy Press. 415 pp. National Research Council. 1982. Drinking Water and Health. Vol. 4. Report of the Safe Drinking Water Committee, Board on Toxicology and Environmental Health Hazards, Assembly of Life Sciences. Washington, D.C.: National Academy Press. 299 pp. National Research Council. 1983. Identifying and Estimating the Genetic Impact of Chemi- cal Mutagens. Report of the Committee on Chemical Environmental Mutagens, Board on Toxicology and Environmental Health Hazards, Commission on Life Sciences. Washing- ton, D.C.: National Academy Press. 295 pp. Shepard, T.H. 1980. Catalog ofTeratogenic Agents. Baltimore: TheJohns Hopkins Univer- sity Press. 410 pp.