National Academies Press: OpenBook

Air Pollution, the Automobile, and Public Health (1988)

Chapter: Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust

« Previous: Effects of Automotive Emissions on Susceptibility to Respiratory Infections
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 519
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 520
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 521
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 522
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 523
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 524
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 525
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 526
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 527
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 528
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 529
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 530
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 531
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 532
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 533
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 534
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 535
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 536
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 537
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 538
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 539
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 540
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 541
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 542
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 543
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 544
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 545
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 546
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 547
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 548
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 549
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 550
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 551
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 552
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 553
Suggested Citation:"Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust." National Research Council. 1988. Air Pollution, the Automobile, and Public Health. Washington, DC: The National Academies Press. doi: 10.17226/1033.
×
Page 554

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.

Assessment of Carcinogenicity: Generic Issues and Their Application to Diesel Exhaust DAVID G. KAUFMAN University of North Carolina Mechanisms of Carcinogenesis Relevant to Assessment of Mobile Source Emissions / 520 Experimental Models in Chemical Carcinogenesis / 520 Role of DNA Replication and Repair / 522 Genetic Effects of Carcinogen Damage to DNA / 522 Atypical Carcinogens / 524 Promotion, Cocarcinogenesis, and Enhancement / 525 Multistep Processes / 526 Variations in Susceptibility / 527 Metabolic Conversion and Carcinogen Activation / 530 Qualitative Assessments of Carcinogenicity / 530 Epidemiologic Evaluation / 530 Bioassays in Experimental Animals / 531 Short-Term Tests in Vivo and in Vitro / 533 Methods for Quantitative Extrapolations to Human Cancer Risk / 534 Estimation of Quantitative Risk in Laboratory Animals / 534 Extrapolations Among Species / 535 Extrapolations Among Routes of Administration or Exposure / 535 Extrapolation to Dose Levels of Human Exposure / 537 Experimental Evidence on Carcinogenicity of Diesel Exhaust / 539 Short-Term Tests of Activity of Diesel Emissions / 540 Data on Carcinogenic Activity of Diesel Exhaust Emissions / 541 Quantitative Assessment of the Cancer Risk of Diesel Exhaust in Humans / 543 Summary / 546 Summary of Research Recommendations: Priorities, Purposes, and Responsibilities / 547 Summary of Research Recommendations: A Research Plan / 549 Air Pollution, the Automobile, and Public Health. (it) 1988 by the Health Effects Institute. National Academy Press, Washington, D.C. 519

520 Assessment of Carcinogenicity Despite our limited understanding of carci- nogenesis, practical concerns in the "real world" confront us with the need to assess the potential significance of diesel exhaust as a human carcinogen. Such an assessment requires progressing from fragmentary the- oretical insights into the process of carcino- genesis to estimates of the human risk posed by diesel exhaust. Confounding this effort is the fact that diesel exhaust is an imprecisely characterized and inconsis- tently constituted product composed of chemicals that may trigger carcinogenesis individually, cooperatively, or even se- quentially. Researchers are now confronting the dif- ficulties of understanding the etiology and pathogenesis of multifactorial, multistep disease processes, and they are just begin- ning to recognize general principles that may operate in most typical cases of cancer. There is awareness of the relationship be- tween the dose of carcinogens and the resulting tumor response, and recognition of the importance of the metabolism of a carcinogen into reactive intermediates that may cause damage. Cellular mechanisms such as DNA replication may provide op- portunities for carcinogens to transform genetic information, and targets in DNA may include specific genes or sites at which chromosomes are prone to breakage. En- hancing factors, such as promoters, may increase the likelihood of cancer develop- ment. Variations in human susceptibility to cancer make evaluation of the activity of specific carcinogens difficult, although it is clear that certain human tissues or certain individuals are more susceptible to cancer than others. Certain familial tendencies or acquired illnesses are also thought to pre- dispose people to cancer. In this chapter, the evidence on the car- cinogenicity of diesel engine exhausts and the methods used to make quantitative risk estimates from these data are evaluated. Specific evidence concerning carcinogen- esis of diesel exhaust in experimental sys- tems is reviewed, and relationships be . . , . . . tween t. his Information anc . reviews In other chapters are identified. Current knowledge as well as areas of ignorance influence efforts to estimate human risks by extrapolation from the experimental data on animals. A discussion of these issues serves as an outline for making such esti- mates in the future. Mechanisms of Carcinogenesis Relevant to Assessment of Mobile Source Emissions Chemical carcinogenesis is a very complex topic. Thus, this review is selective in its consideration of carcinogenesis, focusing on several general concepts rather than on specific details. The constructive role of studying cancer development in animal models is considered, and certain aspects of the general principles operating in most typical cases of carcinogenesis are exam- ined. The review also touches on unusual cases that appear not to fit the typical pattern of cancer development. It considers the evidence for and the problems associ- ated with evaluating a disease that develops as the result of a multistep process. Finally, the factors that define individual variations in susceptibility are discussed, and features of carcinogen metabolism and translocation are reviewed. Experimental Models in Chemical Carcinogenesis Experimental animal models have been employed to reproduce tumors of the his- tologic types and organs of origin that commonly occur in humans. Such models permit direct experimental study of factors that influence the development of the most common cancers in humans and the mech- anisms of action of particular carcinogens. Examples of valuable animal models and their applications are listed in table 1. Some unique insights have been derived from comparisons of the properties of animal tissues in which the tumor response is a good model for the human disease, to tissues of other species in which the re- sponse is very different from that of hu- mans. Studies of particular tissues have been facilitated by using organ cultures (Saff~otti

David G. Kaufman 521 Table 1. Examples of Valuable Animal Models of Human Carcinogcncsis Rodent Species Type of Human Cancer Modeled References Application L)osc/ response Metabolic mechanisms Hormonal influences Dietary influences Hamster Rat Rat Mouse Lung Pancreatic Breast Colon Skin Saff~otti et al. (1968) Pour (1984) Scarpelli ct al. (1984) Muggins et al. (1961) Gullino et al. (1975) Ward et al. (1973) Rcddy et al. (1974) Berenblum and Shubik Promotion (1 947) and Harris 1979~. In this technique, pieces of intact tissue representative of the sam- pled organ are grown in culture. Many features of the tissue that exist in viva, including the interrelationship between the epithelial components and the supportive cells, are preserved. Such cultures can be used to assess morphological features, mac- romolecular synthesis, and responses to hormones as well as capacity to metabolize carcinogens and to repair DNA damage. Use of organ cultures has been a principal approach used for analysis of properties related to carcinogenesis in human tissues. Some of the attractive features of organ cultures for example, their maintenance of natural relationships between epithelial and supportive cells are related to some of their major shortcomings. In contrast to cell cultures, organ cultures cannot be propagated, and material from an individ- ual human subject is rapidly exhausted. Cell culture overcomes this limitation be- cause cells may be propagated in culture. However, the very process of propagation exerts a selective pressure, and the cell type that emerges may be unlike that predomi- nating in the intact tissue. Nevertheless, isolated cells have proven very useful in the study of common and unique features of carcinogenesis under far more controlled environmental conditions than is possible . . . In an intact an1ma. I. Although direct experimentation with the objective of inducing carcinogenesis is clearly unethical in humans, a broader, deeper information base is needed on the properties of human cells and tissues that relate to carcinogenesis. This goal has been approached by undertaking culture studies of human cells and tissues obtained at im- mediate autopsies or from surgical speci- mens (Harris and Trump 1983~. Studying the properties of human tissues in vitro allows examination of the human diversity in cancer development. For exam- ple, in vitro techniques can be used to explore the individual variability in metab- olizing carcinogens, repairing DNA dam- age, responding to various hormones, and perhaps even to determine the degree to which various nutrients serve as cofactors . . . In carcmogenesls. Although in vitro carcinogenesis with human cells in culture is rather new, trans- formation of normal cells to neoplastic ones has been accomplished with a number of cell types. Results from such studies permit the direct comparison of the stages in the presumed multistep process of carcinogen- esis in humans and in animals. For exam- ple, the apparently greater difficulty in transforming human cells than animal cells may parallel the comparative susceptibility to cancer of these various species. If the determinants of the various stages in carci- nogenesis are successfully characterized in human cells, it may be possible to develop improved methods for early detection of preneoplastic or early neoplastic lesions. Some human tissues have been main- tained as viable xenotransplants in nude mice (Valerio et al. 1981~. Such models are an ethically acceptable method for in viva study of the process of carcinogenesis in human tissue (Shimosato et al. 1980~. This

522 Assessment of Carcinogenicity model may provide for direct comparisons of features of carcinogenesis between hu- mans and experimental animals that are commonly used in bioassays. Such infor- mation would clearly be valuable in de- termining the risks to humans of agents demonstrated as carcinogenic in animal bioassays. Role of DNA Replication and Repair It is a well-recognized clinical observation that cancer typically occurs in tissues that have a high rate of cell proliferation or in tissues in which cell proliferation occurs in response to injury. Conversely, cancer is extremely rare in adult tissues or cell types in which cell proliferation does not occur. It was the opinion of classical pathologists that chronic irritation or injury was the etiologic factor for the development of cancer. Subsequently, a variety of specific carcinogenic etiologic agents have been rec- ognized. Nonetheless, cell proliferation plays a significant role in the evolution of cancers (Grisham et al. 1983~. This is well illustrated in the case of liver cancers in- duced in rats by chemical carcinogens. Typical liver carcinogens at effectiv'e doses are also hepatotoxic, and they induce re- storative hyperplasia to replace cells lost as the result of the toxicity. The influence of cell proliferation as a contributing factor in the development of cancer presumably results from effects on the mitotic process and on DNA synthesis. Replicating DNA is vulnerable for a variety of reasons. First, replicating DNA is af- fected to a greater extent by chemical car- cinogens than is nonreplicating DNA (Cordeiro-Stone et al. 1982~. Second, rep- lication of DNA that contains carcinogen adducts may cause incorporation of incor- rect nucleotides at sites of altered or excised bases. Third, some carcinogens may mod- ify nucleotide precursors, and altered pre- cursors may be incorporated into DNA. Fourth, DNA replication itself occurs with a low, but nonzero, error rate. Situations that increase cell replication are likely to cause mutations strictly as the result of these errors. Mammalian cells have a number of mechanisms to repair DNA damage and to reduce the likelihood of errors during DNA replication. Treatments of cells or animals with chemical carcinogens or radi- ation cause the onset of DNA repair pro- cesses. In studies in which cell proliferation has been inhibited and DNA repair has been allowed to remove some or most carcinogen-induced DNA adducts, the transforming effects of the carcinogen dam- age have been reduced (Ikenaga and Kaku- naga 1977~. In contrast, in patients with defective DNA repair processes, such as the genetically determined syndrome known as xeroderma pigmentosum, increased inci- dences of tumors have been observed (Setlow 1978; Hanawalt and Sarasin 1986~. Thus, DNA repair processes appear to be protective against tumor development, whereas defects of DNA repair appear to be associated with increased risks of cancer. There appears to be a critical interrela- tionship between the repair and replication of DNA as factors in the etiology of cancer (Kakunaga 1975~. If DNA replication pro- ceeds within a damaged region prior to repair, there is a substantial risk of error- making during replication, which may cause a mutation to occur as the result of alteration of the base sequence of the com- plementary DNA strand. Of course, this does not occur if the repair of the damage precedes replication. Consequently, the re- lationship in time of the repair and replica- tion of DNA may be a major determinant of the potential for the occurrence of mu- tations and also oresumablv. of carcino- genes~s. , ~,, Genetic Effects of Carcinogen Damage to DNA Chemical carcinogens have been shown to produce a variety of types of DNA damage that can lead to genetic effects on cells (table 2) (Sarma et al. 1975; Drake and Baltz 1976; Singer and Grunberger 1983~. Point muta- tions and frameshift mutations can alter the regulatory or coding regions of genes. On a larger scale, carcinogens can directly af- fect chromatics and chromosomes (Evans 1983~. By still unknown mechanisms, car- cinogen damage can cause the exchange of

David G. Kaufman 523 Table 2. Genetic EEects of Carcinogen Damage to DNA Point mutations transitions and transversions Frameshift mutations small deletions or additions Mutations at "hot spots" Chromosomal breakage at "fragile sites" Recombinations and rearrangements Sister chromatic exchanges Translocations of portions of chromosomes Gene amplification Aneuploidy DNA segments between sister chromatics, and chromosomal breakage that leads to large deletions or transposition of chromo- somal segments to other chromosomes. Pre- sumably, such damage may lead to failures of mitotic division with unequal distribution of chromosomes between daughter cells, result- ing in abnormal DNA content. DNA dam- age is also thought to be one mechanism for the amplification of segments of DNA. The significance of many or all of these forms of damage to DNA does not concern the chemical composition of this molecule but rather its content of genetic informa- tion. Valuable insights about these genetic effects, particularly with regard to onco- genes, have arisen from recent studies in viral carcinogenesis and molecular biology. Investigations of the mechanism of cell transformation by oncogenic retroviruses have shown that their transforming genes, designated as oncogenes, are derived from the coding regions of cellular precursor genes known as proto-oncogenes (Bishop 1983~. Proto-oncogenes are believed to play an important, though as yet unknown, role in normal cellular function or differen- tiation because they are highly conserved in widely divergent species from yeast to humans. Recent studies have shown that proto-oncogenes can acquire transforming activity as the result of genetic alterations that affect their DNA sequence or place them under abnormal genetic regulation by chromosomal rearrangements, insertion of promoters, or gene amplification (Wein- berg 1985; Barbacid 1986~. The number of known retroviral oncogenes is quite lim- ited about two dozen. Even when the proto-oncogenes from which they are de- rived and the closely related cellular genes (for example, N-myc and N-ras) are added to the sum, the total of retrovirus-related oncogenes is still small. Although further studies of human and animal tumors have identified additional genes with transform- ing activity, it is not yet possible to esti- mate the number of cellular genes that have transforming activity induced by genetic alteration. It is well known that mutations occur at exceptionally high rates at specific sites in DNA of viruses and other prokaryotic or- ganisms. This nonhomogeneous effect is recognized for spontaneous mutations as well as mutations induced by radiation or chemicals. The location of these so-called "hot spots" relates to the specific form of radiation or chemical carcinogen that in- duces the mutations. DNA sequence as well as the structural features of DNA, including bending and association with proteins, appears to influence the spectrum of hot spots. Clearly the DNA sequence in higher organisms such as mammals is far less completely defined, and the means for cataloging the spectra of hot spots in DNA of these organisms are very limited. None- theless, some evidence suggests that there are sites selectively affected by carcinogens where mutations occur at high frequency. Fragile sites are locations in chromo- somes that are particularly prone to break- age. When cell growth conditions are al- tered, such as through deprivation of thymidine and folic acid, chromosomes have been found to break consistently at the same sites. These sites are closely related to sites where chromosomal rearrangements occur in human cancers (Yunis and Soreng 1984), suggesting that structural peculiari- ties that make these sites prone to breakage may be important factors in the develop- ment of cancers. Another notable point is the chromosomal location of these fragile sites relative to several of the known proto- oncogenes. Although the power of the scientific methods used to compare the locations of these sites is not great, the apparent statistical relationship within the experimental error of the methods suggests that some very important feature of cancer development is related to the structure of DNA at these sites.

524 Assessment of Carcinogenicity Techniques for identifying subregions (bands) within chromosomes now allow abnormal chromosomes in cancer cells to be examined with far greater resolution and specificity than previously possible. Sur- veys of the chromosomal banding patterns of a wide spectrum of cancer cells have shown some consistent patterns of chro- mosomal abnormalities for many different types of cancer (Sandberg 1983; Mitelman 1986~. For some cancers- for example, Burkitt's lymphoma there is a very high degree of consistency in the type of alter- ation observed. Most Burkitt's lymphomas show balanced translocations of portions of specific chromosomes. In other cases, such as the development of the Philadelphia chromosome (loss of a portion of the long arms of chromosome 22) in chronic my- elogenous leukemia, the appearance of the chromosomal abnormality accompanies the chronic phase of the disease. Another very common feature of cancer cells is the development of aneuploidy, with cells having more or less than the normal diploid number of chromosomes or an abnormal DNA content. In fact, the large size and the hyperchromaticity char- acteristic of cancer cell nuclei are largely due to the increased DNA content of typ- ical aneuploid cells. Aneuploidy is pre- sumed to develop as a consequence of unequal mitotic divisions during the evolu- tion of cancer cells. The presence of abnor- mal mitotic figures is one feature of cancers used to arrive at a pathological diagnosis. One of the consequences of the abnormal chromosomal content of cancer cells is that particular genes are present in low or high copy number. One can speculate how the loss of a normal inhibiting function can occur with the loss of a chromosome in hypodiploid cancer cells. Hyperdiploid cells can greatly overexpress particular gene products, or they may generate insufficient inhibitory activity to balance the high copy number of some cancer-related gene. Atypical Carcinogens A number of substances very different from the typical chemical carcinogens have been shown to be carcinogenic in humans and in experimental animals. With atypical carcin- ogens, carcinogenesis can be induced by physical agents and chemicals that do not directly alter DNA. The differences be- tween these atypical carcinogens and the common carcinogens challenge the classical concepts of carcinogenesis and demand the development of theories of carcinogenesis that can include their mode of action. For some time, asbestos has been recog- nized as carcinogenic, first in humans (Doll 1955) and later confirmed in experimental animals (Wagner et al. 1973~. When di- rectly instilled into the pleural cavity of experimental animals it has been shown to produce tumors like those that follow as- bestos exposure in humans (Wagner et al. 1973~. The critical property of asbestos best associated with carcinogenicity is the phys- ical dimensions of fibers (Stanton and Wrench 1972) rather than the chemical composition of the asbestos or the sub- stances adsorbed on it. This was confirmed by showing that glass fibers, prepared in length and width comparable to asbestos fibers, were also carcinogenic. The cellular response to asbestos fibers and other foreign bodies involves the for- eign-body inflammatory reaction wherein the fibers are surrounded by macrophages and fibroblasts (Brand et al. 1975~. Current hypotheses suggest that the inflammatory cells or epithelial cells produce reactive forms of oxygen molecules which may affect the DNA of the epithelial cells, and this damage is fundamental to the carcino- genic process. Others suggest that asbestos acts by affecting chromosomal segregation during mitosis. On a practical level, asbes- tos is relevant to the topic of mobile source emissions. It is known that asbestos expo- sure is associated with mesothelial cancer in humans. However, in individuals in whom asbestos exposure is combined with ciga- rette smoking, the risk of cancer is greatly increased, and the leading type of cancer is bronchogenic carcinoma (Selikoff et al. 1968~. It is conceivable that individuals who have been exposed to asbestos will represent a group at increased risk from the combined effects of asbestos and mobile . . source em1sslons. A number of studies have shown that

David G. Kaufman 525 unusual substances, functioning as atypical carcinogens, can produce cancers in exper- imental animals. Plastic films have been shown to produce tumors when implanted into animals. However, when the films were sufficiently fenestrated, or when they were ground to a powder, the material was not carcinogenic. Several metals, in the forms of ores, refinery process by-prod- ucts, and ions and salts, have been shown to be carcinogenic in humans or experimental animals (International Agency for Research on Cancer 1980~. Examples include various forms of arsenic, chrome, and nickel. A number of reports in recent years have noted that chemicals, including therapeutic agents that cause proliferation of peroxi- somes, are carcinogenic (Ready et al. 1980~. Unlike chemicals such as phenobar- bital, these agents appear to function as complete carcinogens rather than just as promoters. Investigations of examples of this class of chemicals have shown that they do not form adducts with DNA. Several other chemicals and drugs are carcinogenic in animals or humans, but are not known to interact with or form adducts with DNA. Among these are agents that affect enzymes involved in the metabolism of DNA precursors or that more directly affect DNA precursor pools. These prop- erties make some of these chemicals effec- tive therapeutic agents for treating cancer. The action of some of these agents is be- lieved to be a consequence of imbalances in DNA precursor pools, which also cause mutations. Promotion, Cocarcinogenesis, and Enhancement . Exposure to carcinogens is not the only determinant of cancer development. Other substances or other processes can influence the risk for cancer development, particu- larly when they complement exposures to carcinogens or act on animals that have a high spontaneous tumor incidence. These factors must be recognized when the obser- vational data derived from carcinogenicity tests in animals are being interpreted mech- anistically. The terms "enhancers" and "enhancement" describe effects that include those typically classified as promoters or cocarcinogens but without attribution of a mechanism of action. Promotion is defined operationally, based on classical experiments in which tumors were induced in mouse skin by a two-step treatment protocol (Berenblum 1975~. The first treatment involved the application of a subcarcinogenic dose of a strong carcinogen to the mouse skin, fol- lowed by a prolonged series of applications of a noncarcinogenic agent. The combined treatments produced a tumor response, whereas the same dose of the carcinogen or the second agent, which has come to be known as a promoter when used alone, was ineffective or vastly weaker. The two steps of the treatment process were designated as . . . . . . nltlatlon anc . promotion, ant . have come to be interpreted as separate events or pro- cesses in the evolution of cancers. In contrast to the separate application of . . . . initiator anc promoter, cocarclnogens are agents that enhance the development of cancers when administered concurrently with a carcinogen. Cocarcinogens act through a variety of mechanisms. They may modify the metabolism of carcinogens to yield a greater quantity or proportion of ultimate carcinogenic metabolites. They may act by causing cell or tissue toxicity with accompanying accelerated cell prolif- eration; this in turn may increase the risk of malignant transformation. They may also act by interfering with normal defense mechanisms that function to counteract the detrimental effects of carcinogens. Enhancing or inhibiting effects from ex- posure to a wide variety of substances (for example, certain constituents of mobile source emissions), not just exposure to car- cinogens (such as the possibly carcinogenic constituents of such mobile source emis- sions), determine the tumor response. Our understanding of these effects and the inter- actions between substances is very limited. More-specific enhancing effects, in some cases affecting individual tissues, may come from exposures to noncarcinogens that ex- ert a promoting or cocarcinogenic effect. Individual genetics, prior or concurrent medical conditions, and diet all may con- tribute to an individual's specific risk from

526 Assessment of Carcinogenicity a given level of exposure. Such enhancers presumably increase the effects of other carcinogens. It is particularly important to ascertain whether mobile source emissions contain constituents with enhancing activ- ~ty. Mobile source emissions may represent a serious public health problem if they en- hance carcinogenesis initiated by other exposures. This is an important general problem that requires further attention. Different methods of bioassay from those used to detect carcinogens will have to be developed to determine whether these emissions have enhancing activity. Such methods are needed to explore the possibil- ities that enhancing activities are specific in augmenting the activity of particular types of carcinogens or that their activity differs according to tissue sites at which they act. It is clear that standard carcinogenicity bioassavs are not cure tests for either can cer-~nit~at~ng activity or activity as a com- plete carcinogen. They are phenomeno- logic studies that associate excess cancers with particular treatments, but they do not indicate the mechanism by which the can- cers are produced. Particularly in the case of tumors in tissues with a high spontane- ous tumor incidence in untreated animals, increases in the incidence of tumors may reflect a toxic or promoting activity of the tested chemical. If this effect is not the result of toxicity associated with high exposure levels, then the result may be a demonstration of pro- motion activity. Such a conclusion might distinguish these compounds from stan- dard carcinogens, but it does not indicate that they are without risk. In view of the hazard posed by chemicals of this type, it is important to develop methods to demon- strate how they cause tumors. Since some of the constituents of diesel exhaust may also have this kind of activity, it would be useful to have the means of identifying and quantitating these chemicals. Promoting or enhancing activity may involve a number of organs and tissues other than the skin. For this reason it will be necessary to evaluate the differences in enhancing effects in different tissues. For example, 12-0-tetradecanoylphorbol-13 acetate (TPA) is a good promoter for mouse skin, but is not a good promoter for rat liver; conversely, phenobarbital is very effective in rat liver, but not in skin. The possibility also exists that enhancing activ- ity relates to the type of initiator. The type of initiator may determine which tissues or organs will be sensitized to promoter or enhancer effects. It is entirely conceivable that the broad spectrum of chemicals in diesel exhaust contains enhancing agents with dis- tinctive patterns of organ selectivity. To test these hypotheses it will be neces- sary to examine the enhancing activity of materials such as diesel exhaust following an initiating treatment with any of a variety of carcinogens with a range of organ spec- ificities. In this manner it may be possible to develop a standard panel of animal test models that would have the ability to detect and quantify promoters and enhancers that are active in any of a number of tissues. · Recommendation 1. . . . . The role of pro- moters and enhancers In human carcinogen- esis should be determined. Multistep Processes On the basis of a variety of clinical and experimental evidence, the development of cancer is believed to be a multistep process (Armitage 1985~. Clinical experience has shown that the incidence of most cancers rises with age and most are seen to pass through premalignant stages prior to the development of clinically overt cancer (Doll 1971~. The most thoroughly studied case is that of the multistep evolution of squamous cell carcinoma of the uterine cervix. The validity of a multistep interpre- tation is attested to by the fact that clinical intervention at an early stage vastly reduces the incidence of the overt, invasive tumors of this type. A similar sequence of prema- lignant lesions of the bronchial epithelium precedes invasive lung cancer (Auerbach et al. 1961~. One study followed uranium miners with repetitive sputum cytologies for many years (Schreiber et al. 1974~. Progressive cnanges In Cyrologlc nnalngs proceeded from squamous metaplasia through various stages of dysplasia, in situ . ~. . . . . .. ..

David G. Kaufman 527 . . . . . carcinoma, and Invasive carcinoma over the course of several years. Subsequently, comparable sequences of epithelial lesions have been found to precede overt cancers in a number of sites (Farber 1984~. Initiation and promotion in the mouse skin bioassay is an example of carcinogen- esis as a two-step process (Berenblum 1975~. Comparatively recent studies have shown that the process of promotion itself can be divided into stages (Slaga et al. 1980~. In the case of the evolution of tu- mors of rat liver, cancer is believed to be the end result of a process in which foci or areas of enzyme-altered hepatocytes and neoplastic nodules precede malignant tu- mors (Farber 1980~. In the respiratory tract of hamsters, carcinogen treatment has been found to cause a progressive sequence of histologic alterations that culminate in invasive, malignant tumors (Saff~otti and Kaufman 1975~. These lesions demonstrate a spectrum of morphological changes very close in appearance to the lesions of the respiratory tract seen in humans. In fact, the evolving lesions shed cells analogous to those observed in the cytology preparations from the uranium miners cited above (Schreiber et al. 1974~. Methods to study the transformation of cells by chemical carcinogens in tissue cul- ture have been available for about two decades. These studies first were successful in rodent embryo and fibroblast cells. More recently, similar results have been achieved using rodent epithelial cells, and in the past few years human cells have also been trans- formed with chemical carcinogens. A num- ber of morphological, biological and phe- notypic changes have often been observed in these in vitro transformation systems as the cells progress from the original cell population to demonstrably malignant cells. In some cases, for example in studies using Syrian hamster embryo cells, a spe- cific sequence of changes in the culture has been linked to malignancy (Barrett and Ts'o 1978; Smith and Sager 1982~. With cultured rat tracheal epithelial cells, a se- quence of progressive changes in the bio- logical behavior of carcinogen-treated cells has been observed (Nettesheim and Barrett 1984~. In this system, the cultured cells can be evaluated for their relationship to the morphological alterations observed in vivo by allowing them to repopulate a trans- planted rat trachea which had been deepi- thelialized (Klein-Szanto et al. 1982~. Other evidence of the multistep nature of trans- formation found in this system is a two . . . . . . . step transformation envoy Ding an ~n~t~at~ng carcinogen treatment followed by in vitro promotion with TPA (Steele et al. 1980~. Studies of the transformation of human cells in vitro have also shown that several distinct alterations occur consistently and in a generally similar order (Kakunaga et al. 1983). Variations in Susceptibility Rates of development of spontaneous be- nign and malignant tumors vary in animals of different species and strains (Grasso and Hardy 1974~. Some animals are highly re- sistant to tumor development, and even after a long life few will die with tumors. In contrast, some species and strains of ani- mals have a very high incidence of cancers, in some cases 100 percent. In these species and strains, the type and quantity of these background tumors are characteristic of the animal and are presumed not to be the result of unusual exposures to environmen- tal factors. Among the animals typically chosen for carcinogenesis bioassays, mice have an exceptionally high incidence of liver tumors, particularly in males. In fe- male rats, mammary tumors are very com- mon. Treatment of these animals with chemical carcinogens results in tumors at various locations and of types that depend on the activity and dose of the carcinogen as well as the route of administration and other factors. Commonly, these treatments also affect the incidence and multiplicity of the tumors characteristic of the untreated animals, indicative of the sensitivity of these tissues to transformation. The human population, in comparison, appears to have a relatively low back- ground level of cancer, as determined from cancer incidence data for certain low-risk groups in underdeveloped nations or in specific populations such as Mormons or Seventh Day Adventists in the United

528 Assessment of Carcinogenicity Table 3. Factors Affecting Human Susceptibility to Carcinogenesis E. xposures to carcinogens Diet composition and nutritional status Personal habits including cigarette smoking and alcohol consumption Determinants of geographic variations in cancer development Genetic diseases or heterozygous carrier states Acquired illnesses and infections Unknown factors determining familial . . . prec .lsposltlons Variations in metabolic activation or inactivation of carcinogens States in which there are religious restric- tions on smoking or certain dietary prac- tices. Despite the low overall cancer rates in these groups, certain cancers are seen and these may represent the background tu- mors of humans. These include leukemias and lymphomas, soft-tissue sarcomas, skin tumors, and a low rate of tumors of several epithelial tissues. Above this background, several factors appear to affect the suscep- tibility of humans to the development of cancers (table 3~. The incidence in humans of tumors of various organs differs by country and even by population group within countries. In the United States, cancer of the lung is the most common significant cancer in males and females, whereas in Egypt and Japan, cancers of the urinary bladder and stomach, respectively, are the most common. Within the United States, there appear to be geo- graphic differences in incidence of tumors of various organs (Pickle et al. 1987~. Clearly, a large proportion of these tumors are induced rather than spontaneous and are ~ . . . . 01 environmental orlgln. In contrast to the general population, there are individuals who are genetically predisposed to the development of cancers. Examples of genetic diseases associated with a high incidence of cancer include xeroderma pigmentosum, ataxia telangiec- tasia, familial retinoblastoma, Fanconi's anemia, Gardner's syndrome, familial polyposis coli, and many others. Studies have shown that for the recessively inher- ited genetic disease ataxia telangiectasia, close relatives who do not have the disease, but are heterozygous carriers, also have an elevated cancer risk (Swift et al. 1976~. In fact, the heterozygous carriers of the ataxia . . . te anglectasla trait may represent up to a few percent of the human population. At present, the biological basis for these ge- netic diseases and their link to cancer are unknown. However, it is known that there are defects of DNA repair functions, pre- sumably different defects, for xeroderma pigmentosum, ataxia telangiectasia, and Fanconi's anemia. Familial retinoblastoma has been shown to be consistent with a deletion or mutation of chromosome 13. Familial polyposis cold and Gardner's syn- drome are associated with abnormalities of cellular growth control. These observations provide clues to pos- sible steps in the presumed multistep pro- cess of malignant transformation. To the extent that carrier states for these diseases are common in the population, these ge- netic traits may be factors that influence individual risk for developing cancer (Swift et al. 19764. It is likely that other genetically determined factors may influence cancer risks even if they do not yield recognized genetic diseases. For example, there may be genetically determined influences on the rate or route of metabolism of chemicals. The racial differences in alcohol metabo- lism illustrate that such differences occur in the human population. Individual varia- tions in other factors, such as those affecting responses to injury, may also influence cancer risk. Knowledge about such factors is limited at present but may be an impor- tant and useful area for continued research. There are a variety of illnesses and infec- tions that predispose people to the devel- opment of cancer. For examples, hepatitis B and schistosomiasis of the bladder are important factors in the causation of liver and bladder cancers, respectively. Certain lymphomas are associated with infectious diseases (for example, human T-cell lym- photrophic virus types I and III, or Epstein- Barr virus), and colon cancers are associ- ated with ulcerative colitis. These diseases and conditions cause a high level of cell proliferation in specific target cell popula- tions which may predispose to cancer de- velopment in the affected tissue.

David G. Kaufman 529 Some tumors of the lung are associated with scars of the parenchyma. There has also been speculation that other lung con- ditions predispose people to lung cancer development (Kuschner 1985~. Although it is likely that these conditions predispose people to lung cancer because of increased cell proliferation rates, it is also possible that these conditions affect the capacity of the lung to clear exogenous materials, in- cluding potential environmental carcino- gens. Omitting the known genetic diseases that predispose to cancer development, and in the absence of acquired diseases that are associated with cancer, there are still a number of families with an unusually high incidence of cancer. In most of these fami- lies there is variable penetrance of tumor risk with less than 100 percent incidence of cancer in these populations. In some cases there are distinct patterns of tumor devel- opment with particular organs affected to unusual extents and with different tumors predominating in males and females. It is unclear whether these occurrences are pri- marily the result of unrecognized genetic diseases or a heterozygous carrier state for a recessive genetic disease. Alternatively, these families may develop these cancers because of elusive environmental factors passed socially from generation to genera- tion, such as diet or personal habits. Clinical and experimental evidence sug- gest that there are important differences in susceptibility to cancer among individuals in the human population. This could be a very significant factor in efforts to control specific types of cancer, including any re- lated to exposure to diesel exhausts. The population of susceptible individuals may account for a disproportionate share of particular types of cancer. It may be pos- sible to significantly change the overall in- cidence of specific types of cancer by identifying susceptible individuals and con- centrating cancer prevention activities on this population. It might be possible to identify individuals for whom specific types of carcinogens or diesel exhaust rep- resent a particular hazard and protect them from such hazards. We have limited knowledge of biological and enzymatic factors that determine these states of unusual susceptibility. More data are needed about the range of variation of metabolic processes, DNA repair pro- cesses, constitutive and induced cell prolif- eration rates, and responsiveness to hor- mones in tissues from human subjects. These data are needed for each tissue in which cancer is common, and this informa- tion should be obtained where possible to determine the variability according to age, gender, genetic background, and personal factors such as diet, therapeutic drug use, and personal habits (for example, cigarette smoking). Accomplishing these goals will require development of methods to obtain human tissues in an acceptable manner. Furthermore, human subjects will have to be chosen scientifically so that they are representative of the population as a whole or the subpopulations that appear to be at unusual risk. If these studies are successful, the next step will be development of meth- ods to test these characteristics in samples of tissue that can be obtained from normal individuals with little or no risk. On the basis of epidemiologic observa- tions that there is a range of responses within populations apparently exposed to the same levels of a carcinogenic substance, it is conceivable that there are individual factors that are major determinants of risk. Identifying the portion of the population at exceptional risk and concentrating protec- tive efforts on that population might have a major impact on the overall cancer inci- dence at particular tissue-specific sites. Such an approach has proven notably effec- tive in reducing myocardial infarction rates in individuals with genetic abnormalities of low-density lipoprotein metabolism and in individuals with acquired coronary artery disease. Developing methods to determine the elements of individual risk will require great attention. The development of appro- priate and acceptable methods for obtain- ing cells from various body sites with little or no risk should be included in this method. Methods to test various cellular characteristics that have been associated with the development of cancer will also have to be devised. Such factors as the

530 Assessment of Carcinogenicity capacity for carcinogen metabolism, DNA . . . repair, oncogene activation, or c ~romosom- al breakage might all be included among the individual characteristics evaluated. It will be particularly important to determine how to scale methods down to fit the number of cells that can be made available for testing. The availability of monoclonal antibodies, enzyme-linked immunosorbent assay (ELISA) techniques, in situ nucleic acid hybridization, chromosomal banding methods, and flow cytometry offer many new and unexplored approaches to deter- mining these individual characteristics with- in the numbers of cells that may be obtained from persons without demonstrable disease. · Recommendation 2. Methods should be developed to identify individuals at high risk. Metabolic Conversio,2 and Carcinogen Activation Many chemical carcinogens are inert and unreactive in their native form. Cells have a wide variety of metabolic enzymes which may inactivate and modify endogenous as well as exogenous chemicals, including those that are potentially harmful. In many cases these metabolic processes increase the water solubility of these chemicals as a way of improving the body's ability to excrete them. The process of metabolic alteration proceeds via a number of reactions which often produce a variety of products. Some of these products are reactive and interact with cellular constituents to pro- duce damage (Miller 1970~. An example is benzo~a~pyrene (BaP) metabolism: sponta- neously reactive diol-epoxide intermediates are formed as minor products whereas the . . . . . majority are converter . Into unreactive ex- cretion products. The diol-epoxides are the major, and perhaps the exclusive, proxi- mate carcinogenic forms of the parent car- cinogen. Additional examples are provided in chapters by Hecht and by Marnett (both in this volume). Metabolic processes are highly regulated and responsive to the environmental con- ditions of the organism and their consti- tuent cells. In most cases the metabolic pathways are induced or activated by sub- strates. At the same time, many of the substrates metabolized by the same enzyme systems compete for metabolism, particu- larly before inductive mechanisms increase metabolic capacity. Since these metabolic processes use molecular oxygen, NADH, or other cofactors in the reactions, factors that influence the supply of these cofactors may also modify metabolic rates. These factors include oxidative poisons, alcohol, dietary constituents, and a variety of drugs. Some carcinogens, even if administered systemically, produce tumors in specific target tissues (Merletti et al. 1984~. For most of these chemicals the organ speci- ficity is a manifestation of the metabolic properties of the particular tissue. For ex- ample, certain N-nitrosamine compounds with propyl, butyl, or phenyl substituents produce tumors in the esophagus or uri- nary bladder of rats with great selectivity. In contrast, diethylnitrosamine principally produces tumors in the liver and lungs of rats, and dimethylnitrosamine under ap . . . . . . proprlate cone .ltlons primary y proc uces tu- mors of the liver and kidneys. Differences in specificity are presumed to relate to the absence of certain enzymes in particular species or organs where some of these compounds are ineffective. In other cases, particular metabolic pathways may show exceptional affinity for specific forms of a given class of compounds in a partic- ular tissue and lesser activity for other forms. This results in large rate differences in metabolism that correlate with the spec- trum of distribution of tumors. Qualitative Assessments of Carcinogenicity Epidemiologic Evaluation The most powerful and convincing evi- dence for the carcinogenicity of a substance is the demonstration of its association with an excess of cancer in epidemiologic stud- ies. Epidemiologic studies vary in type and design, with different levels of scientific power, but all seek to compare the rates of cancer in a defined test group versus that in

David G. Kaufman 531 O . a control group. These studies may be broadly based, as in the case of geographi- cally based investigations of differences in cancer incidence, or they may focus on large population groups such as those of a certain religion with particular constraints on foods and habits, or groups with partic- ular ethnic or genetic backgrounds. Studies of this type are likely to provide only the most general information. More-specific information is derived from experiments that focus more narrowly on particular chemicals or occupational conditions of ex- posure. The general approach in epidemiologic studies involves a comparison of the risk for the development of cancer in a defined experimental group to that in an appropri- ate and scientifically defensible control group (or two or more control groups) (see Bresnitz and Rest, this volume). Trends in cancer incidence may also be determined by comparisons between cohorts selected as a function of graded intervals of time of exposure or rates of exposure. Trends of cancer incidence with time or extent of exposure support an association between the trend-related factors and cancer devel- opment. Statistically significant differences in cancer incidence between experimental and control groups indicate probable, causal relationships to cancer development. The extent of the increase in incidence and the strength of the statistical signifi- cance are used to weigh the importance of the relationship. In cases where there is good epidemiologic evidence linking an environmental exposure to cancer develop- ment and there are good quantitative data on exposure of the human subjects, it is possible to make reasonable predictions of the risks associated with future exposures at similar dose levels. Such information is available for very few chemicals or envi- ronmental exposures. In most cases, evi- dence for carcinogenicity is derived from experimental data obtained from carcino- genicity bioassays performed with rodents. Bioassays in Experimental Animals For most chemicals there is no epidemio- logic evidence to evaluate carcinogenicity. Epidemiologic evidence may lack the power to adequately assess the carcino- genicity of the substance because of insuf- ficient numbers of subjects, confounding factors, or inadequate documentation of exposure. In these cases the only evidence for carcinogenicity may be derived from the results of long-term tests in experimen- tal animals. The carcinogenicity of substances is de- termined from experiments in which test animals, typically rodents, are exposed to the subject chemical, most commonly for long periods of time at high doses that approach the maximally tolerated dose. Rodents are the preferred animals because of their comparatively small size, their rather short life span, and their sensitivity to the development of cancer. Since the life span of these animals generally is two to three years, it is necessary to expose the animals to high enough doses of the test substance to compensate for the compar- atively short period of treatment in the rodents relative to the lifetime of humans. The philosophical basis for testing at comparatively high dose levels is the belief that carcinogenesis is an unusual form of chronic toxicity. Very high doses of chem- icals given either acutely or chronically are most commonly toxic and kill the test animals without causing tumors. Chemi- cals tested for carcinogenicity are usually selected because they are thought to have a high probability of being carcinogenic; even among these selected chemicals, many or most do not show a carcinogenic re- sponse, or they show an equivocal re- sponse. The exception to this is the group of known human carcinogens which have all proven to be carcinogenic in animal tests. This activity of human carcinogens in animal experiments is one of the observa- tions that supports this method of identifi . . cation ot carcinogens. Long-term carcinogenicity test methods have evolved over time. Earlier studies tended to be done in a single species, often with very small numbers of animals, with incompletely defined experimental meth- ods and materials, and with incomplete methods for pathological evaluation of the test animals. Nonetheless, in cases in which

clear strong evidence of carcinogenicity re- sulted, these studies still serve as data for assessment of the substance. More recently, particularly through the efforts of the National Toxicology Pro- gram and with the adoption of the Good Laboratory Practices Act, better docu- mented and more thoroughly designed car- cinogenicity tests have become the standard of performance. Potential carcinogens are tested in at least two species, usually rats and mice, in sufficient numbers to allow adequate statistical evaluation of the results, and usually with several dose rates. Pathol- ogy is thoroughly evaluated in the test animals and is subject to independent veri- fication. The overall results are also re- viewed by an independent group of experts who are familiar with tests of this type. Analysis of the results of carcinogenesis bioassays involves statistical as well as biological evaluations. Statistical analyses compare the incidence and multiplicity of tumors between experimental and control groups. Test results are analyzed for statis- tical significance. In cases where multiple experimental groups are tested at different dose levels, analyses are included to seek statistically significant dose-related trends. Where possible, control group results are compared to the historical record of previ- ous control groups of the same species, strain, and gender, maintained under simi- lar conditions. Biological considerations include exami- nation of the study for possible confound- ing factors such as the route or method of administration, group size, and survival. Statistically significant results that stand without biological reservation are included in the qualitative assessment of carcinoge- nicity. Results that show the development of unusual tumors are given greater weight than those that show an increase of tumors that occur at a high spontaneous back- ground level in controls. Additional weight is given to cases where tumors arise with exceptional rapidity or with unusually high incidence. All of these factors gain strength if the pattern of results is consistent be- tween species. The International Agency for Research on Cancer (IARC) has been the leading international source for assess Classification Nature of Evidence 532 Assessment of Carcinogenicity Table 4. IARC Classification of Evidence of Carcinogenicity No evidence Inadequate Limited Sufficient No studies of the material. Experimental data are scientifically compromised or are inconclusive Positive evidence for carcinogenicity is limited to one species and is not exceptional. Positive results in two species, particularly with evidence of dose/ response relationships, and if there are multiple confirmatory results NOTE: IARC = International Agency for Research on Cancer. ing carcinogenicity of chemicals, environ- mental mixtures, and occupational expo- sures. The evaluation approach used by IARC incorporates the considerations noted above and concludes with a judg- ment about the strength of the evidence for the carcinogenicity to animals of the sub- stance of exposure (table 4~. Results showing carcinogenicity can be judged "sufficient" if they show an unusual type of tumor rarely seen in the species, particularly if these tumors occur with un- usually high incidence, or if their latency period (time to first tumor) is exceptionally short. This assessment considers the strength of the evidence that the compound is carcinogenic (qualitatively), but does not assess the strength or potency of the sub- stance as a carcinogen. Cell proliferation, either constitutive or induced in response to injury, is an essential characteristic of tissues with a high risk of developing cancer. The information pre- sented earlier indicates that cell prolifera- tion, and in particular DNA synthesis, may have a crucial role in the various mecha- nisms of carcinogenesis. Thus, factors that increase cell proliferation rates may poten- tiate the development of cancer. In long-term bioassays for carcinogenic- ity, the doses of chemicals are chosen in relation to the highest dose found to have a minimal effect in terms of subacute toxic- ity, that is, no reduction of survival in the subacute phase up to several weeks follow- ing treatment. The doses chosen, therefore, do not sufficiently compromise the func

David G. Kaufman 533 lion of any vital tissue or organ so as to appreciably reduce the survival of test ani- mals during the subacute phase. This does not imply that the dose of the test agent has no effect on the tissues or organs of the test animals. Certainly, the goal is to demon- strate a carcinogenic effect of the substance if this is a property of the compound. Since the doses used to test for this activity are generally much higher than doses usually experienced by the human population, it is conceivable that toxicity may be a signifi- cant factor in tumor development. A chemical that tests positively in such a bioassay may require cell toxicity and the resulting increased cell proliferation of re- generative processes in order to evoke a tumor response. Thus, increased cell pro- liferation may be an essential feature of the apparent carcinogenicity of the chemical in the bioassay. This may not occur under realistic conditions of exposure at levels to which humans are subjected. In species and strains of animals used for bioassays, there are some tissues in which there is a notable spontaneous incidence of tumors. In these tissues it may suffice to produce an increased rate of cell prolifera- tion, strictly as a matter of cell toxicity and regeneration, to demonstrate an increased tumor incidence. Thus the toxicity of the high dose of chemical may be modulating (accelerating) the development of tumors. These tumors may have developed sponta- neously, but more slowly. Alternatively, these tumors may have arisen from cells that were already initiated as an intrinsic property of the tissue of the specific animal, although they would not have been ex- pressed as tumors without the accelerating (promoting) effect of the chemical. In such cases, it is conceivable that the bioassay that yielded an increased incidence of tumors did so only because the high dose of the test chemical was toxic in these susceptible tis- sues and increased the cell proliferation rate. It is possible that in such a circumstance cellular protective mechanisms may pre- vent toxicity at the doses of these com- pounds to which humans are exposed, and may also prevent stimulation of cell prolif- eration. Thus a chemical that produced an . . . . . . apparent positive carcinogenic result In an animal bioassay might not represent any appreciable carcinogenic risk at realistic levels of exposure in humans. For this reason some recommendations have been made to test chemicals for carci- nogenicity at levels at which humans are exposed. Testing at doses that produce no detectable toxicity might be more justifi- able. Still, the underlying hypothesis for this line of argument is largely untested. Whether such mechanisms represent a tan- gible factor in the positive test for carcino- genicity of any compounds is unknown. In order to identify chemicals accurately as carcinogens in bioassays, it is essential to distinguish true positive results from ap- parent positives based on this confounding effect for compounds which in fact are only false positives. To assess this possibility it may be desir- able to undertake studies utilizing doses selected for the extent of their toxicity, ideally including some specific dose that causes no toxicity. If it were possible to determine that a correlation existed be- tween tissue toxicity and tumorigenesis, then it might be possible to provide some correction for cancer risks based on the degree of toxicity in the tissue as deter- mined even from short-term in viva stud- ies. Such an effort would identify more clearly the kinds of compounds that might require testing in larger numbers or at lower doses to relate carcinogenicity risks in humans to results in animal studies. ~ Recommendation 3. The role of tox- icity in carcinogenesis should be evaluated. Short-Term Tests in Viva and in Vitro Short-term tests have come to be used in the qualitative assessment of carcinogenic- ity and in the prioritization of chemicals for in viva testing. Typical assays evaluate the mutagenicity of the test chemical either directly or following metabolic activation, usually using the S9 microsome-enriched fraction from livers of rats whose metabo- lism has been induced with agents such as Aroclors. Other assays evaluate the ability

534 Assessment of Carcinogenicity of the compounds to produce large-scale disruption of DNA, specifically chromo- somal damage, including the induction of sister chromatic exchanges, aneuploidy, and micronuclei. Induction of DNA repair is another type of end point used to assess a chemical's ability to cause DNA damage. Perhaps the short-term test that most closely ap- proaches a test of carcinogenicity in vivo is evaluation of the ability of a chemical to transform cells in vitro. A variety of cells and organisms from prokaryotic orga- nisms, to plants and insects, and to mam- malian cells (including human cells) are used in these studies. Assessment of chem- ical activity often is based on results ob- tained in multiple test systems. The more systems in which the chemical has shown activity, the greater the evidence of its . ~ . . . potentla tor carcmogenlclty. These assessments can be used to priori- tize chemicals for testing in long-term car- cinogenicity studies in viva. Results of short-term tests may also be used to con- firm the results of carcinogenicity testing in vivo. The short-term tests generally cannot be used alone to determine carcinogenicity because validation studies for these tests are not sufficiently accurate and precise. A1- though most carcinogens have been shown to be genotoxic, not all are genotoxic. Consequently, these tests are prone to false negatives. Other tests are being evaluated for their ability to detect carcinogens that act via nongenotoxic mechanisms or through mechanisms that only generate genotoxlc mediators indirectly in specific cells. At present, this area must be regarded as one in development rather than a defined area that can be used routinely for regulatory purposes. Methods for Quantitative Extrapolations to Human Cancer Risk In this section, the methods used to esti- mate quantitatively the risk for human can- cer development resulting from "real Table 5. Factors Involved in Quantitative Estimates of Human Risk Estimates of quantitative risk in experimental animals Extrapolation among species Extrapolations among routes of administration and exposure Extrapolation to dose levels of human exposure world" exposures to substances judged qualitatively to be carcinogenic are consid- ered. In the event that sound epidemiologic evidence for carcinogenicity exists and ad- equate data on exposure rates in study populations are available, it may be possi- ble to directly estimate cancer risks for humans. Because such data are not avail- able in most cases, extrapolations must be made from results of animal studies (table 5~. Both the procedures and philosophy of the methods in use and alternative methods that have been proposed to make these extrapolations are considered in this sec- t~on. Estimation of Quantitative Risk in Laboratory Animals The first step in quantitative evaluation of carcinogenicity is to determine the quanti- tative tumor response in long-term studies in animals. Such studies must determine and verify tumor responses in various or- gans and tissues of test animals and con- trols. Results are analyzed to determine whether there is an excess incidence of tumors, cataloged on a site-by-site basis, in experimental animals compared to con- trols, and whether these excesses meet sta- tistical tests of significance. Any errors in the diagnosis of tumors will profoundly affect calculated incidences and subsequent extrapolations. Judgments regarding inclu- sion of animals dying prematurely or oth- erwise lost from the study will also affect the population size (denominator) in the incidence calculation. The type of tumor used in the quantita- tion presents another problem. Some tu- mors are widely diagnosed by similar cri- teria and are not controversial. Pathological diagnosis can pose a far more difficult prob- lem, particularly when attempts are made

David G. Kaufman 535 to distinguish an early neoplastic lesion from a nonneoplastic but undifferentiated area of tissue repair. Also, only a single observation is made of the lesion. Lack of the extensive natural history detail and clin- ical follow-up known for most human tu- mors complicates diagnosis of many types of tumors in animal bioassays and influ- ences their quantitative reliability. In most cases only a single positive result is evalu- ated, and questions of quantitative repro- ducibility are not considered; Conse- quently, there is no way to estimate accurately the error in derived estimates of tumor incidence, nor is this error propa- gated through the quantitative extrapola- tion to yield the probable error in the risk estimate. , Extrapolations Among Species Carcinogenicity studies have traditionally been performed in strains and species of animals sensitive to tumor development. Rodents are commonly used because they can be inbred with high uniformity, have a comparatively short life-span, and are com- paratively inexpensive to maintain. The choice of sensitive species and strains facil- itates the qualitative detection of carcino- genic activity, but raises problems with regard to quantitative aspects of the tumor response. The principal issue is whether the assay is determining an inherent carcinogenic activ- ity of the material tested or whether it is influencing the rate or extent of occurrence of tumors that occur spontaneously in the animal subjects, perhaps facilitated by the toxicity of the compound and the extent of the resulting regenerative process. This is not a problem in judging the qualitative carcinogenicity of a substance, because in- crease of tumor incidence by any mecha- nism is consistent with the definition of carcinogenicity. The problem does arise when this type of data is used for quantita- tive assessments of cancer risks in humans. Human sensitivity to carcinogens is not well understood. We do not know how the human population compares with typically used experimental animals with regard to sensitivity to carcinogens. it Is not known whether small or large proportions of the human population are at high risk from carcinogens. For that matter, there are only fragmentary data comparing sensitive ro- dent strains with more-resistant strains. Sensitivity and resistance are probably var- iables dependent on the types of carcino- gens considered. Since human exposures are to complex mixtures of chemicals rather than to a single pure compound, our general lack of knowledge regarding inter- actions among carcinogens and toxins raises a further, major complication. This problem is of particular concern in consid- ering the risk from complex mixtures such as diesel exhaust. Extrapolations Among Routes of Administration or Exposure Carcinogenicity tests are used to evaluate the activity of substances; they are not necessarily designed to reproduce precisely the conditions of human exposure. Most commonly, potential carcinogens are tested by being incorporated into the diet of experimental animals and administered throughout life. In some cases, but less frequently, the substance is added to the drinking water or directly instilled into the stomach by gavage. Skin painting (with or without subsequent applications of pro- moters) and intraperitoneal injections are still the routes of administrations in test systems with skin papillomas or lung ade- nomas, respectively, as end points in mice. Rarely, other less common routes of ad- ministration are used, such as inhalation, subcutaneous injections, or intratracheal administrations. In most cases, the route of administration is chosen for practical rea- sons: to simplify the methods of the study and to limit costs, not because of the sim- ilarity to the typical exposure route of the human population. Because long-term tests often involve routes of administration dissimilar to those of typical human exposure, methods have been developed to relate experimental re- sults in animals to estimates of human cancer risk. If humans are usually exposed to a compound by inhalation, and the carcinogenesis bioassay utilizes ingestion,

536 Assessment of Carcinogenicity there must be a way to relate these studies to compensate for peculiarities of tumor localization or number resulting from the route of administration. The practical solu- tion to this problem has been to make an evaluation by analogy with another com- pound. The test chemical is compared to a refer- ence compound that has been tested by the same route (for example, ingestion), but also has been tested by the route of admin- istration most typical of human exposure (for example, inhalation). Studies using the same route of administration (that is, inges- tion) compare the potency of the two chemicals. The tumor response for the test chemical is related to an equivalent dose of the reference compound yielding the same tumor response. On the basis of this infor- mation, the tumor response to the test chemical via the normal route of human exposure is extrapolated to the tumor re- sponse of the equivalent (calculated) dose of the reference chemical by that route of . . . a' .mlnlstratlon. Essential to making this a workable means of extrapolation is to develop a body of information concerning the carcinoge- nicity of sufficiently diverse reference chemicals tested by various routes of ad- ministration. The selection of the reference chemical to be used for extrapolation would be based on similarities between this chemical and the test substance with regard to chemical structure or metabolic mecha- nisms that convert it to an active carcino- gen. Although this approach offers a rea- sonable and pragmatic solution when practiced with adequate biological insight, its reasonableness does not ensure its cor- rectness or accuracy, since it has not been adequately studied and validated by rigor- ous scientific tests. Even allowing for sub- stantial safety factors in applying this method, it is unclear whether the conclu- sions generated are sufficiently protective or vastly too restrictive when used in ex- trapolations to human cancer risks. The extrapolation process presently used to make quantitative risk assessments re- quires that several estimates be made to adjust for differences between human ex- posures and the conditions of animal exper imentation. Examples include corrections for differences in the route of administra- tion, dose rate, and species. These correc- tion factors usually exaggerate the esti- mates of human risk or the relationship to humans so that the extrapolations do not underestimate the actual human risk. The corrections are applied to the best estimate of the actual excess of tumors in the test group as compared to appropriate controls. Extrapolated estimates of excess cancer deaths in the human population are calcu- lated from the application of these correc- tion factors to the quantitative estimate of excess tumors. The problem with this process concerns the validity and verification of estimation and extrapolation. The approach is logical and offers a well-intentioned attempt to maximize the protection of the population at risk. However, the elements of the ex- trapolation process are all estimates based on inadequate amounts of data and incor- porating a wide safety margin. Further- more, these estimates and extrapolations have been accepted and used without ade- quate testing and validation. To extrapolate from available data con- cerning the carcinogenicity of substances tested in animals to quantitative estimates of cancer risk in the human population is difficult and problematic. Much essential data are not available to support the meth- ods that are used for correcting differences in doses; in species, including the use of sensitive species; in routes of administra- tion; and between the test compound and the chemical on which extrapolation meth- ods are based. Distinguishing experimental tumors in- duced by complete carcinogens from tu- mors promoted by the biological environ- ment of the particular animal tissue would improve significantly our current methods of extrapolating and quantitating risks. These two different development mecha- nisms of specific types of tumors may determine the choice between different methods for extrapolation to low doses. Additional information is also required for understanding the implications of the differences in carcinogenicity between spe- cies. To what extent is the sensitivity of the

David G. Kaufman commonly used test animals a function of differences in carcinogen metabolism or defense mechanisms such as DNA repair capacity? Only with such information can it be concluded with confidence that the means of correcting for differences in routes of administration for chemicals of different types have been established. Ad- ditional data and a sound scientific valida- tion study would allow greater confidence in making these corrections for route of . . . ac .m~n~strahon. Extrapolation methods, although a logi- cal and responsible approach, are still largely untested. For example, extrapola- tions concerning diesel exhaust have used topside coke oven emissions as a compari- son standard. Objective tests of the validity of the process of extrapolation by analogy must be undertaken. If the currently used extrapolation methods are precise or even reasonably effective, then they should be used to make testable estimates of human cancer risks, or, conversely, predict effects in experimental animals from human data. A rigorous scientific test is needed to vali- date the currently used method for risk extrapolation. · Recommendation 4. Critical data should be gathered for quantitative assess- ments. Extrapolation to Dose Levels of Human Exposure For practical reasons-basically the need to produce tumors within the short life span of comparatively small numbers of ro . . . c tents carc~nogen~c~ty tests are usually performed with high doses of the test sub- stances. These dose levels are usually quite different from the levels to which humans are exposed. It is therefore necessary to extrapolate from the high dose levels of the animal test to the much lower levels of human exposure. Unfortunately, little is known about the effects of carcinogens at the low dose rates typical of human expo- sures. A particular concern about high doses is that their effect may be more than that of a pure carcinogen. Doses may be sufficiently high to cause toxicity in the 537 target tissue; cell necrosis and restorative regeneration with elevated cell proliferation may potentiate the intrinsic carcinogenic effect of the agent. At the low doses at which humans are exposed, this toxicity may not occur, and tumor induction, if it occurs at all, may be at vastly lower rates. Carcinogens generally show a cancer- causing potential in relation to the dose of the chemical administered. At high doses, the tumor response is usually directly re- lated to dose, with progressively more tu- mors as the dose is increased. The nature of dose/response relationships at low doses is the subject of considerable debate. For most studies, the experimental design- principally the numbers of animals used is suited to detect strongly or moderately . . caranogenlc su Stances. There are usually too few animals in an experiment to show a statistically signifi- cant, small increase in tumors caused by a less carcinogenic substance or a low dose of a more active substance. As a consequence, typical tests to detect carcinogens have pro- duced little information about tumor re- sponses at low dose levels. The dearth of information on this topic has made it im- possible to determine the shape of the dose/response curve at low doses, to ascer- tain whether there is a certain minimal dose (threshold dose) below which there is no effect, or whether there is an effect at any level down to zero exposure. In order to obtain further information about the shape of the dose/response curve at lower exposures, the National Center for Toxicological Research undertook a study using large numbers of animals and extend- ing the dose range lower by an order of magnitude (Cairns 19801. This study tallow) used two strong carcinogens di- ethylnitrosamine and 2-acetylaminofluo- ren~and evaluated their effects in C3H x BL6 hybrid mice (Gaylor 1980; Littlefield et al. 1980~. Study results suggested that for spontaneously occurring liver tumors in this strain of mice, the induced tumor incidence was linear to the dose of carcin- ogens administered. In the case of bladder tumors, where the carcinogen is assumed to function both as the initiator and the promoter, tumor incidences decreased rap

538 Assessment of Carcinogenicity idly as the dose was reduced, with what could be interpreted as a threshold effect. It is possible that in the case of spontaneously occurring tumors, the animal strain or the conditions of maintenance provide suff~- cient promoting stimulus to produce can- cers in proportion to the extent of initiation by carcinogen treatment. If this interpreta- tion is accepted, then this study provides a measure of support for both the linear extrapolation model and, at the other ex- treme, the threshold model. In spite of this large-scale effort, our knowledge has advanced only modestly. The results do not provide enough infor- mation to predict the shape of the dose/re- sponse curve for a particular chemical at the very low doses of human exposures. In fact, the results show that each model may have some measure of validity, and the appropriate extrapolation may relate to the carcinogenic agent and the species (and strain) at risk. Such a conclusion would greatly increase the complexity of achiev- ing reliable extrapolations of animal carci- nogenicity data to human cancer risk esti- mates. It would point out the deficiencies of adoption of any general and universal extrapolation method for human cancer risk of all chemicals. The effects of small doses of carcinogens have been the subject of considerable con- jecture, because of the absence of critical data concerning carcinogenic effects at very low doses and a still quite incomplete knowledge of the mechanistic details of carcinogenesis. The development of strongly entrenched opposing positions re- flects the lack of real knowledge in this area and the significant practical consequences resulting from the use of either of the extremes of extrapolation. It is not germane here to consider the relative merits of the various extrapolation analyses that have been made. Various methods of extrapolation yield estimates of cancer risk in animals differing by as much as several orders of magnitude. When these uncertainties are combined with uncertain- ties about the comparative sensitivity of humans, the accuracy of the estimate of tumor incidence in bioassays, and the ef- fects of mixed exposures, the estimation of cancer risks in humans is clearly far from accurate or reliable. Carcinogen treatments or exposures can occur under a wide variety of circum- stances. Human exposures to a particular carcinogen may occur as the result of a single incident, as in the case of atomic bomb survivors, or may result from chronic or episodic exposures at lower lev- els. Experimental studies also have varied the mode of carcinogen exposure. These range from single treatments with either short-lived or persistent substances, to multiple discrete exposures with any of a variety of intervals during a treatment pe- riod of weeks to months; or continuous treatments, usually by incorporation into the animals' food or drinking water. Results of tumor development following discrete treatment protocols with the same substance but using different treatment schedules have been used to compare the effects of dose fractionation. Single treat- ments have generally been found to be less effective than the same total dose given over multiple treatments. More fraction- ated doses tend to produce lower toxicity, allowing more of the treated animals to survive long enough to develop tumors. Protocols that administer the carcinogen in few doses over a short interval early in the animals' life-span can result in reductions in the latency period, but these protocols may result in a lower overall incidence of tu- mors in the treated population. Exposure levels, particularly as they ap- ply to the individuals at risk, are poorly defined in the case of most human expo- sures to carcinogenic or potentially carci- nogenic substances. Estimates of average levels of exposure are usually based on current measurements of ambient concen- trations in air, water, food, and so on, and these are extrapolated back in time for the duration of the study period. Estimates of concentrations may be stratified according to the specific type of work classification, the amounts of various kinds of food con- sumed, and so on. Although such ap- proaches are very limited in defining indi- vidual exposures, they may be the best way of evaluating these parameters in an en- tirely retrospective manner. More desirable

David G. Kaufman 539 would be careful individual measurements of the presence and action of the substance in the target tissues of the subjects through- out the course of the epidemiologic study. It would be valuable to have data on exposure at the target tissues in order to estimate the activity of environmental compounds as carcinogens in the organs of human subjects. Such data would also vastly improve our understanding of ex- perimental studies in animals. If these data were available for experimental and epide- miologic studies, it might be possible to predict cancer risks better, and these esti- mates of risks might be provided on an individual basis. With such information, it might be possible to monitor work condi- tions or even modify the design of internal combustion engines or their fuel, in order to minimize levels of harmful agents, or to reduce their levels to acceptable limits. One of the major limitations of most . . . . . . . . epic .em1o Ogle Investigations ot envlron- mental factors as determinants of cancer is the inadequacy of quantification of expo- sure to specific agents. Consequently, these studies have had to rest on estimates of exposure levels that are based on such parameters as geographic location, classifi- cation of job or other function, and retro- spective estimates as averages based on current measurements. These limitations all reduce the power of epidemiologic stud- ies to detect positive carcinogenic effects, or, circumscribe the statistical range for which a no-effect result can be substanti- ated for negative studies. Performing dosimetry studies on exper- imental subjects would improve the power and specificity of epidemiologic studies of the carcinogenic effects of environmental exposures. Recent scientific advances have made this goal far more feasible. It is now possible to detect the adducts formed by the binding of carcinogens to DNA or other macromolecules within cells, using specific antibodies and sensitive methods for amplifying the signals they produce. ELISA methods can provide this amplifica- tion and have been shown to detect envi- ronmental levels of some substances. Since these methods are largely untested as a means of environmental monitoring, tests are needed to verify their validity and discriminatory power. Such tests could be undertaken using well-studied animal model systems in which quantified tumor responses have been observed with carcin- ogens for which these modern techniques exist. If these tests in experimental animals prove the power and sensitivity of this approach under the controlled conditions of the laboratory, these methods should be applied to the study of epidemiologically defined groups to determine whether they offer distinct improvements to the estima r t1on ot exposures. Such a test could be conducted on groups for which conventional environmental monitoring of the workplace, food con- tamination, cigarette smoking, and so on, can be used as a quantifiable basis for comparison with experimental results. In the event that these methods are shown to be reliable as discriminators of exposures for a few well-defined environmental haz- ards, then these methods could be used to gain further insights about suspected hu- man carcinogens in a prospective study. If this type of study demonstrates that expo- sure levels at the target tissue are highly correlated with the tumor incidence rates, then this approach should be used routinely to augment or replace conventional meth- ods of environmental monitoring. · Recommendation 5. Acceptable meth- ods should be developed for dosimetry in humans. Experimental Evidence on Carcinogenicity of Diesel Exhaust In this section the information available about the carcinogenicity of diesel exhaust emissions is reviewed, and the data from animal studies and human epidemiologic studies and the risk extrapolation based on these data are described. Other studies that describe the results of short-term tests that provide additional insight about the poten- tial carcinogenic activity of diesel exhaust emissions are also reviewed.

540 Assessment of Carcinogenicity Table 6. Short-Term Tests Used to Assess Diesel Engine Exhaust Reverse mutations in Sal~no,~ella typhi~n``ri~'n bioassay (Ames assay) D3 recombinogenic assay in Saccharo~nyces cerevisiae Forward mutagenesis in mouse L5178Y lymphoma cells Mutagenesis in Chinese hamster ovary (CHO) cells DNA fragmentation by alkaline elusion in Syrian hamster embryo (SHE) cells DNA adduct detection using the 32P-postlabeling technique Sister chromatic exchange in CHO cells Sister chromatic exchange in human lymphocytes treated in vitro Sister chromatic exchange in fetal hamster liver exposed transplacentally Cell focus transformation assay in SHE cells Cell focus transformation assay in BALB/c mouse 3T3 cells Short- Term Tests of Activity of Diesel Emissions Short-term in vitro assays that detect mu- tagenesis or other activities that suggest carcinogenic potential (table 6) have been used to evaluate diesel engine exhaust (Lewtas et al. 1981~. In these studies, diesel exhaust particulates from four different en- gines were compared with positive control substances including BaP, and extracts from cigarette smoke, coke oven emis- sions, roofing tar, and gasoline engine . . emlsslons. Each of these substances was tested for mutagenic activity in the Salmonella typhi- murium bioassay (Ames assay) (Claxton 1981~. The assay was performed with and without the addition of induced rat liver microsomes (S9 fraction) as a source of mammalian metabolizing activity. Each of these test substances was mutagenic in the presence of S9, but extracts ot rooting tar and cigarette smoke were inactive in its absence. In comparisons of uniform amounts of extracted organic material, the diesel exhausts produced a 20-fold range of activity, with the exhaust from a Nissan engine being most active, those from Olds- mobile and Volkswagen engines being in- termediate, and that from a Caterpillar engine being least active. The addition of S9 to diesel exhausts did not increase mu tagenic activity, but reduced it in most cases. Mitchell and coworkers (1981) tested these substances using the D3 recombino- genic assay in Saccharomyces cerevisiae, for- ward mutagenesis in mouse L5178Y lym- phoma cells, and sister chromatic exchange in Chinese hamster ovary (CHO) cells. The recombinogenic assay was insufficiently sensitive to detect activity, but positive results were found with the other two assays. Most of the extracts proved positive in both assays in the presence of a metabolic . . activation system. Interestingly, most compounds induced sister chromatic ex- change and were mutagenic in the absence of an exogenous activation system; this suggests that the extracts contain direct- acting mutagens and DNA-damaging agents. These compounds were also assessed for genotoxicity by examining mutagenesis in CHO cells and by studying DNA fragmen- tation by alkaline elusion and cell transfor- mation by focus assay in Syrian hamster embryo (SHE) cells (Casto et al. 1981~. These assays, which depend on the test cells themselves for metabolic activation, gave far fewer positive results. Extracts of Volkswagen and Datsun diesel engine ex- hausts were positive in the CHO mutagen- esis assay, whereas extracts from coke oven emissions or from gasoline engine exhaust were active in this assay and in the SHE cell DNA fragmentation assay. Studies of transformation of SHE cells by these com- pounds were negative. In other studies, these compounds were assayed for their ability to transform mouse BALB/c 3T3 cells when tested with and without an S9 activating system from livers of rats treated with Aroclor-1254 (Curren et al. 1981~. Although there were no clear dose/response relationships, qualitative as- sessments were possible and an approxi- mate ranking of activity could be achieved. By this evaluation, emissions from a coke oven and gasoline engine were most active and approximately equal in activity, Nissan diesel was less active, and roofing tar was least active. Extracts from Oldsmobile and Caterpillar diesel exhausts were inactive. Several concurrent studies have exam

David G. Kaufman 541 ined the mutagenic activity of extracts of diesel engine exhaust particulates and the ability of these extracts to induce sister chromatic exchange. Nachtman and col- leagues (1981) tested extracts from diesel exhaust particulates of unspecified origin using the Ames Salmonella bioassay. They found the extracts to be mutagenic without an induced rat liver S9 metabolic activating system. Addition of the S9 fraction reduced activity, presumably by competing with bacteria for binding of the reactive com- pound. Treatment of the extract with re- ducing agents greatly diminished muta- genic activity; addition of the induced S9 fraction restored the activity. l his was interpreted as suggestive of the presence of nitroarene mutagens in the diesel exhaust particulates. Pitts and coworkers (1982) studied ex- tracts of exhaust particulates from a Nissan diesel engine. They demonstrated the pres- ence of several nitroarene compounds in the extracts and showed the mutagenicity of some of these chemicals in the Ames assay. By comparing a standard Salmonella strain to one deficient in nitroreductase activity, it was possible to recognize the sizable contribution of active nitroarene mutagens to the total mutagenic activity of the extracts. Li and Royer (1982) examined the effect of diesel exhaust extracts on mutagenesis induced by BaP and N-methyl-N'-nitro- N-nitrosoguanidine (MNNG) in CHO cells, testing exhaust extracts from five different brands of diesel engine. CHO cells were treated with the strong mutagen (BaP or MNNG) plus diesel exhaust, and the mutation frequencies were compared to the sum of the mutation frequencies in cells treated with each component separately. For MNNG alone and for BaP together with an exogenous metabolic activating system, the diesel exhaust extracts potenti- ated mutagenesis frequencies in the CHO cells. The results were interpreted as show- ing the presence of comutagens or cocar- cinogens in the extracts of the particle fraction of diesel exhaust. The ability of diesel emissions to induce sister chromatic exchanges has been stud- ied in human lymphocytes in vitro (Lockard et al. 1982) and in fetal hamster liver following transplacental exposure (Pe- reira et al. 19821. Both systems used a Nissan engine to generate exhausts. In the former study, cells were exposed to an organic extract of diesel exhaust particu- lates; in the latter study, hamsters were exposed transplacentally to exhaust emis- sions, exhaust particulate material, or to an organic extract of the particulates. In both studies the diesel exhaust extracts produced a dose-related increase in sister chromatic exchanges. Wong and colleagues (1986) exposed F344 rats to diesel exhausts by inhalation for 30 months using methods that had been shown previously to produce lung tumors in these animals. They evaluated DNA extracted from lung tissue of diesel-ex- posed and control rats for damage by the use of the [32P]-postlabeling technique to detect the presence of adducted nucleo- sides. This technique had the sensitivity to distinguish exposed from unexposed rats, based on the presence and quantity of ad- ducts detected. Data on Carcinogenic Activity of Diesel Exhaust Emissions Kotin and coworkers (1955) tested the car- cinogenicity of extracts of diesel engine exhaust particulates. The concentrations of polynuclear aromatic hydrocarbons (PAHs) in extracts of exhaust particulates were found to vary according to the con- ditions of operation of the engine. Extracts were tested for carcinogenicity by skin painting on the backs of C57 black and strain A mice. The extracts produced nota- ble systemic toxicity in both mouse strains. In both strains the extract induced the formation of skin tumors, whereas none developed in controls. There were few tumors in the C57 black mice, but 50 to 85 percent of the strain A mice developed skin tumors. In one of the groups a large pro- portion of the tumors were carcinomas rather than papillomas. Nesnow and colleagues (1981, 1982, 1983a) and Slaga and coworkers (1981) showed that extracts from Nissan, Olds- mobile, Volkswagen, and Caterpillar die

i 542 Assessment of Carcinogenicity set engines and from a Mustang gasoline . . . . . engine actec as initiators In two-stage car- cinogenesis when promoted with TPA in the skin of SENCAR mice. Exhaust ex- tracts were prepared according to estab- lished protocols using exhausts produced under carefully defined engine operating conditions and with a uniform fuel; these extracts were tested parallel to a number of other substances which were positive con- trols and scaling factors (extracts from cig- arette smoke, coke oven emissions, roofing tar, and so on) (Lewtas et al. 1981~. The exhaust extracts from Nissan and Volkswa- gen engines produced skin papillomas in proportion to dose. The Caterpillar engine extract was regarded as active although no clear dose response was evident. The Olds- mobile and Volkswagen exhausts as well as the gasoline exhaust produced a weak re- sponse. In each case, the total tumor-initi- ating activity exceeded the activity of the BaP content of the exhaust samples. This indicates that there are components of the mixture in addition to BaP that are in- volved in the total tumor-inducing activity. A later study by this same group (Nesnow et al. 1983b) showed that Nissan diesel exhaust extract was a complete carcinogen as well as an initiator, but not a tumor promoter, in SENCAR mouse skin. The carcinogenic activity of diesel engine exhaust was evaluated in a separate study using the lung adenoma in strain A mice as an end point (Orthoefer et al. 1981~. Expo- sures were to diesel exhaust by inhalation or to an exhaust particulate fraction by intraperitoneal injection for a period of 8 weeks followed by sacrifice at 2~30 weeks. Other animals were exposed by inhalation for up to 7 months. None of these treatments produced a notable in- crease in lung adenoma incidence or multi- plicity as compared to controls. In separate experiments, low doses of urethane were added to the treatment protocol for diesel- exposed and for unexposed mice. In these studies, there was a significant increase in the number of adenomas as a result of the combined treatments with urethane and diesel exhaust. Diesel exhaust has been studied for its . . . . . . carclnogemclty fly In la .atlon exposure In F344 rats (McClellan et al. 1986a). Inhala- tion exposures at soot concentrations of 0.35, 3.5, and 7.0 mg/m3 were carried out on a 7-hr/day, 5-day/week schedule for up to 30 months; exhaust was generated by an Oldsmobile engine operating under defined conditions. Rats chronically exposed to diesel exhaust developed areas of pulmo- nary fibrosis and accumulations of soot, particularly at the highest exposure level. The rates of clearance of soot particles were shown to be significantly impaired at the two higher dose levels (Wolff et al. 1986~. A significant increase in the incidence of lung tumors, particularly adenocarcinomas and squamous cell carcinomas, was found in the group of rats exposed to the highest concentrations of diesel exhaust (McClellan et al. 1986a,b; Mauderly et al. 1987~. In a separate series of studies, rats, mice, and hamsters were exposed to diesel ex- hausts by inhalation (Heinrich et al. 1982, 1985, 1986~. Hamsters were studied to de- tect whether exposure to diesel exhaust affected tumor responses in animals treated with proven respiratory tract carcinogens: diethylnitrosamine (DEN) was injected subcutaneously, or dibenzo~a,hianthracene was instilled intratracheally (Heinrich et al. 1982, 1985~. Inhalation exposures were to exhaust from a Daimler-Benz diesel en- gine, either whole exhaust or exhaust with the particulate fraction removed; control hamsters breathed air. Lung tumors devel- oped in too few hamsters for meaningful evaluation. Papilloma of the larynx and trachea induced by DEN were potentiated by exposure to diesel exhaust, either with or without particles. Focal proliferations in the lung periphery were seen in all groups exposed to total exhaust; these lesions were also seen in groups treated with the known carcinogen and the particulate-free exhaust. In a separate study, Heinrich and co- workers (1986) exposed rats, mice, and hamsters to total diesel exhaust (from a different engine), particle-free exhaust, or clean air by inhalation. No lung tumors were observed among hamsters, but there was an increased incidence of bronchiolo- alveolar hyperplasia and emphysematous lesions in animals exposed to total exhaust. In mice, exposure to diesel exhaust with or -' ---r

David G. Kaufman 543 without particles increased the incidence of lung adenocarcinomas as compared to controls. Bronchiolo-alveolar hyperplasias were far more common in mice exposed to total exhaust as compared to particle-free exhaust or controls. Differences between control rats and those exposed to particle- free exhaust were not significant. Rats ex- posed to total exhaust had several lesions that were significantly increased, including bronchiolo-alveolar adenomas, squamous cell tumors (one low-grade squamous cell carcinoma and eight benign keratinizing cysts), bronchiolo-alveolar metaplasias, and bronchiolo-alveolar hyperplasias. Oth- er animals were exposed to the exhausts and also were treated with a known carcin- ogen to evaluate the potentiating effect of the inhalation exposures. For hamsters and mice, there were no effects and inconsistent effects, respectively. Total diesel exhaust potentiated the induction, by dipentyl nitrosamine, ot malignant lung tumors, adenocarcinomas as well as squamous cell carcinomas. Diesel exhaust has also been evaluated for its ability to initiate the induction of foci of altered hepatocytes in rat liver. Pereira and colleagues (1981) exposed rats to diesel exhausts generated by a Nissan engine op- erating under defined conditions. They also performed two-thirds partial hepatic resec- tions on the rats and subjected them to the promoting effects of maintenance on a choline-deficient diet. At the end of the experiment, the rats were sacrificed and their livers were evaluated for the presence of the presumptive preneoplastic lesions, ~glutamyltranspeptidase-positive foci. The researchers found no increase in the num- ber of positive foci in the diesel-exhaust treated rats. It is possible that this assay is insufficiently sensitive to detect carcino- genic or initiating activity, for doses of the magnitude that could be received by inha- lation in the brief period of restorative hyperplasia in the liver. Epidemiologic studies of carcinogenicity of diesel exhaust in humans are limited in number. Studies by Hueper (1955), Raffle (1957), Kaplan (1959), Waxweiller et al. (1973), Menck and Henderson (1976), Heino et al. (1978), Luepker and Smith ~.. ~ -, - ~- r - ~ (1978), and Wegman and Peters (1978) have been reviewed by Schenker (1980~. These studies were all regarded as flawed because of limited durations of exposure or fol- low-up, because of small numbers of sub- jects, or because of uncertainties about the extent of diesel exhaust exposures or the types of tumors that developed. In four of these studies (Raffle 1957; Kaplan 1959; Waxweiller et al. 1973; and Menck and Henderson 1976) researchers touna no ~n- creases in lung cancer incidence that could be associated with diesel exhaust exposure. Each of the other four studies reported an excess of lung cancers in the subject groups as compared to control groups. However, flaws make it impossible to reach a definitive conclusion regarding the carcinogenicity of diesel exhaust from these studies. Harris (1983) reviewed the study of Lon- don transport workers by Raffle (1957) and updated information concerning that study to include the years 1950-1974. From these data he observed that there was no increased incidence of lung cancer, but by . . . . . . nitric Acing corrections tor uncertainties and confounding factors, he calculated an upper-limit estimate of potential risk for lung cancer attributable to diesel exhaust. Hall and Wynder (1984) investigated the potential carcinogenicity of diesel engine . . r ~- 7 exhaust exposure using a case-control analysis. They compared rune; cancer pa tients and controls without tobacco- related disease for occupational exposures to diesel exhaust as judged from job classi- fication. Although the study showed a strong relationship between smoking and lung cancer, it did not show an association between exposure to diesel exhaust and lung cancer. Quantitative Assessment of the Cancer Risk of Diesel Exhaust in Humans Some of the abovementioned studies have been used to assess the potential carcinoge- nicity of diesel engine emissions for the human population. Rather than the step- by-step extrapolation approach described earlier, the evaluation rested on earlier cal , . . ..

544 Assessment of Carcinogenicity culations for other compounds and the approach used to make this assessment was an extrapolation by comparisons and anal- ogy (Albert et al. 1983; Lewtas et al. 1983~. Results of skin carcinogenesis and skin ini- tiation in SENCAR mice were compared for diesel exhaust extracts and extracts from gasoline engine exhaust, cigarette smoke condensate, roofing tar vapors, and coke oven emissions. Similarly, results of short-term assays for genotoxicity were compared for these same compounds. From these data the comparative potency of diesel exhaust was estimated on the basis of the most active diesel extract. To relate these values to estimates of risk for the human population, estimates of human lung cancer risk were made for coke oven emissions, roofing tar, and cigarette smoke condensate. For each of these compounds there were epidemiologic data relating ex- posure to human cancer and experimental data in test systems identical to that for diesel exhausts. The risk per unit quantity for diesel exhausts was extrapolated by determining the human risk on the basis of a unit quantity of organic extractable ma- terial. The estimated unit risk obtained for human lung cancer was 0.02-0.60 x 10-4 (lung cancers)/,ug exhaust particulates/m3 of air. To understand this estimate, it is impor- tant to recognize the inherent assumptions of the method (Lewtas et al. 1983~. The method assumes that the relative potency ~ . Or carcinogens in one carclnogenesls assay is directly proportional to that in another bioassay. Further, this comparability is as- sumed to apply across biological systems and species. This assumes that the bioavail- ability of the active compounds at the tar- get tissue is proportional even when ex- trapolations are made between species and between routes of exposure. As Cuddihy and McClellan (1983) note, the estimates derived by this method suggest that ex- tracts of diesel exhaust particles are not "orders of magnitude more potent than other emissions." This risk assessment has a number of limitations. These studies and extrapola- tions are not based on whole, fresh ex- hausts; the exhausts are not acting on a . population exposed to a myriad of other carcinogens and active compounds unre- lated to diesel exhaust; and the exhausts are acting on a homogeneous population where genetic factors, prior illness, and personal habits do not influence the suscep- tibility to these insults. Also, this assess- ment offers an estimate of risk strictly for lung cancer, although cancers in other sites might also be affected. The risk estimate is also provided as specific risk per unit of exposure. This specific risk is not very dissimilar to those for the other materials to which it was compared experimentally. Thus, the comparison with the specific risk for gasoline engines is somewhat mislead- ing when one considers the fact that diesel engines may generate one to two orders of magnitude more particles than a gasoline engine with a catalytic converter. More recently, the comparative potency approach has been used to assess the human cancer risk associated with diesel exhaust in a more comprehensive analysis that in- cludes estimates of population exposures (Cuddihy et al. 1984; McClellan 1986~. Those analyses used previously reported estimates of specific risk of lung cancer development (lung cancers/,ug diesel par- ticulates/m3 of air/year) (Albert et al. 1983; Cuddihy et al. 1983; Harris 1983~. Expo- sure estimates were based on environmen- tal concentrations in various locations and distribution of the population according to concentration levels and assuming a 20 percent proportion of diesel-powered light- duty vehicles. Analyses were also based on the estimates of risk from epidemiologic studies. From these data, calculated values for excess lung cancer deaths per year ranged from 100 to 3,500, a range attribut- able to an increase to a 20 percent light- duty diesel-powered fleet. As with the ear- lier estimate noted above, there are numerous potential sources of error in these calculated risk values. Despite these limitations the risk estimate offers a starting point for determining the overall potential for changes in the rate of cancer deaths as a result of increasing the use of diesel engines in the U.S. transportation fleet. Too little generic information exists on the carcinogenicity of the gaseous and par

David G. Kaufman 545 ticulate emissions of mobile sources. Stud- ies have been performed on representative emissions generated by particular sources operating under specific conditions. Scien- tifically, it is not clear to what extent such results apply to different engines operating with different fuel or other different condi- tions. Further, it is not clear how these . . . results relate to the same engine operating under other conditions or even to the same engine operating under presumably identi- cal conditions at a different time. Differ- ences between individual engines or oper- ating conditions can result in the generation of emissions with quantitative and even qualitative differences in the products formed. These differences in turn can be the major determinant of the activity of the emission in carcinogenicity tests. This sit- uation is quite unlike the testing of a pure chemical, in which case there is a reason- able assurance of repeatability upon retest- ing. Given that there is no standard mixture for mobile source emissions, the question arises whether these mixtures can be eval- uated on the basis of quantity and activity of certain "sentinel" constituents. If these most active components could be moni- tored and minimized, then optimum en- gines and operating conditions might be selected. Although this idea has merit as a comparative measure, the assessment of the actual quantitative risk at any level of these sentinel compounds may be difficult to determine. A further complication is the fact that little is known about the possibil- ity of interactions between carcinogenic compounds. It is unlikely that the risk associated with the mixture of sentinel compounds is simply the sum of the effects of the individual compounds. This uncer- tainty is further amplified when the numer- ous other constituents of emissions are con- sidered as influencing the activities of the sentinel compounds. Further bioassays are needed to provide a sufficient body of information about the carcinogenicity of diesel exhausts in exper- imental animals. In view of the variability of diesel engine exhausts due to engine design, conditions of operation, and the fuel used, it is necessary to perform more studies to evaluate the influence of these variables on tumor responses. Exposure to diesel alone should be complemented by studies in which diesel exhaust exposures follow initiating carcinogen treatment in each of several organs or tissues. Exposures should not be limited to a particular frac- tion of diesel exhaust condensate or to the particulate material, since the complete ex- hausts may have additive or inhibitory ef- fects that would otherwise not be detected. ~ Recommendation 6. Additional stud- ies should be performed on the carcinoge- nicity of diesel exhaust. Assessment of the carcinogenicity of mixtures poses two conflicting problems. The first concerns the nature of mixtures and the fact that each mixture represents a unique case. The second concerns attempts to evaluate mixtures by dividing them into their constituents. In such cases it is difficult to determine how to reconstitute the effects of the individual constituents into the ef- fects of the total mixture. Mixtures such as the diverse combustion products in mobile source emissions can be administered to experimental animals and tested for carcinogenicity. However, these mixtures are not intentionally formulated with precise analytical procedures. They are the products of a process or source that may not have exceptional reproducibility. Thus, the emissions from two different diesel engines may have quantitative differ- ences in the products of combustion. Even the same engine, operating under slightly different, or even unmodified conditions, may yield mixtures of products with some quantitative differences. Despite the slight quantitative differences in the various com- bustion products, it is likely that the mix- tures will have similar qualitative effects: they will prove to be carcinogenic or they will not. The magnitude of the carcino- genic response may be affected by the ac- tual composition of the mixtures. Under any condition of operation or engine design, diesel exhaust is a mixture of many chemicals. The composition of this mixture is highly variable and depends on such factors as the engine, fuel, and oper

546 Assessment of Carcinogenicity ating conditions. It is conceivable that the Summary interaction of the components of this mix ture is the determinant of overall carcino genicity of the complete mixture. There fore, the only currently valid method to determine the carcinogenic activity of each form of diesel exhaust is by a separate animal bioassay. However, animal bioassay is not practi cal for evaluating modifications of diesel engine design or other aspects of their operation as they affect carcinogenicity. It would be valuable to have some method of estimating changes in carcinogenic activity based on knowledge about the changes in composition of diesel exhaust. This would require a better understanding of the in teractions of components of complex mixtures in causing cancer. To learn how constituents of mixtures interact in carcin ogenesis, it will be necessary to determine how carcinogenicity changes with the vari ation of the concentration of individual components. Choices of chemicals to study would presumably be based on the activ ity of the isolated compound or its rela tive abundance in the diesel exhaust mix ture. In addition, constituents that have demonstrated or are suspected of enhanc ing (or inhibiting, for that matter) the action of carcinogens (for example, pro moters or enhancers) will also need to be studied. This problem will require additional study to evaluate the effects of variation in the composition of diesel exhaust on the tumorigenicity of other carcinogens; this serves as a model of the multiple complex exposures associated with human environ ments and lifestyles. Optimally it would be possible to achieve a reasonable estimate for the complex exhaust mixture that is based on measurements of the concentrations of a certain small number of index compounds. This hypothesis and experimental approach should be tested and validated. If it is found to predict certain levels of carcinogenic activity, the predictions should be tested bv performing animal bioassays. , · Recommendation 7. Methods should be developed for assessing the carcinoge . . . . nlclty ot mixtures. This chapter reviews information on mech- anisms of carcinogenesis and considers fac- tors that influence the rate of tumor forma- tion. It also considers the criteria for identifying a chemical qualitatively as a carcinogen, and methods that have been used to extrapolate from these data to quantitative estimates of cancer risk in hu- mans. Finally, data have been reviewed regarding the qualitative assessment of die- sel exhaust as a carcinogen and the extrap- olations made using these data to estimate human cancer risk from diesel exhaust ex- posure. The review of current knowledge about . . . . carclnogenes1s points out great advances that have been made in our understanding of cancer and also reflects the vast remain- ing ignorance. Cancer development has come to be recognized as a slowly pro- gressing, multifactorial, multistep process. Cell proliferation is known to have one or several roles in the process, and abnor- malities of the control of this process are fundamental features of cancer cells. Fac- tors such as chronic injury or toxicity (for example, toxicity that is produced by high, but nonlethal doses of administered drugs) can result in elevated rates of cell proliferation with the attendant increase in cancer risk. Some chemicals are known to act as carcinogens by direct effects on DNA; in some cases, specific mutations induced by chemicals have been identified. Other genes whose effects are recognized in their absence or altered state in certain genetic diseases predisposing to cancer have been localized cytogenetically, and efforts are in progress to isolate the genes. Studies of atypical carcinogens that do not have a direct mutagenic effect have suggested alternate pathways or separate steps in the pathway to the development of cancer. Similarly, studies of the role of promoters in carcinogenesis have pointed out the multistep nature of the process, and the potential influence of factors that may act by selecting cells with abnormal properties. The human population is diverse in its genetic back- ground and its exposures to harmful

David G. Kaufman 547 materials. Many chemicals require enzy- matic activation to become reactive ulti- mate carcinogenic metabolites; these meta- bolic processes may be included among the factors influenced by genetics and environ- mental exposures, that determine the indi- vidual variations in susceptibility to cancer. The list of issues about the process of carcinogenesis considered in this section was necessarily incomplete, limited by space constraints rather than the exhaustion of important facts. The section on qualitative evaluations of carcinogenicity and quantitative estimates of cancer risks in humans considered the criteria for designating a chemical as a carcinogen and how these data are quanti- fied and extrapolated to estimates of human cancer risk. The section shows that there are reasonably well-defined criteria for judging whether a chemical is a carcinogen. That is not to say that this evaluation is not without problems. Weak carcinogenic re- sponses may be difficult to distinguish from background levels of tumors. Increased rates of tumors may be observed as the result of exposure to promoters of carcino- genesis; on the basis of a positive tumor response, the promoter is classified for- mally as a carcinogen. However, the tumor response produced by promoters may be critically dependent on dose and may even have a threshold level for activity. Conse- quently, promoters may become classified as carcinogens although their mode of ac- tion may be quite different from that of strong mutagenic carcinogens. The review of methods for quantitative risk assessment includes discussions of the several factors that must be considered in extrapolating estimates of human cancer risk. The discus- sion of extrapolation to humans from the species used in the carcinogenicity tests includes consideration of differences be . . . . . tween species in t :le sensitivity to tumor formation in particular organs. Also noted are the considerations that must be given to account for differences in dose and in the route of exposure to a compound when extrapolating from animal tests to estimates of human risk. The review of data specifically concerned with diesel engine exhaust emissions dem onstrates that these exhausts have biologi- cal activity. Short-term tests have shown that diesel engine exhausts are mutagenic and can cause chromosomal damage. The activity in these studies was influenced by the source of the emissions tested, for ex- ample, the type of engine used. A variety of studies have evaluated the activity of diesel engine exhaust as a complete carcinogen and as an initiator or promoter of carcino- genesis. Some of the studies failed to pro- duce positive results or were equivocal. Positive results, however, have been found in inhalation studies and in mouse skin painting and lung adenoma formation as- says. The results of these studies have been used with current, though admittedly im- perfect, risk extrapolation methods, and values for projected human cancers have been calculated. The risk for diesel engine exhaust was calculated to be comparable to the approximate range found for other car- cinogenic human exposures such as coke oven emissions and roofing tar. Within the limitations of these estimates, diesel engine exhausts do not appear to be notably more active than these other materials. Review of the epidemiologic studies of the risk of diesel engine exhausts shows that exposure to these exhausts does not cause a strong effect like cigarette smoking. How- ever, because of the limitations of the stud- ies, it is difficult to conclude conversely that the carcinogenic activity is negligible or absent. Summary of Research Recommendations: Priorities, Purposes, and Responsibilities Many factors must be considered in devel- oping a research plan that sets priorities for the pursuit of the various recommended research goals. For example, these prior- ities might be selected on the basis of the unique mission of the Health Effects Insti- tute, or they might be viewed on the basis of the more general need for furthering our knowledge about how to make quantitative risk assessments. From a practical point of view, it might be preferable to place the

548 Assessment of Carcinogenicity highest priority on goals that will require the longest time to accomplish or that are not getting adequate attention and support from other sources. It is also reasonable to place highest priority on goals that might significantly affect the cancer risks that might be attributable to diesel engine ex- haust, even without accomplishing all of the proposed research goals. If accomplishment of the unique mission of the Health Effects Institute is the per- spective from which priorities are deter- mined, then highest priority must be given to performing additional studies on the carcinogenicity of diesel exhaust (Recom- mendation 6) and developing methods for . . . . . assessing t he carclnogenlclty of mixtures (Recommendation 7~. It is unlikely that other sources or organizations will place comparable emphasis on the direct study of diesel exhaust as a carcinogen. The issue of the carcinogenicity of mixtures is a more general problem, but it is essential for the evaluation of diesel exhaust, although only a secondary concern in the evaluation of many other materials. If the view is taken that adequate assess- ment of the hazards of diesel exhaust will not be possible without obtaining more knowledge about how to make quantita- tive risk assessments in general, then prior- ities might be set somewhat differently. In this case, the highest priority might be placed on evaluating the role of toxicity in carcinogenesis (Recommendation 3) and gathering critical data for quantitative as- sessments (Recommendation 4~. By learn- ing how to make quantitative risk assess- ments that account for effects of toxicity, and which involve extrapolations to low doses, among routes of administration, anc among species, In a manner more firmly founded on scientific knowledge, better estimates of human cancer risks in general will become possible, and this will benefit the assessments of diesel ex- haust. Another basis for setting priorities might be consideration of practical issues. For example, priorities could be set so that the research goals all might be accomplished in the shortest time. From this perspective, highest priority might be placed on goals that require the longest time to accomplish. Thus, priority might be given to evaluating the role of toxicity in carcinogenesis (Rec- ommendation 3) Lathering critical data for . . ~ ~ 7 quantitative assessments (Recommenda- tion 4), and developing methods for assess- ing the carcinogenicity of mixtures (Rec- ommendation 7~. Each of these is a complex problem that will require the per- formance of long-term studies to accom- plish, and may require several such studies in sequence. By beginning these studies at the earliest time and phasing in other goals later, it might be possible to have the more complete body of knowledge with which to make scientific risk assessments at the . . . earliest time. Another perspective is to place the high- est priority on goals that are not receiv- ing adequate attention and support from other sources. It could be argued that many or most of the Research Recommen- dations are not being pursued with the vigor that might be desired. The conclu- sion from this, however, is that all of the Recommendations should be given a high priority. This point of view may be accu- rate but it does not contribute to a practical plan. Another view might be predicated on the idea that early availability of certain critical knowledge might make it possible to affect the cancer risks from diesel exhaust signif- icantly even before all of the needed infor- mation for scientific risk assessments has been obtained. A possible scenerio that might fit this perspective would place the highest priority on developing methods to identify individuals at high risk (Recom- mendation 2) and developing acceptable methods for dosimetry in humans (Recom- mendation 5~. For example, if one could identify the individuals who were at high risk for the development of cancer if they are excessively exposed to diesel engine exhaust, then it would be possible to focus preventive health measures on this group. If it were possible to carry out dosimetry on exposed individuals, then preventive measures might be developed that would reduce exposure and risk.

David G. Kaufman 549 Summary of Research Recommendations: A Research Plan From the preceding discussion, it is clear that there are many ways to assign priorities for the pursuit of the various Research Recommendations. The following plan considers these different perspectives in defining a preferred set of priorities. HIGH PRIORITY No other organization will commit comparable attention or resources to the study of diesel engine exhaust, and therefore this must be done by the Health Effects Institute. One would hope that research on the scientific problems in making critical extrapolations in quantitative risk assessment and in validating the process would be widely supported and actively pursued. Unfortunately, this need has been clear for some time, yet there has been less progress in solving this problem than might have been expected. Accom- plishment of the following two goals will provide the most urgently needed information to perform better assessments of the human risks resulting from diesel exhausts. Recommendation 4 Critical data should be gathered for quantitative assessments. Recommendation 6 Additional studies should be performed on the carcinogenicity of diesel exhaust. MEDIUM PRIORITY Development of methods for human dosimetry may benefit from investigator-initiated basic research and even from the re search of commercial enterprises. Therefore, the pursuit of these goals may be given somewhat lower priority. A similar lower priority may be given to the evaluation of the carcinogenicity of mixtures. This problem is not unique to the assessment of diesel exhausts and knowledge may be gained from studies supported by other regulatory programs. Recommendations Acceptable methods should be developed for dosimetry in humans. Recommendation 7 Methods should be developed for assessing the carcinogenicity of mixtures. LOW PRIORITY The remaining recommendations are important but are generic goals that would improve our general ability to make risk assess ments. These issues touch on basic research that is being pursued in investigator-initiated studies. Investigations of this type may also be pursued by other agencies that are required to make risk assessments. Thus, although these are important goals, they may deserve lower priority in this program.

550 Assessment of Carcinogenicity Recommendation 1 The role of promoters and enhancers in human carcinogenesis should be determined. Recommendation 2 Methods should be developed to identify individuals at high risk. Recommendation 3 The role of toxicity in carcinogenesis should be evaluated. Acknowledgments The author thanks Dianne Shaw for excel- lent technical editing and Brigitte Cooke for skillful secretarial assistance. References diesel and related environmental emissions: in vitro mutagenesis and oncogenic transformation, Envi- ron. Int. 5:403-409. Claxton, L. D. 1981. Mutagenic and carcinogenic potency of diesel and related environmental emis- sions: Salmonella bioassay, Environ. Int. 5:389-391. Cordeiro-Stone, M., Topal, M. D., and Kaufman, D. G. 1982. DNA in proximity to the site of replication is more alkylated than other nuclear DNA in S Phase 10T1/2 cells treated with N- methyl-N-nitrosourea, Carcinogenesis 3:1119-1127. Cuddihy, R. G., and McClellan, R. O. 1983. Evalu- ating lung cancer risks from exposures to diesel engine exhaust, Risk Anal. 3:119-123. Cuddihy, R. G., Griffith, W. C., and McClellan, R. O. 1984. Health risks from light-duty diesel vehicles, Environ. Sci. Technol . 18:14A-21 A. Curren. R. D.. Kouri. R. E.. Kim C. M.. and Albert, R. E., Lewtas, J., Nesnow, S., Thorslund, T. W., and Anderson, E. 1983. Comparative po- tency method for cancer risk assessment: applica- tion to diesel particulate emissions, Risk Anal. 3:101-117. Armitage, P. 1985. Multistage models of carcinogen- esis, Environ. Health Perspect. 63:195-201. Auerbach, O., Stout, A. P., Hammond, E. C., and Garfinkel, L. 1961. Changes in bronchial epithelium in relation to cigarette smoking and in relation to lung cancer, N. Engl. J. Med. 265:253-267. Barbacid, M. 1986. Oncogenes and human cancer: cause or consequence? Carcinogenesis 7:1037-1042. Barrett, J. C., and Tsto, P. O. P. 1978. Evidence for the progressive nature of neoplastic transformation in vitro, Proc. Natl. Acad. Sci. USA 75:3297-3301. Berenblum, I. 1975. Sequential aspects of chemical carcinogenesis: skin. In: Cancer: A Comprehensive Treatise (F. F. Becker, ed.), pp. 323-344, Plenum Press, New York. Berenblum, I., and Shubik, P. 1947. A new, quanti- tative, approach to the study of the stages of chem- ical carcinogenesis in the mouse's skin, Br. J. Cancer 1 :383-391. Bishop, J. M. 1983. Cellular oncogenes and retrovi- ruses, Ann. Rev. Biochem. 52:301-354. Brand, K. G., Buoen, L. C., Johnson, K. H., and Brand, I. 1975. Etiological factors, stages, and the role of the foreign body in foreign body tumorigen- esis: a review, Cancer Res. 35:279-286. Cairns, T. 1980. The EDo~ study: introduction, ob- jectives, and experimental design, J. Environ. Pathol. Toxicol. 3:1-7. Casto, B. C., Hatch, G. G., and Huang, S. L. 1981. Mutagenic and carcinogenic potency of extracts of Correspondence should be addressed to David G. Kaufman, Department of Pathology, School of Med icine, University of North Carolina, Brinkhous-Bul litt Building, 228H, Chapel Hill, NC 27514. , ~7 - - 7 Schechtman, L. M. 1981. Mutagenic and carcino- genic potency of extracts from diesel related envi- ronmental emissions: simultaneous morphological transformation and mutagenesis in BALB/c 3T3 cells, Environ. Int. 5:411-415. Doll, R. 1955. Mortality from lung cancer in asbestos workers, Br. J. Ind. 12:81-86. Doll, R. 1971. The age distribution of cancer: impli- cations for models of carcinogenesis, J. Roy. Soc. Med. 134:133-166. Drake, J. W., and Baltz, R. H. 1976. The biochemis- try of mutagenesis, Ann. Rev. Biochem. 45:11-37. Evans, H. J. 1983. Effects on chromosomes of carci- nogenic rays and chemicals, In: Chromosome Muta- tions and Neoplasia a German, ed.), pp. 253-279, A. R. Liss, New York. Farber, E. 1980. The sequential analysis of liver cancer induction, Biochim. Biophys. Acta 605:149-166. Farber, E. 1984. Chemical carcinogenesis: a current biological perspective, Carcinogenesis 5:1-5. Gaylor, D. W. 1980. The EDo, study: summary and conclusions, J. Environ. Pathol. Toxicol. 3:179-183. Grasso, P., and Hardy, J. 1974. Strain differences in natural incidence and response to carcinogens, In: Mouse Hepatic Neoplasia (W. H. Butler and P. M. Newberne, eds.), pp. 111-132, Elsevier Press, Amsterdam. Grisham, J. W., KauLmann, W. K., and Kaufman, D. G. 1983. The cell cycle and chemical carcinogen- esis, Sure. Synth. Pathol. Res. 1:49-66. Gullino, P. M., Pettigrew, H. M., and Grantham, F. H. 1975. N-Nitrosomethylurea as mammary gland carcinogen in rats, J. Nat. Cancer Inst. 54:401-414. Hall, N. E., and Wynder, E. L. 1984. Diesel exhaust

David G. Kaufman 551 exposure and lung cancer: a case-control study, Environ. Res. 34:77-86. Hanawalt, P. C., and Sarasin, A. 1986. Cancer-prone hereditary diseases with DNA processing abnor- malities, Trends Genet. 2:12~129. Harris, C. C., and Trump, B. F. 1983. Human tissues and cells in biomedical research, Sure. Synth. Pathol. Res. 1:16~171. Harris, J. E. 1983. Diesel emissions and lung cancer, Risk Anal. 3:83-100. Heino, M., Ketola, R., and Makela, P. 1978. Work conditions and health of locomotive engineers. I. Noise, vibration, thermal climate, diesel exhaust constituents, ergonomics, Scand. J. Work Environ. Health 4:3-14. Heinrich, U., Peters, L., Funcke, W., Pott, F., Mohr, U., and Stober, W. 1982. Investigation of toxic and carcinogenic effects of diesel exhaust in long-term inhalation exposure of rodents, In: Toxicologic Effects of Emissions~rom Diesel Engines U Lewtas, ed.), pp. 22~242, Elsevier Science Publishing, Inc., Amster- dam. Heinrich, U., Pott, F., Mohr, U., and Stober, W. 1985. Experimental methods for the detection of the carcinogenicity andfor cocarcinogenicity of inhaled polycyclic-aromatic-hydrocarbon-containing emis- sions, In: Carcinogenesis, A Comprehensive Survey. Vol. 8, Cancer of the Respiratory Tract: Predisposing Factors (M. J. Mass, D. G. Kaufman, J. M. Sieg- fried, V. E. Steele, and S. Nesnow, eds.), Vol. 8, pp. 131-146, Raven Press, New York. Heinrich, U., Muhle, H., Takenaka, S., Ernst, H., Fuhst, R., Pott, F., Mohr, U., and Stober, W. 1986. Chronic effects on the respiratory tract of hamsters, mice and rats after long-term inhalation of high concentrations of filtered and unfiltered diesel en- gine emissions, J. Appl. Toxicol. 6:383-397. Hueper, W. C. 1955. A Quest into the Environmental Causes of Carcinoma of the Lung. Public Health Monograph No. 36, U.S. Department of Health, Education and Welfare, Public Health Service. Huggins, C., Grand, L. C., and Brillantes, F. P. 1961. Mammary cancer induced by a single feeding of polynuclear hydrocarbons and its suppression, Na- ture 189:20~207. Ikenaga, M., and Kakunaga, T. 1977. Excision of tnitroquinoline 1-oxide damage and transforma- tion in mouse cells, Cancer Res. 37:3672-3678. International Agency for Research on Cancer. 1980. IARC Monographs on the Evaluation of the Carcino- genic Risks of Chemicals to Humans, Vol. 23, Some Metals and Metallic Compounds, IARC, Lyon, France. Kakunaga, T. 1975. The role of cell division in the malignant transformation of mouse cells treated with 3-methylcholanthrene, Cancer Res. 35:1637- 1642. Kakunaga, T., Crow, J. D., Hamada, H., Hirakawa, T., and Leavitt, J. 1983. Mechanisms of neoplastic transformation of human cells, In: Human Carcino- genesis (C. C. Harris and H. N. Autrup, eds.), pp. 371-399, Academic Press, New York. Kaplan, I. 1959. Relationship of noxious gases to carcinoma of the lung in railroad workers, J. Am. Med. Assoc. 171 :20302043. Klein-Szanto, A. J. P., Terzaghi, M., Mirkin, L. D., Nartin, D., Shiba, M. 1982. Propagation of normal human epithelial cell populations using an in viva culture system, Am. J. Pathol. 108:231-239. Kotin, P., Falk, H. L., and Thomas, M. 1955. Aro- matic hydrocarbons. III. Presence in particulate phase of diesel-engine exhausts and the carcinoge- nicity of exhaust extracts, AMA Arch. Ind. Hyg. Occup. Med. 11 :113-120. Kuschner, M. 1985. Perspective on pathologic predis- position to lung cancer in humans, In: Carcino- genesis, A Comprehensive Survey. Vol. 8, Cancer of the Respiratory Tract: Predisposing Factors (M. J. Mass, D. G. Kaufman, J. M. Siegfried, V. E. Steele, and S. Nesnow, eds.), pp. 17-21, Raven Press, New York. Lewtas, J., Bradow, R. L., Jungers, R. H., Harris, B. D., Zweidinger, R. B., Cushing, K. M., Gill, B. E., and Albert, R. E. 1981. Mutagenic and carcinogenic potency of extracts of diesel and re- lated environmental emissions: study design, sam- ple generation, collection and preparation, Environ. Int. 5:383-387. Lewtas, J., Nesnow, S., and Albert, R. E. 1983. A comparative potency method for cancer risk assess- ment: clarification of the rationale, theoretical basis and application to diesel particulate emissions, Risk Anal. 3:133-137. Li, A. P., and Royer, R. E. 1982. Diesel-exhaust- particle extract enhancement of chemical-induced mutagenesis in cultured Chinese hamster ovary cells: possible interaction of diesel exhaust with environmental carcinogens, Mutat. Res. 103:349- 355. Littlefield, N. A., Farmer, J. H., and Gaylor, D. W. 1980. Effects of dose and time in a long-term, low-dose carcinogenic study, J. Environ. Pathol. Toxicol . 3:17-34. Lockard, J. M., Kaur, P., Lee-Stephens, C., Sab- harwal, P. S., Pereira, M. A., McMillan, L., and Mattox, J. 1982. Induction of sister-chromatic ex- changes in human lymphocytes by extracts of par- ticulate emissions from a diesel engine, Mutat. Res. 104:355-359. Luepker, R. V., and Smith, M. C. 1978. Mortality in unionized truck drivers,J. Occup. Med. 20:677-682. Mauderly, J. L., Jones, R. K., Griff~th, W. C., Hen- derson, R. F., and McClellan, R. O. 1987. Diesel exhaust is a pulmonary carcinogen in rats exposed chronically by inhalation, Fundam. Appl. Toxicol. 9:208-221. McClellan, R. O. 1986. Health effects of diesel ex- haust: a case study in risk assessment, Am. Ind. Hyg. Assoc. J. 47: 1-13. McClellan, R. O., Bice, D. E., Cuddihy, R. G., Gillett, N. A., Henderson, R. F., Jones, R. K., Mauderly, J. L., Pickrell, J. A., Shami, S. G., and Wolff, R. K. 1986a. Health effects of diesel exhaust, In: Aerosols: Research, Risk Assessment and Control Strategies (S. D. Lee, T. Schneider, L. D. Grant, and P. J. Verkerk, eds.), pp. 597-615, Lewis Publishers, Inc., Chelsea, Mich.

552 McClellan, R. O., Mauderly, J. L., Jones, R. K., Henderson, R. F., and Wolff, R. K. 1986b. Lung tumor induction in rats by chronic exposure to diesel exhaust, Abstract of lecture presented at the Second International Aerosol Conference, Berlin, West Germany, September 22-26, 1986. Menck, H. R., and Henderson, B. E. 1976. Occupa- tional differences in rates of lung cancer, J. Occup. Med. 18:797-801. Merletti, F., Heseltine, E., Saracci, R., Simonato, L., Vainio, H., and Wilbourn, J. 1984. Target organs for carcinogenicity of chemicals and industrial ex- posures in humans: a review of results in the IARC Monographs on the Evaluation of the Carcinogenic Risk of Chemicals to Humans, Cancer Res. 44:2244-2250. Miller, J. A. 1970. Carcinogenesis by chemicals: an overview-G. H. A. Clowe's Memorial Lecture, Cancer Res. 30:559-576. Mitchell, A. D., Evans, E. L., end Jotz, M. M. 1981. Mutagenic and carcinogenic potency of extracts of diesel and related environmental emissions: in vitro mutagenesis and DNA damage, Environ. Int. 5: 393-401. Mitelman, F. 1986. Clustering of chromosomal breakpoints in neoplasia, Cancer Genet. Cytogenet. 19:67-71. Nachtman, J. P., Xiao-bai, X., Rappaport, S. M., Talcott, R. E., and Wei, E. T. 1981. Mutagenic activity in diesel exhaust particulates, Bull. Environ. Contam. Toxicol. 27:463-466. Nesnow, S., Triplett, L. L., and Slaga, T. J. 1981. Tumorigenesis of diesel exhaust, gasoline exhaust and related emission extracts on SENCAR mouse skin, In: Short- Term Bioassays in the Analysis of Complex Environmental Mixtures II (M. D. Waters, S. S. Sandhu, J. L. Huisingh, L. Claxton, and S. Nesnow, eds. ), pp. 277-297, Plenum Publishing Corp., New York. Nesnow, S., Triplett, L. L., and Slaga, T. J. 1982. Comparative tumor-initiating activity of complex mixtures from environmental particulate emissions on Sencar mouse skin, J. Nat. Cancer Inst. 68:829- 834. Nesnow, S., Triplett, L. L., and Slaga, T. J. 1983a. Mouse skin tumor initiation-promotion and com- plete carcinogenesis bioassays: mechanisms and bi- ological activities of emission samples, Environ. Health Perspect. 47:255-268. Nesnow, S., Triplett, L. L., and Slaga, T. J. 1983b. Mouse skin carcinogenesis: application to the anal- ysis of complex mixtures, In: Short-Term Bioassays in the Analysis of Complex Environmental Mixtures III (M. D. Waters, S. S. Sandhu, J. N. Lewtas, L. Claxton, N. Chernoff, and S. Nesnow, eds.), pp. 367-390, Plenum Publishing Corp., New York. Nettesheim, P., and Barrett, J. C. 1984. Tracheal epithelial cell transformation: a model system for studies on neoplastic progression, Crit. Rev. Toxi- col. 12:215-239. Orthoefer, J. G., Moore, W., Kraemer, D., Truman, F., Crocker, W., and Yang, Y. Y. 1981. Carcino- genicity of diesel exhaust as tested in strain A mice, Environ. Int. 5:461-471. Assessment of Carcinogenicity Pereira, M. A., Shinozuka, H., and Lombardi, B. 1981. Test of diesel exhaust emissions in the rat liver foci assay, Environ. Int. 5:455-458. Pereira, M. A., McMillan, L., Kaur, P., Gulati, D. K., Sabharwal, P. S. 1982. Effect of diesel exhaust emissions, particulates and extract on sister chromatic exchange in transplacentally exposed fe- tal hamster liver, Environ. Mutagen. 4:215-220. Pickle, L. P., Mason, T. J., Howard, N., Hoover, R., and Fraumeni, J. F. 1987. Atlas of U.S. Cancer Mortality Among Whites: 195~1980, pp. 1-149, Na- tional Institutes of Health, Bethesda, Md. Pitts, J. N., Lokensgard, D. M., Harger, W., Fisher, T. S., Mejia, V., Schuler,J.J., Scorziell, G. M., and Katzenstein, Y. A. 1982. Mutagens in diesel exhaust particulate: identification and direct activities of 6-nitrobenzota~pyrene, 9-nitroanthracene, 1-nitro- pyrene and 5H-phenanthro [(4,5-BCD)] pyran-5- one, Mutat. Res. 103(3-6) :241-249. Pour, P. M. 1984. Histogenesis of exocrine pancreatic cancer in the hamster model, Environ. Health Per- spect. 56:229-243. Raffle, P. A. B. 1957. The health of the worker, Br.J. Ind. Med. 14:7~80. Reddy, B. S., Weisburger, J. H., Narisawa, T., Wynder, E. L. 1974. Colon carcinogenesis in germ- free rats with 1,2-dimethylhydrazine and N- methyl-N'-nitro-N-nitrosoguanidine, Cancer Res. 34:2368-2372. Reddy, J. K., Azarnoff, D. L., and Hignite, C. E. 1980. Hypolipidemic hepatic peroxi-some prolif- erators form a novel class of chemical carcinogens, Nature 283:397-398. Saffiotti, U., and Harris, C. C. 1979. Carcinogenesis studies on organ cultures of animal and human respiratory tissue, In: Carcinogens: Identi~fcation and Mechanisms of Action (A. C. Griff~n and C. R. Shaw, eds.), pp. 6~82, Raven Press, New York. Saff~otti, U., and Kaufman, D. G. 1975. Carcino- genesis of laryngeal carcinoma, Laryngoscope 85: 454 457. Saffiotti, U., Cefis, F., and Kolb, L. H. 1968. A method for the experimental induction of broncho- genic carcinoma, Cancer Res. 28:100124. Sandberg, A. A. 1983. A chromosomal hypothesis of oncogenesis, Cancer Genet. Cytogenet. 8:277-285. Sarma, D. S. R., Rajalakshmi, S., and Farber, E. 1975. Chemical carcinogenesis: interactions of car- cinogens with nucleic acids, In: Cancer, A Compre- hensive Treatise (F. F. Becker, ed. ), Vol. I, pp. 23~287, Plenum Press, New York. Scarpelli, D. G., Rao, M. S., and Reddy, J. K. 1984. Studies of pancreatic carcinogenesis in different an- imal models, Environ. Health Perspect. 56:219-227. Schenker, M. B. 1980. Diesel exhaust an occupa- tional carcinogen?J. Occup. Med. 22:41-46. Schreiber, H., Saccomanno, G., Martin, D. H., and Brennan, L. 1974. Sequential cytological changes during development of respiratory tract tumors induced in hamsters by benzola]pyrene-ferric ox- ide, Cancer Res. 34:689~98. Selikoff, I. J., Hammond, E. C., and Churg, J. 1968. Asbestos exposure, smoking and neoplasia, J. Am. Med. Assoc. 204: 106-112.

David G. Kaufman 553 Setlow, R. B. 1978. Repair deficient human disorders and cancer, Nature 271:71~715. Shimosato, Y., Kodama, T., Tamai, S., and Kameya, T. 1980. Induction of squamous cell carcinoma in human bronchi transplanted into nude mice, Gann 71 :402007. Singer, B., and Grunberger, D. 1983. Molecular Biol- ogy of Mutagens and Carcinogens, pp. 4~219, Plenum Press, New York. Slaga, T. J., Fischer, S. M., Nelson, K., and Gleason, G. L. 1980. Studies on the mechanism of skin tumor promotion: evidence for several stages in promo- tion, Proc. Nat. Acad. Sci. USA 77:3659-3663. Slaga, T. J., Triplett, L. L., and Nesnow, S. 1981. Mutagenic and carcinogenic potency of extracts of diesel and related environmental emissions: two- stage carcinogenesis in skin tumor sensitive mice (Sencar), Environ. Int. 5:417023. Smith, B. L., and Sager, R. 1982. Multistep origin of tumor-forming ability in Chinese hamster embryo fibroblast cells, Cancer Res. 42:389-396. Stanton, M. F., and Wrench, C. 1972. Mechanisms of mesothelioma induction with asbestos and fibrous glass, J. Nat. Cancer Inst. 49:797-821. Steele, V. E., Marchok, A. C., and Nettesheim, P. 1980. Enhancement of carcinogenesis in cultured respiratory tract epithelium by 12-O-tetradecanoyl- phorbol-13-acetate, Int. J. Cancer 26:34~348. Swift, M., Sholman, L., Perry, M., and Chase, C. 1976. Malignant neoplasms in the families of pa- tients with ataxia-telangiectasia, Cancer Res. 36: 209-215. Valerio, M. G., Fineman, E. L., Bowman, R. L., Harris, C. C., Stoner, G. D., Autrup, H., Trump, B. F., McDowell, E. M., end Jones, R. T. 1981. Long-term survival of normal adult human tissues as xenografts in congenitally athymic nude mice, J. Nat. Cancer Inst. 66:849-858. Wagner, J. C., Berry, G., and Timbrell, V. 1973. Mesotheliomata in rats after inoculation with asbes- tos and other materials, Br. J. Cancer 28:17~185. Ward, J. M., Yamamato, R. S., and Brown, C. A. 1973. Pathology of intestinal neoplasms and other lesions in rats exposed to azoxymethane, J. Nat. Cancer Inst. 51 :1029-1039. Waxweiller, R. J., Wagner, J. K., and Archer, W. C. 1973. Mortality of potash workers, J. Occup. Med. 15:406-409. Wegman, D. H., and Peters, J. M. 1978. Oat cell cancer in selected occupations, J. Occup. Med. 20:79~796. Weinberg, R. A. 1985. The action of oncogenes in the cytoplasm and nucleus, Science 230:770-776. Wolff, R. K., Henderson, R. F., Snipes, M. B., Sun, J. D., Bond, J. A.; Mitchell, C. E., Mauderly, J. L., and McClellan, R. O. 1986. Lung retention of diesel soot and associated organic compounds, Abstract of lecture presented at the International Symposium on Toxicological Effects of Emissions from Diesel Engines, Tsukuba Science City, Japan, July 2~28, 1986. Wong, D., Mitchell, C., Wolff, R. K., Mauderly, J. L., and Jeffrey, A. M. 1986. Identification of DNA damage as a result of exposure of rats to diesel engine exhaust, Proc. Am. Assoc. Cancer Res. 27:84. Yunis, J. J., and Soreng, A. L. 1984. Constitutive fragile sites and cancer, Science 226:1199-1204.

Next: Potential Carcinogenic Effects of Polynuclear Aromatic Hydrocarbons and Nitroaromatics in Mobile Source Emissions »
Air Pollution, the Automobile, and Public Health Get This Book
×
Buy Paperback | $170.00
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

"The combination of scientific and institutional integrity represented by this book is unusual. It should be a model for future endeavors to help quantify environmental risk as a basis for good decisionmaking." —William D. Ruckelshaus, from the foreword. This volume, prepared under the auspices of the Health Effects Institute, an independent research organization created and funded jointly by the Environmental Protection Agency and the automobile industry, brings together experts on atmospheric exposure and on the biological effects of toxic substances to examine what is known—and not known—about the human health risks of automotive emissions.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!