Document xzjYJMQeONpBJwdJYqNVM7B41

Page 1 of 59 ToxSci AdvanceTAocxicceoslosgpicuabl lSischieendceDsecember 8, 2010 Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 Chemical Carcinogenesis 4 5 6 7 8 9 10 11 Samuel M. Cohen1,2 and Lora L. Arnold1 12 13 14 15 16 17 18 1Department of Pathology and Microbiology and the University of Nebraska Medical 19 20 Center/Eppley Cancer Center, University of Nebraska Medical Center, Omaha, Nebraska, 21 22 68198-3135, USA. 23 24 25 26 27 2Havlik-Wall Professor of Oncology; to whom correspondence should be addressed 28 29 30 (scohen@unmc.edu). 31 32 33 34 35 36 37 38 39 40 41 42 43 44 Running Title: Chemical Carcinogenesis 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 The Author 2010. Published by Oxford University Press on behalf of the Society of Toxicology. 60 All rights reserved. For Permissions, please email: journals.pe1rmissions@oup.com Toxicological Sciences Page 2 of 59 Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 4 ABSTRACT 5 6 Understanding the relationship of chemicals to carcinogenesis has progressed significantly since 7 8 the initial observations of Hill and Pott in the 1700s. Distinguishing between DNA reactive 9 10 11 chemicals and those which increase cancer risk by increasing cell proliferation has been a major 12 13 breakthrough in delineating overall mechanisms. Competing processes for activation versus 14 15 inactivation of chemicals occurs at many levels, including metabolism, DNA repair, and cellular 16 17 18 repair processes. These processes can be affected by other agents to decrease carcinogenesis 19 20 (chemoprevention). Increasing knowledge of the multiple steps of carcinogenesis is leading to 21 22 improved methods for screening chemicals for carcinogenic activity and for regulatory decision 23 24 25 making. Improvements in assessment of modes of action involved in animal and in vitro models 26 27 has led to more rational approaches to assessing relevance to humans. The advent of genomics 28 29 30 and high throughput technologies have contributed to investigations of mechanisms, and is 31 32 beginning to impact development of better methods for screening chemicals. Based on 33 34 developments in basic research, epidemiology, and astute clinical observations, the major risk 35 36 37 factors and etiologic agents have been identified for a majority of cancers, which is beginning to 38 39 lead to methods to decrease cancer incidence overall and identify targets for early detection and 40 41 treatment. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 2 Page 3 of 59 Toxicological Sciences Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 4 Historical Introduction 5 6 Chemical exposure has been related to the development of cancer ever since the observation by 7 8 John Hill (1761) that snuff users developed nasal cancer more frequently than the general 9 10 11 population. However, chemical carcinogenesis generally dates specifically to the observation by 12 13 Sir Percival Pott (1775) in 1775 describing the frequent occurrence of cancer of the scrotum in 14 15 chimney sweeps in England. He hypothesized that this was due to their significant exposure to 16 17 18 soot. More importantly, he also proposed a mechanism to reduce the incidence of these cancers 19 20 simply by requiring these individuals to bathe on a regular basis. This was instituted and the 21 22 incidence of scrotal cancer was essentially eliminated. It is important to note, given subsequent 23 24 25 scientific discoveries and emphasis on extrapolation to low exposure levels, that he did not 26 27 recommend eliminating the exposure completely, merely reducing it. Scrotal cancer today is a 28 29 30 rare disease. 31 32 33 34 Astute clinical observations such as that made by Pott have been the basis for the discovery of 35 36 37 many of the currently known classes of chemical carcinogens in humans. Examples include the 38 39 observation by Rehn (1895) in 1895 that workers in the aniline dye industry in Germany 40 41 frequently developed bladder cancer, and more recent observations concerning the induction of 42 43 44 angiosarcomas in patients exposed to contrast material used for radiologic imaging studies (Van 45 46 Kaick et al., 1986) and vinyl chloride exposure in the workplace in Louisville, Kentucky (Creech 47 48 and Johnson, 1974). 49 50 51 52 53 Research based on these observations led to several seminal discoveries in the history of 54 55 56 chemical carcinogenesis. Investigation of coal tar, for example, led to the first experimental 57 58 59 60 3 Toxicological Sciences Page 4 of 59 Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 induction of tumors in animal models by Yamagiwa and Ichikawa (1915) by painting this 4 5 6 material on the skin of rabbits and mice. These models have been used in carcinogenesis research 7 8 ever since. Similarly, the relationship of soot to cancer induction led to the purification and 9 10 11 identification of the first pure chemical carcinogen by Kennaway and Hieger (1930), when they 12 13 purified a small amount of dibenz[a,h]anthracene, and produced tumors by painting the chemical 14 15 on the backs of mice. The observation by Rehn of bladder cancer in the aniline dye industry led 16 17 18 to the discovery by Yoshida (1933) and Kinosita (1936) of the induction of liver cancer in rats by 19 20 o-aminoazotoluene. This research also demonstrated the importance of dietary effects on the 21 22 carcinogenic process. 23 24 25 26 27 Ultimately, 2-naphthylamine was identified as one of the principle chemicals to which workers 28 29 30 in the aniline dye industry were actually exposed, and Hueper, Wiley, and Wolfe (1938) 31 32 demonstrated that it was a bladder carcinogen when administered to dogs. In their study, they 33 34 were also the first to emphasize the importance of latency. Their experiment involved 35 36 37 administration of 2-naphthylamine to dogs for two years. The long latency period related to the 38 39 development of cancer after chemical exposure has been an important consideration in 40 41 theoretical models of carcinogenesis and in epidemiology ever since, and has been a major 42 43 44 barrier to the identification of additional chemicals and their possible relationship to cancer. This 45 46 was illustrated subsequently in numerous studies, such as demonstrating the relationship of 47 48 cigarette smoking to lung cancer (and subsequently to other cancers), even though exposure 49 50 51 frequently began during the teen years of the individuals whereas lung cancers did not usually 52 53 develop until after the age of fifty. Even with high exposures to potent chemical carcinogens, 54 55 56 such as 2-naphthylamine, benzidine, and vinyl chloride, the latency period is frequently twenty 57 58 59 60 4 Page 5 of 59 Toxicological Sciences Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 to thirty years or more. The basis for this long latency has posed a theoretical challenge to 4 5 6 scientists continuing to today, but the necessity for multiple specific genetic errors occurring 7 8 stochastically in a single cell provides the explanation for this latency (see below). 9 10 11 12 13 Metabolic Activation vs. Deactivation 14 15 Numerous chemicals have been identified with carcinogenic activity either in humans and/or in 16 17 18 animal models. It wasn't until the publication by the Millers (DeBaun et al., 1970; Miller and 19 20 Miller, 1977) in 1970 that a common mechanism could be demonstrated for many of the 21 22 chemicals from a variety of chemical classes. They were able to demonstrate that an aromatic 23 24 25 amide, 2-acetylaminofluorene(2-AAF) was metabolically activated to a reactive electrophile 26 27 which bound to DNA, forming DNA adducts and ultimately leading to mutations. Metabolic 28 29 30 activation has been a mainstay of carcinogenesis research ever since. A variety of metabolic 31 32 processes have been identified in the activation of a variety of classes of carcinogens, primarily 33 34 involving cytochrome P450 enzymes, but numerous other enzyme systems have also been 35 36 37 identified in the specific activation of chemicals (Guengerich, 2000). 38 39 40 41 It has also become apparent that nearly all chemicals undergo metabolism through several 42 43 44 competing enzyme pathways, with differences in kinetics and saturation levels (Guengerich, 45 46 2000). A common group of enzymes involved with carcinogen metabolism are the cytochrome 47 48 P450 isozymes. However, numerous other enzyme systems have been identified that participate 49 50 51 in the metabolism of various carcinogens (Mitchell and Smith, 2010). Although some of these 52 53 metabolic processes lead to activation to reactive electrophiles, many actually lead to 54 55 56 inactivation of the chemicals, by increasing aqueous solubility and leading to their increased 57 58 59 60 5 Toxicological Sciences Page 6 of 59 Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 excretion either in the urine or the feces (Figure 1). Thus, exposure to any chemical initiates 4 5 6 competing metabolic pathways for activation versus inactivation. Detailed research over the past 7 8 twenty years has shown that there are marked differences in the level of each of these enzyme 9 10 11 processes in human individuals, due to differences in activity of the enzymes secondary to single 12 13 nucleotide polymorphisms (SNPs), as well as environmental influences on induction or 14 15 inhibition of these enzyme systems (Boddy and Ratain, 1997). Although considerable research 16 17 18 has been performed examining the role of these SNPs on susceptibility to carcinogenesis, this 19 20 has not proven to be as influential as initially suspected. Frequently, the influence on cancer risk 21 22 is less than twofold between the various individuals, even though the metabolic activities 23 24 25 between specific enzymes can vary much more. This relatively small effect on cancer induction 26 27 probably reflects the numerous alternative pathways as well as competing pathways for 28 29 30 activation/inactivation of the chemical in addition to the multistep nature of the carcinogenic 31 32 process. 33 34 35 36 37 One of the most extensively studied metabolic systems influencing activation and inactivation 38 39 has been acetylation of aromatic amines (Hein, 2006). Individuals are generally classified as 40 41 slow or fast acetylators, which is genetically determined. The specific isozymes and genetic 42 43 44 variability have been extensively investigated. Aromatic amine acetylation dimorphism was 45 46 actually the first of the differences identified which led to the development of the field which 47 48 now is commonly referred to as molecular epidemiology (Lower and Bryan, 1973). In humans, 49 50 51 with respect to bladder cancer, acetylation of the aromatic amines is actually a deactivation 52 53 process. Thus, one can hypothesize that individuals who are fast acetylators should have a 54 55 56 decreased risk of developing bladder cancer in response to exposure to aromatic amines, in 57 58 59 60 6 Page 7 of 59 Toxicological Sciences Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 contrast to slow acetylators. Acetylation rates also influence the metabolism of a variety of 4 5 6 drugs, such as the anti-tuberculosis drug isoniazid, which can be clinically important. In several 7 8 epidemiology studies investigating occupational exposure to chemicals and exposure to cigarette 9 10 11 smoke, there is evidence that fast acetylators actually do have a decreased risk of developing 12 13 bladder cancer compared to slow acetylators (Hein, 2006). However, it has been suggested that 14 15 aromatic amines might also be related to the induction of colon cancer and even of breast cancer. 16 17 18 In these instances, it appears that acetylation might actually be an activation step, so that fast 19 20 acetylators would be expected to be at increased risk for developing these tumors (Shin et al., 21 22 2008). However, for all of these tumors, including the urinary bladder, the differences in 23 24 25 susceptibility are relatively weak and have not been observed in all epidemiologic investigations 26 27 examining this issue. Conflicting results have been reported for the examination of the influence 28 29 30 of nearly all SNPs on the development of various cancers. However, this remains an active area 31 32 of investigation and one involving a very complex interaction of variables. Not only is there the 33 34 issue of competing metabolic systems for activation and inactivation of a given chemical, there 35 36 37 are alternative pathways, variability in induction and inhibition of these enzymes due to 38 39 environmental influences, as well as an interaction with numerous other chemicals that might be 40 41 involved in the carcinogenic process for a given tissue in a given individual. 42 43 44 45 46 DNA Repair 47 48 The role of metabolic activation in carcinogenesis and the importance of DNA damage and 49 50 51 mutation have led to additional avenues of research regarding mechanisms of carcinogenesis and 52 53 influences on the carcinogenic process. The first of these is DNA repair (Hoeijmakers, 2009). 54 55 56 Cells constantly deal with the formation of DNA adducts, either from endogenous or exogenous 57 58 59 60 7 Toxicological Sciences Page 8 of 59 Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 influences (Hoeijmakers, 1009; Greenfield et al., 1984; Phillips et al., 2000), including DNA 4 5 6 damage that occurs secondary to radiation. The fact that cells do not either automatically die or 7 8 evolve into cancers because of these adducts is largely due to the fact that there is a remarkable 9 10 11 system of DNA repair processes that eliminate these specific alterations in DNA so that mutation 12 13 does not always occur. In fact, nearly all DNA adducts and damage are repaired, with only the 14 15 rare exception leading to the change in sequence and mutation (Figure 1). 16 17 18 19 20 The importance of DNA repair in the carcinogenic process was also demonstrated utilizing what 21 22 Dr. Robert Good (1968) referred to as experiments of nature, i.e., individuals born with specific 23 24 25 mutations related to specific cellular processes. Cleaver (Cleaver et al., 1975) identified that 26 27 individuals with xeroderma pigmentosum (XP) had a markedly increased rate of development of 28 29 30 skin cancers in response to sun exposure. He was able to demonstrate that the specific enzymes 31 32 involved in the various forms of XP were related to repair of the DNA damage that was induced 33 34 by UV radiation, the major causative factor of skin cancer in humans. Individuals with XP, 35 36 37 depending on the specific site of mutation, had varying degrees of inability to repair this DNA 38 39 damage. Thus, they had varying degrees of susceptibility to damage by the ultraviolet radiation 40 41 and varying susceptibilities to developing skin cancer. Some of these individuals were 42 43 44 exquisitely sensitive to the UV radiation, and developed multiple skin cancers early in life unless 45 46 they were completely protected from the UV radiation. For the development of cancer, 47 48 ultimately DNA mutation must occur, because of these alterations in the DNA sequence, either 49 50 51 due to single nucleotide changes or larger alterations in the chromosomes. Since these 52 53 abnormalities each carry with them specific repair processes, polymorphisms in the enzymes 54 55 56 57 58 59 60 8 Page 9 of 59 Toxicological Sciences Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 involved can lead to differences in susceptibility, just as differences in susceptibility can be due 4 5 6 to variations in metabolic activation and inactivation processes. 7 8 9 10 11 Genotoxicity Assays 12 13 The second major consequence of the observation that carcinogens were metabolically activated 14 15 to bind to DNA leading to mutation was the development of a variety of genotoxicity assays, 16 17 18 beginning with the Ames assay in various strains of Salmonella bacteria (McCann et al., 1975; 19 20 Gee et al., 1994). Ames and his colleagues developed strains of Salmonella that were particularly 21 22 susceptible to DNA damage, but more importantly they provided an exogenous source of 23 24 25 metabolic activation systems by adding induced liver microsomes to the medium along with the 26 27 test chemical and the test strain of bacteria. This addressed the issue of metabolic activation for 28 29 30 many of the chemicals being analyzed, since the enzymes involved in such processes frequently 31 32 are highest in the liver. This took advantage of the requirement for metabolic activation for most 33 34 DNA reactive carcinogens. Utilizing the Ames assay, numerous investigators were able to show 35 36 37 a strong correlation between positivity in this assay with a positive response in rodent bioassays 38 39 for carcinogenesis. This led to the now infamous statement that carcinogens are mutagens and 40 41 mutagens are carcinogens (Ames et al., 1973).. 42 43 44 45 46 A plethora of genotoxicity assays have been developed during the past four decades which were 47 48 designed not only as a means of identifying a mechanism by which a chemical can produce 49 50 51 cancer, but also as a short term screening test for identifying chemical carcinogens (Kirkland et 52 53 al., 2007). Although not a major focus of this discussion, besides the specific mutation assays 54 55 56 such as the Ames assay, many of the other assays have proven extremely difficult to interpret, 57 58 59 60 9 Toxicological Sciences Page 10 of 59 Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 especially regarding their relevance for predicting carcinogenicity (Kirkland et al., 2007. Many 4 5 6 have a predictive value for carcinogenesis less than 50%, less than a coin flip. This is because 7 8 many of these assays are in vitro and involve considerable levels of cytotoxicity. They also 9 10 11 involve indirect effects on DNA and chromosomes, rather than a direct interaction with DNA. 12 13 Such indirect assays include assessments for sister chromatid exchange, chromosomal 14 15 aberrations, and micronuclei. Furthermore, many of these in vitro assays cannot be verified in 16 17 18 vivo. Several in vivo genotoxicity assays have also been developed, including specific point 19 20 mutation assays involving specially designed mice and rats, such as the Big Blue rat, Big Blue 21 22 mouse and Mutamouse (Lambert et al., 2005), as well as in vivo micronucleus assays, the Comet 23 24 25 assay, and others. However, a more restrictive definition of genotoxicity is to limit the effects to 26 27 DNA reactivity specifically. This would include chemicals that react directly with DNA or 28 29 30 following metabolic activation to form DNA adducts (Phillips et al., 2000). Only those adducts 31 32 that are mutagenic (not all adducts have mutagenic potential) will lead to mutation and have the 33 34 potential to increase the incidence of cancer. 35 36 37 38 39 Extensive computer analyses relating three dimensional chemical structure to DNA reactivity has 40 41 resulted in a refined structure activity relationship (SAR) process (Tong et al., 2003). Several 42 43 44 computerized models are now available for this analysis, which can relate not only correlation 45 46 between chemical structure and metabolic activation to DNA reactivity, but also for a variety of 47 48 other biological targets, such as specific cell receptors. However, for a correlation to 49 50 51 carcinogenicity, the only significant correlation appears to be with identifying a chemical that 52 53 can be directly or metabolically activated for DNA reactivity. For chemicals that induce cancer 54 55 56 by mechanisms that do not involve DNA reactivity these models are of little value. However, 57 58 59 60 10 Page 11 of 59 Toxicological Sciences Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 such SAR models are of considerable value in screening new chemicals for commercial 4 5 6 development, with the ability to exclude those that have the potential for DNA reactivity. 7 8 Ultimately, the ability of a chemical to be DNA reactive can be demonstrated by an evaluation of 9 10 11 DNA adduct formation either in vitro, or better, in vivo. Especially now that there are 12 13 extraordinarily sensitive techniques available for assessing DNA adducts, this is considered the 14 15 best and most definitive method for demonstrating DNA reactivity. Demonstrating DNA adduct 16 17 18 formation in vivo in the target tissue is best since in vitro findings do not always translate to in 19 20 vivo effects. Furthermore, possible reactive chemical substituents identified by SAR do not 21 22 always relate to actual metabolic activation and DNA adduct formation in vivo. For example, 23 24 25 although numerous aromatic amines can be metabolically activated to DNA reactive substances, 26 27 there are numerous aromatic amines that are not metabolically activated and do not form DNA 28 29 30 adducts. These include not only substances to which we are exposed exogenously, but substances 31 32 that are generated endogenously through normal intermediary metabolism, such as kynurenine, 33 34 anthranilic acid and other metabolites (Bryan, 1969; Seifried et al., 2006). 2-Naphthylamine is 35 36 37 metabolically activated to a reactive electrophile and is mutagenic and carcinogenic. Its isomer, 38 39 1-naphthylamine, is not metabolically activated and is neither mutagenic nor carcinogenic 40 41 (Clayson and Cooper, 1970). 42 43 44 45 46 DNA Reactive vs. Non-DNA Reactive 47 48 Although these genotoxicity assays, specifically DNA reactivity assays, have been a major 49 50 51 development in chemical carcinogenesis, both with respect to determination of mechanism and 52 53 for screening new chemicals for potential genotoxic activity, it was already apparent at the time 54 55 56 when many of these assays were being developed in the 1970's that there were a large number of 57 58 59 60 11 Toxicological Sciences Page 12 of 59 Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 chemicals that produced cancer in animal models (as well as in humans) that were not positive in 4 5 6 any of these assays. These included such chemicals as phenobarbital (Whysner et al. 1996), 7 8 saccharin (IARC, 1999a), as well as others. Weisburger and Williams (1981) were the first to 9 10 11 publish an article stating that there were essentially two classes of chemical carcinogens, 12 13 genotoxic and non-genotoxic. They referred to the non-genotoxic carcinogens as epigenetic. 14 15 Refinement of this dichotomy is to separate chemicals on the basis of DNA reactivity. As 16 17 18 discussed below, those chemicals increasing the risk of cancer that are not DNA reactive do so 19 20 by increasing the number of DNA replications in the target cell population (increased cell 21 22 proliferation). Distinction between classes of chemicals based on their ability to generate DNA 23 24 25 reactivity is the basis for the classification of chemical carcinogens to this day and forms the 26 27 basis for the distinction of potential risks to humans in regulatory decision making. 28 29 30 31 32 Multistage Carcinogenesis 33 34 A major focus of carcinogenesis research has been the issue of latency. A major basis for this 35 36 37 delay in the development of cancer secondary to exposure to chemicals or other carcinogenic 38 39 agents is that cancer is a multi-step process. The concept that cancer involves multiple steps was 40 41 already being investigated as early as the 1930's, but it was not until the seminal publication of 42 43 44 Berenblum and Shubik (1947) in 1947 that a model distinguishing steps in carcinogenesis was 45 46 actually demonstrated. This model was referred to as initiation and promotion, and was later 47 48 amended to include a third step, progression. In their model, Berenblum and Shubik showed that 49 50 51 promotion actually involved clonal expansion of initiated cells to form benign lesions, with 52 53 progression being the step needed to convert these benign lesions to malignancy. 54 55 56 57 58 59 60 12 Page 13 of 59 Toxicological Sciences Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 The two steps of the carcinogenic process, initiation and promotion, were shown to have distinct 4 5 6 qualities (Berenblum and Shubik, 1947; Boutwell, 1964; Dragan et al., 1993). To begin with, it 7 8 was shown that the sequence of administration was critical, so that the initiator had to be 9 10 11 administered first followed by the promoter. Also, it was subsequently shown that initiation was 12 13 an irreversible event, what we now realize to be a permanent alteration in the DNA, whereas 14 15 promotion could be reversible, at least until the step of progression to a malignancy occurred. 16 17 18 Even the benign tumors that were produced in the Berenblum and Shubik model, skin papillomas 19 20 on the backs of mice, could revert to normal. Another characteristic of initiation was that the 21 22 effect appeared to be additive. If a dose of the initiator was fractionated into multiple parts, but 23 24 25 the cumulative amount administered remained the same, similar tumor incidences and numbers 26 27 were induced. In contrast, the dose of the promoter was found not to be additive. There appeared 28 29 30 to be a specific minimum fractional exposure that was required for promotion to occur, i.e., a 31 32 threshold. These basic concepts have evolved into current perspectives on mechanism as well as 33 34 regulatory decision making regarding chemicals. 35 36 37 38 39 In reality, we now know that initiation (and progression) is essentially the process of producing 40 41 irreversible DNA damage, and initiators are chemicals that are DNA reactive, either directly or 42 43 44 following metabolic activation (Dragan et al., 1993). Promoters, in contrast, are non-DNA 45 46 reactive chemicals, and as discussed below, produce their effect by increasing cell proliferation. 47 48 Nearly all DNA reactive carcinogens produce toxicity when administered at high doses, which 49 50 51 will lead to increased cell proliferation, a promoting effect. This has given rise to the term 52 53 complete carcinogen. As described in detail below, the terms initiation, promotion and complete 54 55 56 carcinogen become difficult to define except to classify chemicals as having the properties of 57 58 59 60 13 Toxicological Sciences Page 14 of 59 Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 DNA reactivity, increased cell proliferation or both. Also, DNA replication by itself can lead to a 4 5 6 "spontaneous" alteration in the DNA, effectively acting to produce initiation. 7 8 9 10 11 Although the model of initiation and promotion has served the carcinogenesis community well 12 13 with respect to investigating the multi-step nature of carcinogenesis and the carcinogenic 14 15 properties of chemicals, there actually are several difficulties with this model (Cohen and 16 17 18 Ellwein, 1991; Cohen 1998b). To begin with, administration of the promoter prior to the initiator 19 20 has been shown to actually increase the carcinogenic effect. Furthermore, initiation-promotion 21 22 assays have traditionally been performed in short term experiments, usually twenty weeks or less 23 24 25 in mouse skin studies, whereas evaluation of these chemicals, initiators and promoters, in a full 26 27 two year bioassay, today's standard, showed they were carcinogenic in their own right. 28 29 30 Furthermore, this model requires the presence of an intermediate clonal expansion of initiated 31 32 cells for the development of the ultimate malignancy. Although in many animal models this 33 34 occurs, such as mouse skin, rat liver, mouse and rat colon, mouse and rat lung, and mouse and rat 35 36 37 urinary bladder, this is not always the case either in rodents or especially in humans. There are 38 39 numerous malignancies in humans that occur without an intermediate benign lesion identifiable, 40 41 such as certain leukemias, most sarcomas, small cell carcinoma of the lung and other tissues, and 42 43 44 others. The malignant potential in these tumors appears to occur with the initial transformation of 45 46 the cell without an intervening benign proliferation. As described below, a more generalizable 47 48 multi-stage model can incorporate the concepts of the initiation-promotion model, but also 49 50 51 includes other factors. 52 53 54 55 56 57 58 59 60 14 Page 15 of 59 Toxicological Sciences Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 A second multi-stage model was proposed by Armitage and Doll (1954) based on epidemiologic 4 5 6 observations. They recognized that the incidence of many cancers increased exponentially with 7 8 age. They derived an equation to address this observation and noted that by addressing the 9 10 11 specific data for a given tumor, that the number of stages involved in the carcinogenic process 12 13 could be identified (Figure 2). This formula fitted the data well for cancers of many organs, such 14 15 as lung and urinary bladder. However, there are many examples for which this model does not fit 16 17 18 (Figure 3). For example, in breast cancer there is a bimodal appearance in the curve for times of 19 20 increasing incidence, the peri-menopausal time period and later in age (Moolgavkar and 21 22 Knudson, 1981). Another example is Hodgkin's disease in the United States (Watanabe et al., 23 24 25 1986). Again, there is a bimodal distribution of increased incidences, one in the twenties and 26 27 thirties and then again in later age. An even more striking example are testicular germ cell 28 29 30 tumors in males, where there is a dramatic increase in incidence between ages twenty through 31 32 forty, but these tumors rarely occur after the age of fifty (Kodama and Kodama, 1998). And, of 33 34 course, there are the tumors that occur only or predominantly in childhood, such retinoblastoma 35 36 37 (Knudson, 1971). 38 39 40 41 In the approach developed by Armitage and Doll (1954), implicit assumptions are made that 42 43 44 frequently are valid, but in specific instances are not. One of these assumptions is that the rate of 45 46 cell proliferation in a given tissue remains constant throughout lifetime, and another is that the 47 48 number of target cells remains constant throughout the lifetime. Both of these variables can be 49 50 51 influenced by normal growth and development and by exogenous factors. For example, 52 53 osteosarcomas have peak incidences during the major growing period of humans, ages fifteen to 54 55 56 57 58 59 60 15 Toxicological Sciences Page 16 of 59 Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 twenty-five. This of course is when the bones are growing and osteoblasts are proliferating at a 4 5 6 much greater rate and number than in the adult. 7 8 9 10 11 A third multi-stage model was proposed by Knudson (1971) in a landmark 1971 publication 12 13 describing the occurrence of retinoblastomas in children. Until Knudson's publication, 14 15 retinoblastoma was frequently considered an autosomal dominant disorder since nearly all 16 17 18 children that inherited the disorder developed tumors. However, he posed the question as to why 19 20 individuals who had inherited this disorder did not have all of their retinoblasts develop into 21 22 tumors. Although individuals who had inherited retinoblastomas usually had bilateral disease and 23 24 25 multiple tumors in each eye, it was still a limited number, with most of the retinoblasts in the 26 27 children's eyes being morphologically normal, never becoming malignant. In contrast, 28 29 30 individuals who had the sporadic disease usually had unilateral disease and usually only had one 31 32 tumor. 33 34 35 36 37 Knudson (1971) postulated that the retinoblastoma gene acted as an autosomal recessive gene 38 39 and that individuals who had the inherited genetic abnormality actually had damage to one of the 40 41 alleles in all of their retinoblasts, whereas the other retinoblastoma allele that was inherited was 42 43 44 normal. During normal growth and development, retinoblasts proliferate. In an occasional cell, 45 46 the second allele spontaneously developed an abnormality, leading to the development of the 47 48 malignancy. In contrast, in individuals developing retinoblastomas on a sporadic basis, a 49 50 51 spontaneous error had to occur in both alleles during normal growth and development during cell 52 53 replication of the retinoblast. Even though spontaneous errors in DNA are rare events during 54 55 56 replication, the number of replications occurring normally during development provides 57 58 59 60 16 Page 17 of 59 Toxicological Sciences Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 sufficient opportunity for an allele to become abnormal. In an individual with one allele already 4 5 6 abnormal, the child virtually always will develop an abnormality in the second allele at least 7 8 once, and usually in multiple such cells, leading to the development of multiple tumors. 9 10 11 However, having two rare events occur in same cell during normal development would be an 12 13 extraordinarily rare event; approximately one individual per million births would be expected to 14 15 develop such an abnormality. This is roughly the incidence of retinoblastoma in the general 16 17 18 population on a worldwide basis. The necessity for cell replication becomes obvious once the 19 20 eye fully develops and the retinoblasts stop replicating; tumor induction no longer occurs! 21 22 Although for many cancers, more than two events are required, in contrast to retinoblastoma, 23 24 25 nevertheless, the general principles identified by Knudson in his model readily address the 26 27 inadequacies of both the initiation-promotion model and the Armitage-Doll model. 28 29 30 31 32 It is obvious that the retinoblastoma gene as depicted by Knudson foresaw the definition of a 33 34 tumor suppressor gene. That is, under normal circumstances, when the gene is functional, it 35 36 37 actually inhibits the formation of cancer. It is only with the development of an abnormality in 38 39 both alleles that this inhibition on cancer development is released and cancer occurs. 40 41 42 43 44 45 46 In contrast to tumor suppressor genes are the oncogenes (Hahn et al., 1999; Knudson, 1993). 47 48 These are a group of genes that appear to act in a dominant fashion, and when activated, either 49 50 51 by a mutation or by other cellular processes, increase cell proliferation and produce cancers. In 52 53 general, all tumor suppressor and oncogenes that have been identified are related in one way or 54 55 56 another to cell proliferation or to the preservation of the DNA sequence, either directly or 57 58 59 60 17 Toxicological Sciences Page 18 of 59 Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 indirectly. A sequence of oncogene activation and tumor suppressor gene inactivation has been 4 5 6 best defined for colonic adenocarcinoma by Vogelstein and others (Wood et al., 2007). These 7 8 models provide a molecular basis for the Knudson hypothesis more generally and for the model 9 10 11 described below. However, in the colon model and in others, it remains difficult to define exactly 12 13 which genes are essential to the carcinogenic process and which ones modify the cell number 14 15 and/or replication rate and/or mutation frequency. 16 17 18 19 20 An incredibly astute series of observations published by Boveri (1914) in 1914, actually 21 22 described the fundamental concepts of oncogenes, genes activating cancer, as well as tumor 23 24 25 suppressor genes, genes inactivating cancer development. Keep in mind that this was at a time 26 27 when chromosomes were first being identified as genetic material, and DNA had not yet been 28 29 30 identified as the source of heredity, nor had modifications in DNA structure or sequence by 31 32 exogenous or endogenous factors been discovered. These seminal, astute observations were 33 34 essentially ignored by the scientific community for several decades. 35 36 37 38 39 In the early 1980's, a more generalized multi-stage model of carcinogenesis was developed by 40 41 Moolgavkar and Knudson (1981) utilizing epidemiology data and by my laboratory (Greenfield 42 43 44 et al, 1984). These models are based on a two-step process, but can easily be broadened to a 45 46 multi-step process, as it has become apparent that for most cancers more than two steps are 47 48 involved (Figure 4). 49 50 51 52 53 Cell Proliferation and DNA Reactivity in Carcinogenesis 54 55 56 57 58 59 60 18 Page 19 of 59 Toxicological Sciences Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 The target cell population in a given tissue is assumed to be the pluripotential cells in that tissue 4 5 6 (Moolgavkar and Knudson, 1981; Greenfield et al., 1984; Cohen and Ellwein, 1990a). Cells that 7 8 have already committed to differentiation or are fully differentiated are not the targets for 9 10 11 carcinogenesis, since they are ultimately destined to die. It is only the pluripotential cells (also 12 13 referred to as tissue stem cells) that have the potential to develop as malignancies. In contrast, 14 15 many benign tumors actually arise from cells that are already committed to differentiation. 16 17 18 19 20 Under normal circumstances, when a pluripotential cell divides it produces another cell identical 21 22 to itself, and another cell that is committed to differentiation, and ultimately undergoes terminal 23 24 25 differentiation (Figure 4A). Committed cells and terminally differentiated cells are destined to 26 27 die. Thus, inducing mistakes in the DNA of these cells will not produce a tumor that can evolve 28 29 30 with all of the characteristics of a malignancy. Stem cells also can die, commonly killed by a 31 32 variety of processes including necrosis and apoptosis. If these cells die, the remaining daughter 33 34 cells can divide to form two daughter stem cells to replace the lost cells, or stem cells can be 35 36 37 directly stimulated to replicate into two daughter stem cells (direct mitogenesis) (Figure 4B). If 38 39 there is injury, this is the way tissues are repaired and how this population can be replenished. 40 41 Direct mitogenesis will lead to an increase in the number of cells that are present as well as 42 43 44 usually inducing replication at an increased rate. 45 46 47 48 During DNA replication in the pluripotential cells, whether under normal circumstances 49 50 51 generating one daughter cell and a committed cell, or whether duplicating itself into two 52 53 pluripotential cells, mistakes can occur in the DNA which are fixed permanently and inherited by 54 55 56 the cells (Figure 4C). If the mistake occurs in the portion of a gene that is critical to the 57 58 59 60 19 Toxicological Sciences Page 20 of 59 Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 development of a given cancer for that tissue type, then the cell has taken a step toward the 4 5 6 development of malignancy and enters into what we refer to as the intermediate cell population. 7 8 This population can be clonally expanded, as described in the initiation-promotion model, but 9 10 11 this does not have to occur. 12 13 14 15 Cells in the intermediate population can undergo the same processes as normal cells, i.e., they 16 17 18 can replicate yielding one of themselves and one committed cell with terminal differentiation, or 19 20 they can replicate yielding two identical daughter cells. Again, during DNA replication of these 21 22 intermediate pluripotential cells, mistakes can occur in the DNA. If the mistake is in the critical 23 24 25 portion of a gene that is critical in the pathway to developing cancer in that tissue type, then that 26 27 step will progress further toward malignancy. If it is a two step process, then the second 28 29 30 abnormality that occurs in this intermediate cell population will lead to a malignancy. If there are 31 32 more than two steps, there will be multiple intermediate cell populations that are required for the 33 34 ultimate development of malignancy. 35 36 37 38 39 Once the final alteration occurs and malignancy develops, in contrast to what has long been held 40 41 as true for malignancies, not all of these malignant cells will replicate and some of them can 42 43 44 yield fully differentiated cells. This is the basis for tumors ranging from well-differentiated to 45 46 poorly-differentiated. In well-differentiated tumors, the sequence of pluripotential cell to 47 48 committed cells to fully differentiated cells is maintained, albeit at a slower rate than under 49 50 51 normal circumstances so that a greater proportion of the cells in the tissue remain as the 52 53 pluripotential cells rather than the fully differentiated cells. In contrast, poorly differentiated 54 55 56 malignancies have few cells that progress in this direction, andmost remain in the pluripotential 57 58 59 60 20 Page 21 of 59 Toxicological Sciences Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 stem cell population. Even in these cells, the proliferation rate is not uniform, and only a 4 5 6 relatively small proportion of the cells will replicate in any given day. There is only one tumor 7 8 that I am aware of that actually yields replication of nearly all of the pluripotential cells on a 9 10 11 daily basis and that is Burkitt's lymphoma (Cooper et al., 1965). 12 13 14 15 Fundamental to this model is that the transitions from normal to intermediate to malignancy 16 17 18 occur during DNA replication, each with a certain probability that is usually very low. It is a 19 20 stochastic process. The stochastic nature of carcinogenesis combined with the multiple DNA 21 22 abnormalities that are required form the basis for the latency period required between chemical 23 24 25 exposure and cancer development. 26 27 28 29 30 Based on this sequence of events, there are fundamentally only two mechanisms by which a 31 32 chemical, or any other agent can increase the risk of cancer. Either the chemical can increase the 33 34 probability that at each replication of the pluripotential stem cell there will be a mistake in a 35 36 37 critical gene, or the number of replications in this population can increase, or a combination of 38 39 these two events (Moolgavkar and Knudson, 1981; Greenfield et al., 1984; Cohen and Ellwein, 40 41 1990a). Chemicals that increase the probability of a genetic abnormality with each DNA 42 43 44 replication are essentially those that are DNA reactive. Those that increase the amount of DNA 45 46 proliferation as the basis for increasing cancer risk are those classified as non-DNA reactive. For 47 48 nearly all DNA-reactive carcinogens, if the dose is high enough, there is also an increase in self- 49 50 51 proliferation, usually due to regeneration following cytotoxicity. 52 53 54 55 56 57 58 59 60 21 Toxicological Sciences Page 22 of 59 Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 Increased cell proliferation can occur either by an increase in cell births or a decrease in cell 4 5 6 deaths (Cohen, 1998b). A decrease in cell deaths will lead to an accumulation of cells. Even if 7 8 they are replicating at the normal rate, this still represents an increase in the number of cells that 9 10 11 are replicating. This is a common mistake made in interpreting data on cell proliferation since 12 13 many investigators focus on cell rate rather than on the actual number of replications. For 14 15 example, with chemicals such as phenobarbital there is a transient increase in DNA replication 16 17 18 rate in the liver, but this is accompanied by a growth in the liver with an increase in the number 19 20 of target hepatocytes (Whysner et al., 1996). Thus, for the remainder of the animal's time 21 22 exposed to phenobarbital, there actually is an increase in the number of cells replicating, but not 23 24 25 the rate. Furthermore, it is critical that this proliferation occur in the stem cell population. An 26 27 increase in proliferation in those cells that are committed to differentiation or even fully 28 29 30 differentiated, will not lead to an increase in cancer risk. This is another common difficulty in 31 32 interpreting cell proliferation data. 33 34 35 36 37 Synergy Between Cell Proliferation and DNA Reactivity 38 39 A landmark study in carcinogenicity research is the ED01 mega-mouse experiment performed in 40 41 the 1970's at the National Center for Toxicological Research. In this study, , 2-AAF was 42 43 44 administered to female mice for up to thirty-three months, with group sizes sufficiently large to 45 46 detect an increased incidence of tumors of one percent, rather than the usual two year bioassay 47 48 which has a detection limit of approximately ten percent (SOT, 1981; Cohen and Ellwein, 49 50 51 1990b). The doses used were substantially lower than those typically used in 2-AAF 52 53 experiments, ranging from a low dose of 30 ppm to a high dose of 150 ppm. At these doses there 54 55 56 was approximately a linear dose response relationship for liver tumors, but for urinary bladder 57 58 59 60 22 Page 23 of 59 Toxicological Sciences Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 tumors there was an increased incidence of tumors only at doses of 60 ppm and above. However, 4 5 6 studies of DNA adduct formation showed steady-state levels in both tissues that were linear with 7 8 dose, extending to doses much lower than those used in the carcinogenicity bioassay (Poirier et 9 10 11 al., 1989). In the liver, the tumor effect was due only to the interaction of 2-AAF with the DNA 12 13 of the normal hepatocytes, with this interaction occurring much less frequently in the 14 15 intermediate cell population (foci cells) (Cohen and Ellwein, 1990b). There was no apparent 16 17 18 increase in cell proliferation in either population compared to untreated controls. In contrast, at 19 20 low doses of 2-AAF, the effect of DNA reactivity on the bladder was insufficient to produce a 21 22 detectable incidence of tumors. Keep in mind that the normal mouse bladder only has 23 24 25 approximately twenty-five thousand pluripotential cells, and therefore, only by increasing this 26 27 target cell population and their rate of replication, which occurred at the higher doses, was a 28 29 30 detectable incidence of tumors observed. 31 32 33 34 This synergistic interaction between DNA reactivity and cell proliferation has been identified in 35 36 37 several model systems and in humans. For example, FANFT is a bladder carcinogen in rats 38 39 similar to 2-AAF, with a detectable incidence in the two year bioassay occurring only at doses 40 41 that also increase cell proliferation (hyperplasia) (Murasaki and Cohen, 1983). At lower doses, 42 43 44 the DNA reactivity is still present but the effect does not produce a detectable incidence of 45 46 tumors. In contrast, sodium saccharin at very high doses is not DNA reactive, but increases cell 47 48 proliferation in the urinary bladder epithelium. In a standard two year bioassay, it increases 49 50 51 tumor incidences by approximately one percent. When FANFT administered at a dose that 52 53 produces DNA reactivity but does not increase cell proliferation or a detectable incidence of 54 55 56 57 58 59 60 23 Toxicological Sciences Page 24 of 59 Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 tumors, is co-administered with a dose of sodium saccharin that increases cell proliferation, a 4 5 6 detectable incidence of tumors is identified. 7 8 9 10 11 In humans, examples also have been identified. For example, the aromatic amine 4- 12 13 aminobiphenyl (4-ABP) is present in cigarette smoke and is excreted in the urine in forms that 14 15 lead to DNA adduct formation (Cohen et al., 2006). However, the amounts present in cigarette 16 17 18 smoke are at levels that would be expected to be insufficient to induce the incidence of bladder 19 20 tumors in humans that is actually observed. In addition to the DNA adduct formation, however, 21 22 there is also an increase in proliferation of the urothelium in cigarette smokers. The cause of this 23 24 25 increased proliferation has not been identified, but would not be expected to be due to the 4- 26 27 ABP. However, this combination of increased mutagenic DNA adduct formation by 4-ABP and 28 29 30 increased cell proliferation leads to a significant incidence of bladder tumors, with cigarette 31 32 smoking being the major cause of bladder cancer in the United States. 33 34 35 36 37 The interaction of multiple agents has also been identified in liver carcinogenesis involving 38 39 aflatoxin and hepatitis-B virus (HBV) (Chuang et al., 2009). In parts of China there is exposure 40 41 to high levels of aflatoxin, a DNA reactive carcinogen, but without HBV exposure. The 42 43 44 estimated increase in risk of liver cancer is approximately two to three fold compared to 45 46 individuals not exposed to high levels of aflatoxin. In other parts of China, HBV is prevalent but 47 48 without high levels of aflatoxin exposure. In these circumstances, the estimated increased risk is 49 50 51 approximately ten to twelve times. However, in populations where exposure to HBV is 52 53 combined with high levels of aflatoxin, an exposure that combines increased cell proliferation 54 55 56 with DNA reactivity, there is approximately a sixty-five fold increase in liver cancer risk, again 57 58 59 60 24 Page 25 of 59 Toxicological Sciences Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 illustrating the synergistic response that occurs when DNA reactivity is combined with an 4 5 6 increase in cell proliferation. 7 8 9 10 11 Chemoprevention 12 13 To increase cancer risk, a chemical either increases DNA reactivity leading to an increase of 14 15 mutation with each DNA replication or increases the number of DNA replications in the target 16 17 18 cell population. To reduce cancer, the reverse of this must occur. Either the amount of DNA 19 20 reactivity must be decreased and/or the amount of DNA replication needs to be reduced (Cohen, 21 22 1991). This has been the basis for chemoprevention since the initial postulate by Wattenberg 23 24 25 (1985) created the concept of chemoprevention. This concept is based on the idea that 26 27 consumption of exogenous substances could actually prevent certain types of cancer, especially 28 29 30 those caused by chemicals. Research on chemoprevention in animal models and in human 31 32 clinical trials has continued extensively ever since. However, in all of these systems, the ultimate 33 34 effects are to reduce the formation of DNA adducts, either by interfering with the metabolic 35 36 37 activation, increasing metabolic deactivation, or binding to the reactive electrophile once it is 38 39 formed so that it cannot react with the DNA, or the effect reduces cell proliferation in the target 40 41 tissue. However, like carcinogenesis itself, the issue is much more complex than originally 42 43 44 imagined. Many of the examples of chemoprevention require doses that are in excess of what can 45 46 be safely consumed by humans. Furthermore, in some examples where substances were used for 47 48 chemoprevention, side effects of the agent actually led to an increase in cancer risk, as was seen 49 50 51 in the beta-carotene trials in cigarette smokers (Hannekens et al., 1996), or led to an increase in 52 53 side effects with other disease processes, such as the cyclooxygenase-2 (COX-2) inhibitors 54 55 56 57 58 59 60 25 Toxicological Sciences Page 26 of 59 Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 which had chemopreventive activity in animal models, but in humans have led to an increased 4 5 6 risk of significant cardiovascular events (Marnett, 2009; Wu et al., 2010). 7 8 9 10 11 There is epidemiologic evidence that certain exposures can reduce cancer risk, such as increased 12 13 consumption of fruits and vegetables, although this has been challenged more recently (Vistag, 14 15 2009). Caloric restriction was identified as a preventive process in animal models nearly seven 16 17 18 decades ago (Tannenbaum, 1940; Kritchevsky, 1999), and has been demonstrated to be related to 19 20 a reduced risk of certain cancers in humans (Lutz and Schlatter, 1992). In the tissues that are 21 22 affected, caloric restriction reduces cell proliferation (Lu et al., 1993), but also has numerous 23 24 25 other effects, such as effects on DNA repair, that could contribute to the inhibitory process (Hart 26 27 et al., 1995). Although chemoprevention in the clinical setting has thus far been disappointing, 28 29 30 this area of research continues. 31 32 33 34 Immunosuppression and Carcinogenesis 35 36 37 In the late 1950's and early 1960's, several investigators made observations that suggested that 38 39 neoplastic cells developed foreign, specific antigens as they formed, and that malignancy 40 41 developed as a consequence of suppression of the normal immunosurveillance of foreign 42 43 44 antigens, allowing the malignant cell to escape surveillance and develop into a cancer (Burnet, 45 46 1957; 1964; Thomas, 1959; Klein and Klein, 1988; Kripke and Borsos, 1974; Schwartz, 1975; 47 48 Baldwin, 1973). Specific antigens were identified in mouse tumors that were related to certain 49 50 51 malignancies and were immunogenic upon transplantation (Kripke and Borsos, 1974; Klein and 52 53 Klein, 1988). Furthermore, many of the chemicals identified as carcinogens, such as polycyclic 54 55 56 aromatic hydrocarbons, were shown to be immunosuppressive in the animal models (Baldwin, 57 58 59 60 26 Page 27 of 59 Toxicological Sciences Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 1973). Lastly, Good (1968) and his colleagues at the University of Minnesota observed that 4 5 6 children born with inherited immune deficiencies had an increased risk of developing cancer, and 7 8 others began to note that kidney transplant patients also had an increased risk of cancer while 9 10 11 they were on immunosuppressive drugs (Vajdic et al., 2006). More recently, with the advent of 12 13 other immunosuppressive agents and other organ transplants, this increased risk has been 14 15 corroborated, and also observed in treatment with these agents for a variety of neoplastic and 16 17 18 non-neoplastic disorders (Grulich et al., 2007). Additionally, acquired immunodeficiency disease 19 20 syndrome (AIDS) is associated with an increased risk of developing cancers (Grulich et al., 21 22 2007). Kaposi's sarcoma was one of the original defining criteria for AIDS. 23 24 25 26 27 Although all of these pieces of evidence suggested that there was immunosurveillance of 28 29 30 neoplastic cells, in reality it turns out that the immunosurveillance is of infectious organisms, just 31 32 that some of these infections can produce cancer (Cohen et al., 1991; Cohen, 1999b; Schwartz, 33 34 1975). Thus, the tumors in immunosuppressed patients, whether inherited, secondary to 35 36 37 chemotherapy for transplantation, malignancy, or autoimmune diseases, or in AIDS, occur only 38 39 in certain tissues. These include B-cell lymphomas secondary to Epstein-Barr virus (EBV) 40 41 infections (or secondary to regeneration itself of the B-cell population), squamous cell tumors 42 43 44 secondary to human papilloma virus (HPV) or Kaposi's sarcoma secondary to human herpes 45 46 virus-8 (HHV8). There is some suggestion that there is an increased risk of liver cancer in 47 48 patients with hepatitis B or C virus infections, also. However, there is not an increased risk of the 49 50 51 more common tumors, such as cancer of the lung, colon, breast, or prostate, or other tumors 52 53 which do not have an infectious etiology. 54 55 56 57 58 59 60 27 Toxicological Sciences Page 28 of 59 Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 Furthermore, the antigenicity of the mouse tumors was due to the presence of viruses that were 4 5 6 used to induce the tumors or were activated in the tumorigenic process. Also, the doses necessary 7 8 to produce immunosuppression, such as with polycyclic aromatic hydrocarbons, were actually 9 10 11 higher than the doses necessary to produce tumors. 12 13 14 15 Based on the research over the past five decades, it is clear that the immunosurveillance is not of 16 17 18 the malignant cells themselves, but of the infections that can increase the risk of certain tumors. 19 20 21 22 Non-DNA Reactive Carcinogens 23 24 25 In animal models, hormones and various treatments that affect the endocrine system frequently 26 27 lead to an increased risk of tumors in the target population. This includes tumors of the rat 28 29 30 thyroid, the rat testicular Leydig cells, and the endocrine cells of the stomach (which lead to 31 32 carcinoid tumors). However, the only endocrine-related tumors in animal models that appear to 33 34 be pertinent to humans are those associated with estrogen (Cohen, 2004). Estrogen is known to 35 36 37 increase the risk of breast cancer and endometrial cancer in humans, and there is some 38 39 suggestion that it also increases the risk of ovarian cancer. There is some evidence that estrogens 40 41 also increases the risk of hepatocellular carcinomas (Preston-Martin et al., 1990), although more 42 43 44 commonly the effect in humans is on the induction of hepatocellular adenomas, which are not 45 46 premalignant lesions. The effect of estrogen on these target tissues is to increase the number of 47 48 DNA replications in the target tissue pluripotential stem cell population (Preston-Martin et al., 49 50 51 1990), However, there is also evidence that estrogen can be metabolically activated to 52 53 metabolites that react with DNA leading to adducts which result in apurinic sites (Cavalieri et al., 54 55 56 2002). Whether the increase in cancer risk is due only to the increase in cell proliferation or due 57 58 59 60 28 Page 29 of 59 Toxicological Sciences Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 to a synergistic interaction between the DNA reactive and cell proliferative effects remains a 4 5 6 source of intense investigation. Nevertheless, estrogenic activity must be considered a risk factor 7 8 for human cancer, at least of certain cell types. 9 10 11 12 13 As discussed above in detail, carcinogens can be classified as either DNA reactive or non-DNA 14 15 reactive, relying on increased cell proliferation. The metabolic activation of DNA reactive 16 17 18 carcinogens has been studied extensively and includes classes of compounds such as the 19 20 polycyclic aromatic hydrocarbons, aromatic amines, N-nitrosamines and related chemicals (such 21 22 as hydrazines and N-nitroso-amides), aflatoxin, and numerous others. In each of these instances, 23 24 25 exposures at high levels produce an increase in cell proliferation in the target tissue that greatly 26 27 accentuates and accelerates the process. However, at lower exposure levels the increase in 28 29 30 incidence of tumors is proportionately considerably lower, giving a strikingly non-linear dose 31 32 response, the so-called hockey stick dose response curve. Extrapolating estimate of risk in 33 34 humans for DNA reactive carcinogens has traditionally assumed a linear, non-threshold dose 35 36 37 response, based to a large degree on the radiation model of mutagenesis and carcinogenesis 38 39 (Calabrese, 2009) forming the basis for regulation of such chemicals. The shape of the dose 40 41 response for these chemicals, especially extending to low exposure levels, continues to be the 42 43 44 focus of considerable heated debate. 45 46 47 48 Based on extensive investigations over the last five decades, there have also been a large number 49 50 51 of chemicals that have been identified that increase the risk of cancer in animal models but are 52 53 not DNA reactive (Weisburger and Williams, 1981; Cohen, 2004). In each of these instances 54 55 56 studied in detail, the carcinogenic effect is due to an increase in cell proliferation. This can either 57 58 59 60 29 Toxicological Sciences Page 30 of 59 Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 be by a direct mitogenic effect (involving hormones and/or growth factors) or can be due to 4 5 6 toxicity and regeneration. Toxicity can be produced either by induction of necrosis or by an 7 8 increase in apoptosis in the target cell population. A decrease in cell deaths can occur by 9 10 11 inhibiting apoptosis or by inhibiting differentiation. A combination of these effects can certainly 12 13 occur. However, in each of these instances the increase in cell proliferation represents a 14 15 preneoplastic step in the process which is essential for the ultimate development of tumors. In 16 17 18 many instances, it is the actual toxicity that is of critical importance to human exposure whereas 19 20 the malignant consequence is either rare or the animal model of malignancy development is not 21 22 relevant to humans. Attention to the actual toxicity that is involved is more relevant to human 23 24 25 risk. A few examples that illustrate specific points are described below. 26 27 28 29 30 Sodium saccharin was shown in the 1970's to increase bladder cancer in rats when administered 31 32 beginning at birth or earlier and continuing in the offspring generation for their lifetime (Cohen 33 34 et al., 2008). The effect was greater in male rats than female rats, and mice were unaffected. 35 36 37 Administration to monkeys beginning at birth and continuing for up to twenty-four years also did 38 39 not increase cancer risk. Sodium saccharin in the rat leads to pronounced alterations in the 40 41 composition of the various normal urinary constituents, leading to the formation of calcium 42 43 44 phosphate-containing precipitate. This precipitate is cytotoxic to the urothelium which leads to 45 46 regenerative proliferation and ultimately to the development of a low incidence of tumors. The 47 48 formation of this precipitate occurs more readily in male rats than in female rats predominantly 49 50 51 because of the much higher level of protein in the male rat (2u-globulin). Mice do not develop 52 53 the precipitate because of a much lower concentration of calcium and phosphate in the urine, and 54 55 56 primates do not develop the precipitate because of the much lower density of constituents in the 57 58 59 60 30 Page 31 of 59 Toxicological Sciences Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 urine overall (osmolality of rodents is 1500-3000 mosmol, in contrast to humans below 400, and 4 5 6 commonly below 250). Thus, the mechanism involved, calcium phosphate-containing precipitate 7 8 is not relevant to humans (Cohen, 1999a; IARC, 1999b). A similar precipitate occurs with the 9 10 11 administration to rats of high doses of the sodium salt of any moderate to strong acid. Thus, 12 13 calcium phosphate-containing precipitate and bladder tumors are induced following 14 15 administration of comparably high doses of sodium ascorbate, chloride, bicarbonate, glutamate, 16 17 18 and other sodium salts (Cohen et al., 2000). 19 20 21 22 A related phenomenon has been identified in response to numerous chemicals leading to the 23 24 25 production of urinary solids. One of these, melamine, leads to urinary calculi when administered 26 27 to rats (IARC, 1999a; 1999b; Meek et al., 2003). If the dose is insufficient to produce the calculi, 28 29 30 there is no increase in proliferation or tumor formation. Thus, it is a high dose phenomenon only, 31 32 dependent on the physical property of solubility of the chemical. It was postulated that if humans 33 34 were exposed environmentally to levels substantially lower than required for calculus formation, 35 36 37 they would not have a tumorigenic response to melamine. However, in a recent unfortunate 38 39 episode in China where baby formula was specifically adulterated with extraordinarily high 40 41 levels of melamine, these children developed urinary calculi similar to what is seen in the rat, 42 43 44 with urinary obstruction, hydronephrosis, and renal injury (Meek et al., 2003; Guan et al., 2009). 45 46 Whether or not this would also lead to an increase of tumor risk can be argued, based on the 47 48 transient exposure of humans to these calculi. However, the reality is that the toxic effect, 49 50 51 calculi, was similar between the rat and the human, and actually at relatively comparable 52 53 exposure levels. These calculi were composed predominantly of melamine with variable amounts 54 55 56 57 58 59 60 31 Toxicological Sciences Page 32 of 59 Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 of uric acid also present. Thus, in this instance, the risk assessment for humans involves an 4 5 6 evaluation of the dose response and identification of a threshold. 7 8 9 10 11 High doses of d-limonene produce an increased risk of renal tubular tumors in male rats (IARC 12 13 Working Group, 1999b; Hard and Whysner, 1994). The d-limonene is metabolized to an epoxide 14 15 which binds to a protein, 2u-globulin, which is absorbed into the proximal tubular cells after 16 17 18 passing through the glomerulus. In its bound form, the 2u-globulin-d-limonene epoxide cannot 19 20 be readily degraded so that there is an accumulation in lysosomes and ultimately an increase in 21 22 cell death and consequent regeneration leading to tumors. This does not occur in female rats 23 24 25 since they do not have substantial amounts of 2u-globulin present. It also does not occur in mice, 26 27 which have an analogous protein referred to as mouse urinary protein (MUP), but the epoxide 28 29 30 does not bind to this protein and it does not lead to renal tubular cytotoxicity and regeneration. 31 32 Humans do not have a protein that is comparable to 2u-globulin to which the epoxide can bind. 33 34 Thus, humans are not at risk for developing tumors in response to this mode of action. 35 36 37 38 39 These three chemicals all act by increasing cytotoxicity and regeneration. In contrast, there are 40 41 numerous chemicals that have been identified in rats that produce a direct mitogenic effect such 42 43 44 as increasing thyroid stimulating hormone (TSH) by one means or another (IARC Working 45 46 Group, 1999b; Capen, 1998; Hill et al., 1989). This leads to a direct mitogenic stimulus of the rat 47 48 thyroid follicular cells and ultimately the development of benign and malignant tumors. 49 50 51 Although humans have a similar feedback mechanism involving circulating thyroid hormones 52 53 and TSH, the quantitative aspects are quite different. Humans have a circulating thyroid binding 54 55 56 globulin so that thyroid hormone is readily available if a stimulus occurs that leads to a decrease 57 58 59 60 32 Page 33 of 59 Toxicological Sciences Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 in circulating thyroid. In contrast, the rat does not have the circulating, readily available thyroid 4 5 6 hormone, so its response is to increase TSH to stimulate the thyroid to produce more hormone by 7 8 the follicular cells. Furthermore, the response to a hypothyroid stimulus in rats is to produce 9 10 11 TSH, stimulating proliferation of follicular cells, leading to tumors. In contrast, hypothyroidism 12 13 in humans leads to an increase in TSH, but this does not lead to an increase follicular cell 14 15 proliferation. It has been concluded that this mode of action in rats is not relevant to humans, 16 17 18 based predominantly on a quantitative assessment of the process, but also involving some 19 20 qualitative issues. Again, the tumors arise from a process that leads to cell proliferation, and this 21 22 process occurs early in the overall carcinogenicity of these chemicals. Epidemiologic 23 24 25 investigations have not shown increased thyroid cancer associated with hypothyroidism nor has 26 27 it been shown to be related to chemical exposure, only radiation. Many chemicals have been 28 29 30 shown to be toxic to the thyroid in animal models and in humans, but not thyroid carcinogens in 31 32 humans. 33 34 35 36 37 Chloroform produces liver and renal tumors in rats and mice (Meek et al., 2003; Andersen et al., 38 39 1998). The mechanism involves metabolism to phosgene which induces cytotoxicity and 40 41 ultimately regenerative proliferation in both liver and kidney. Chloroform does not produce 42 43 44 DNA reactivity, similar to all of the other chemicals that have just been described regarding 45 46 increased cell proliferation. It is thus clear that chloroform is carcinogenic only when the 47 48 exposure is sufficiently high to produce cytotoxicity. 49 50 51 52 53 In humans, chloroform was once used as an anesthetic, and based on observations at those high 54 55 56 levels of exposure it was readily apparent that chloroform could produce hepatocellular toxicity 57 58 59 60 33 Toxicological Sciences Page 34 of 59 Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 as well as nephrotoxicity in humans. However, the effect was transient, disappearing once the 4 5 6 anesthetic had been removed and the patient had the chance to recover from the toxicity. 7 8 Whether such a high exposure level would produce tumors in humans in the liver and kidney is 9 10 11 unknown, but the reality is that exposure to this chemical would not continue for a sustained 12 13 period of time at high levels because of such toxicity. Since prolonged exposure appears to be 14 15 required for carcinogenicity by these non-DNA reactive carcinogens, it is unlikely that 16 17 18 chloroform would pose a carcinogenic hazard even with short-term high-dose exposures, similar 19 20 to the melamine story with calculi in infants. However, humans are also exposed to chloroform 21 22 in chlorinated drinking water. The exposure in the drinking water is well below that which is 23 24 25 anticipated to produce hepatocellular toxicity or nephrotoxicity and thus does not pose a 26 27 carcinogenic hazard to humans. The actual toxicity assessment for chloroform is based on its 28 29 30 cytotoxicity effects on the liver and kidney, not on the carcinogenicity, which is essentially a 31 32 bystander effect resulting from the cytotoxicity. 33 34 35 36 37 There are numerous other examples that have been identified of carcinogenesis by non-DNA 38 39 reactive substances and in many instances the mode of action has been identified. These include 40 41 the cytochrome P450 inducers, which appear to involve interaction with constitutive androstane 42 43 44 receptor (CAR), which produce hepatocellular and thyroid tumors in rats and liver tumors in 45 46 mice (Ross et al., 2009). Although humans have a similar receptor, activation of this receptor 47 48 appears to lead to an increase in hepatocellular proliferation in rodents whereas in human cells 49 50 51 this proliferative effect does not occur. A similar situation occurs with PPAR agonists (Klaunig 52 53 et al., 2003; Yang et al., 2008; Gonzalez and Shah, 2008). Again, humans have a similar 54 55 56 57 58 59 60 34 Page 35 of 59 Toxicological Sciences Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 receptor, but increased cell proliferation is produced in rodents secondary to activation of the 4 5 6 receptor whereas that does not happen in humans. 7 8 9 10 11 Many non-DNA reactive carcinogens have a mode of action involving interaction with specific 12 13 cellular receptors, including nuclear receptors. This interaction with specific receptors, whether 14 15 growth factor receptors or nuclear receptors, ultimately leads to increased cell proliferation. 16 17 18 Examples include the CAR, PPAR estrogen receptors, and numerous others. It has become 19 20 apparent, however, that just identifying an interaction with a specific receptor does not define the 21 22 relevance of rodent carcinogenesis to humans. Effects of activation of the receptor on down- 23 24 25 stream consequences can vary considerably between species, as well as species differences in co- 26 27 activators and co-repressors. For example, certain chemicals interact with CAR or PPAR 28 29 30 receptors in rodents and humans, with similar metabolic consequences, such as activation of 31 32 cytochrome P450s or peroxisomal enzymes, respectively. However, in rodents there is also a 33 34 proliferative response leading to cancer induction,but this does not occur in humans and 35 36 37 consequently those chemicals are unlikely to be human carcinogens (Ross et al., 2009; Klaunig 38 39 et al., 2003; Gonzalez and Shah, 2008). 40 41 42 43 44 In each of these instances, and in many more, it is clear that the tumors arise secondary to some 45 46 early occurring toxic event. By examining the mechanisms involved in the early event, rather 47 48 than having to rely on a two year bioassay, considerable progress can be made more quickly in 49 50 51 delineating the actual mechanisms involved with possible carcinogenesis. 52 53 54 55 56 Mode of Action and Human Relevance 57 58 59 60 35 Toxicological Sciences Page 36 of 59 Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 During the past fifteen years, regulatory agencies around the world have been evolving a 4 5 6 framework to incorporate scientific understanding of the carcinogenic process into regulatory 7 8 decision making. This has led to an evolving framework, starting with a process for evaluation of 9 10 11 mode of action in animal models and then an assessment of its relevance to humans (Sonich- 12 13 Mullin et al., 2001; Meek et al., 2003; Seed et al., 2005; Boobis et al., 2006; 2008). Activities to 14 15 further develop this framework are continuing by incorporating various aspects of dose response, 16 17 18 exposure, cumulative risk, and other aspects involved in overall risk assessment. The important 19 20 part of this process is the definition of mode of action for a given chemical in contrast to what is 21 22 referred to as mechanism of action. A mode of action is a more generalized phenomenon without 23 24 25 understanding the more detailed molecular events, which is referred to as the mechanism of 26 27 action. 28 29 30 31 32 The mode of action for animal models is assessed utilizing a modification of the Bradford Hill 33 34 (1965) criteria originally developed for use in assessing causation in epidemiology 35 36 37 investigations. This includes issues such as temporality, dose response, biologic plausibility, 38 39 reproducibility of the data, and cohesiveness of the data. Once a mode of action is delineated in 40 41 the animal model, there are then qualitative and quantitative assessments of this mode of action 42 43 44 in humans. Frequently, there is no chemical specific data available to assess this in humans, so 45 46 surrogate systems need to be assessed. This can include in vitro models utilizing human cells and 47 48 in vivo models utilizing so-called humanized tissues. These, unfortunately are not definitive, but 49 50 51 provide considerable evidence that can be taken into account in an overall weight of evidence 52 53 evaluation for a chemical. However, there is frequently information in humans that is highly 54 55 56 relevant to the overall mode of action, even if chemical specific data are not available. This can 57 58 59 60 36 Page 37 of 59 Toxicological Sciences Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 take into account, for example, genetic abnormalities, such as inherited immunosuppressive 4 5 6 disorders, as well as basic biochemical and physiological processes, such as thyroid hormone 7 8 homeostasis and thyroid follicular proliferation in response to stimuli that are known to produce 9 10 11 hypothyroidism. Epidemiologic findings, not only of the chemical, but also of various 12 13 physiologic or metabolic dysfunctions like hypothyroidism, can provide valuable information. 14 15 16 17 18 Once the mode of action has been assessed, a qualitative assessment is then made based on the 19 20 question: Can human relevance of the mode of action be reasonably excluded on the basis of 21 22 fundamental, qualitative differences in key events between experimental animals and humans? A 23 24 25 concordance table comparing the effects of the key events in the animal model to the human 26 27 situation is of considerable usefulness in this exercise. If the key events of the mode of action 28 29 30 that are required for the animal model cannot occur in the human situation, such as formation of 31 32 calcium phosphate-containing precipitate with sodium salts or an epoxide metabolite of d- 33 34 limonene binding with 2u-globulin, then the mode of action is not relevant to the human 35 36 37 qualitatively and no cancer risk will occur. However, if the mode of action is relevant to humans 38 39 qualitatively, then a quantitative assessment must be made. This incorporates differences in 40 41 homeostatic control mechanisms, such as for thyroid carcinogenesis in the rat. Again, if the key 42 43 44 events in the mode of action are not quantitatively relevant to humans, then there is no cancer 45 46 risk for that chemical. After this analysis is completed, the data gaps, uncertainties, confidence in 47 48 the analysis, and implication for the overall risk assessment are considered. Even if the mode of 49 50 51 action for a given chemical is relevant to humans, the information gathered in the process of this 52 53 evaluation can then be used in an overall risk assessment, including dose response, mode of 54 55 56 action, and modulating effects on the overall response. 57 58 59 60 37 Toxicological Sciences Page 38 of 59 Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 4 5 6 Carcinogenicity Testing 7 8 As illustrated at the beginning of this article, many of the known human carcinogens have been 9 10 11 identified by astute observations of clinical and occupational settings as well as occasionally by 12 13 detailed epidemiologic evaluations. However, we would rather be able to identify which 14 15 chemicals would be likely to increase our risk of cancer before such exposure occurs and prevent 16 17 18 or reduce the exposure. Numerous methods have been developed to evaluate the potential 19 20 carcinogenic risk to humans. The standard that has been used for more than four decades has 21 22 been the two year bioassay in rats and mice. This evolved at the National Cancer Institute and 23 24 25 then ultimately at the National Toxicology Program in the United States. This has formed the 26 27 basis for screening of chemicals in the chemical, agrichemical and pharmaceutical industries as 28 29 30 well as more generally. However, this is a long, expensive, and resource intensive exercise,and is 31 32 also fraught with sources of controversy regarding relevance of the mode of action to humans or 33 34 the dose used in the study compared to human exposure. In addition, concordance of positivity in 35 36 37 rats compared to mice is only 70-75% (Haseman and Huff, 1987) with target organ concordance 38 39 even less (Maronpot et al., 2004). Because of the high background incidences of certain tumors, 40 41 such as liver, lung, vascular tumors, and lymphomas, in various mouse strains, along with other 42 43 44 factors has led many to recommend not using the mouse for carcinogenicity testing (Alden et al., 45 46 1996; Doe et al., 2006). Short term screens for chemical carcinogens have been sought to try to 47 48 replace the two year bioassay. The first successful step in this process was the identification of 49 50 51 the DNA reactivity of chemicals and the development of genotoxicity assays, particularly the 52 53 Ames assay. This is widely used for screening of chemicals in combination with structure 54 55 56 57 58 59 60 38 Page 39 of 59 Toxicological Sciences Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 activity relationship considerations. However, as discussed above, this is useful only for 4 5 6 detecting potential DNA reactive carcinogens, and there are limitations to that as well. 7 8 9 10 11 A modification of the rodent bioassay has been developed over the last two decades utilizing 12 13 transgenic and knockout mice (Cohen et al., 2001). The ones that have been specifically 14 15 evaluated for carcinogenicity screening include the p53 heterozygous knockout, the dual 16 17 18 XPA/p53 knockout mouse, the rasH2 transgenic mouse and the Tg.AC transgenic mouse for 19 20 evaluation of dermal exposure (Pritchard et al., 2002). Although some of these continue to be 21 22 utilized for screening purposes, the difficulty arises when there is a positive tumor finding as the 23 24 25 mode of action is frequently unknown. Because of interpretation difficulties and sensitivity to 26 27 irritation effects, the Tg.AC model has been essentially abandoned. The XPA/p53 model has also 28 29 30 been abandoned since the observed effects appeared to be the same as those obtained with the 31 32 p53 mouse. Of the remaining two models, the rasH2 seems to have gained more favor for general 33 34 screening whereas the p53 mouse model is used, somewhat erroneously, to assess genotoxicity. 35 36 37 A major advance in utilizing these models is the considerably shorter period of time and 38 39 decreased resources necessary to use them, but they have the same drawbacks as the two year 40 41 bioassay in that they provide no information by themselves regarding mode of action or 42 43 44 relevance to humans. This can only be gained by a shorter term assessment of specific key events 45 46 that are likely to occur in a postulated mode of action. 47 48 49 50 51 Other test systems have been developed to try to identify the chemicals acting to increase 52 53 carcinogenic risk by non-DNA reactive processes. This has included in vitro inhibition of gap 54 55 56 junction communication (Ruch and Trosko, 2001) as well as specific in vivo assays that utilize a 57 58 59 60 39 Toxicological Sciences Page 40 of 59 Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 combination of DNA reactive and proliferative stimuli, such as the Ito medium term assay (Ito et 4 5 6 al., 2003). However, all these models have significant limitations in addressing the ultimate 7 8 concern, human carcinogenic risk. 9 10 11 12 13 Based on the multi-stage model of carcinogenesis described above, and the framework for mode 14 15 of action and human relevance evaluation, another system has recently been proposed that 16 17 18 utilizes shorter-term exposure, assessing specifically the considerations for DNA reactivity, cell 19 20 proliferation, estrogenicity, and immunosuppression (Cohen, 2004; 2010). Utilizing specific 21 22 markers for various tissues, an initial screen for potential carcinogenicity in the rodent can be 23 24 25 effective, with an extremely low likelihood of false negatives. Most importantly, utilizing mode 26 27 of action analysis, short-term evaluations of detailed considerations of the mode of action and 28 29 30 dose response, one can assess the overall human relevance of the mode of action and the likely 31 32 exposure considerations for humans. A difficulty with this process is the high number of false 33 34 positives that need to be evaluated in detail regarding mode of action. However, the data that is 35 36 37 generated is directly applicable to evaluation of human risk. 38 39 40 41 With the explosion of genomics into biologic research it has become a focal point for 42 43 44 considerable investigation to utilize such systems for screening for chemical carcinogenicity and 45 46 other toxins end points (Nie et al., 2006; Ellinger-Ziegelbauer et al., 2008; Thomas et al., 2007; 47 48 Rotroff et al., 2010). Initial results are encouraging regarding DNA reactive chemicals and for 49 50 51 identifying specific modes of action for known non-DNA reactive carcinogens. It remains 52 53 unclear whether it provides a useful screening method in general, especially in detecting 54 55 56 chemicals that might act by a mode of action not yet identified. A major drawback of these 57 58 59 60 40 Page 41 of 59 Toxicological Sciences Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 approaches is the ultimate need to examine by genomic screening a very large number of tissues. 4 5 6 Most of the work done to date has involved the liver. In contrast to rodents, where liver is the 7 8 most common target site for carcinogenicity of chemicals, liver cancer in humans is generally 9 10 11 not related to chemical exposure except in unusual instances such as aflatoxin and ethanol. Based 12 13 on previous screening methodologies, it is very likely that screening of the liver will be useful in 14 15 detecting rodent liver carcinogens, but probably will not be of much use in detecting carcinogens 16 17 18 active at other target sites. It still leaves the question of human relevance. 19 20 21 22 Transgenic and knockout models have generally not been widely used for screening purposes, 23 24 25 and mode of action analysis if a positive is detected in these assays is a difficult exercise. 26 27 However, specific, targeted knockout and transgenic models can be extremely useful in 28 29 30 evaluating specific modes of action. For example, CYP2E1 is essential for the oxidative 31 32 metabolism of chloroform to phosgene (Meek et al., 2003; Andersen et al., 1998). Mice 33 34 specifically knocked out for this gene are unable to oxidatively metabolise chloroform and do not 35 36 37 develop liver or kidney toxicity, prerequisites for the ultimate development of tumors. Similarly, 38 39 in vitro models as well as humanized mouse and rat models can be of considerable usefulness in 40 41 evaluating specific modes of action for a given chemical (Yang et al., 2008). However, as 42 43 44 described above, all of these have limitations. It is critical that all of the information available be 45 46 incorporated into any assessment of the mode of action and its relevance to humans, as well as 47 48 the dose response. 49 50 51 52 53 For toxicology testing, including testing for carcinogenicity, practical considerations need to be 54 55 56 incorporated . Several suggestions have been made to begin modifying regulatory decision 57 58 59 60 41 Toxicological Sciences Page 42 of 59 Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 making so that the tens of thousands of chemicals in commerce and in the environment can be 4 5 6 addressed. For chemicals that we are exposed to at very low amounts, the toxicological 7 8 threshold of concern approach is being used in certain circumstances, such as for flavors and 9 10 11 food and drug contaminants (Munro et al., 1996). Modifications to the testing of agrichemicals 12 13 have been recommended and are being assessed (Carmichael et al., 2006; Doe et al., 2006). 14 15 Other approaches for specific classes of chemicals are being evaluated. Numerous high 16 17 18 throughput approaches have been developed as potential screening tools for the many thousands 19 20 of chemicals that need to be evaluated, such as the TOXCAST project of the EPA (Kavlok and 21 22 Dix et al., 2010). 23 24 25 26 27 A recent publication of the National Research Council, Toxicity Testing in the 21st Century: A 28 29 30 Vision and a Strategy, was published in 2007 (NRC, 2007) which describes a vision for toxicity 31 32 testing in the future. This is based on a biological systems approach to perturbations of the 33 34 complex interactive pathways controlling cellular functions and utilizing in vitro cellular model 35 36 37 systems, ultimately utilizing human cells for an analysis of potential adverse reactions in 38 39 humans. 40 41 42 43 44 Although the vision of the NRC report is laudable and worthwhile pursuing, we are far from 45 46 being able to implement it at this time. We do not have available appropriate cell lines that 47 48 appropriately mimic the response in vivo, all of the metabolic activation and inactivation 49 50 51 processes are not preserved in vitro, and current in vitro approaches are incapable of addressing 52 53 the common occurrence of organ interactions that are involved in many toxic endpoints. 54 55 56 Quantitative extrapolation from in vitro to in vivo is even more precarious without first examing 57 58 59 60 42 Page 43 of 59 Toxicological Sciences Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 toxicokinetics in vivo. In addition, we do not yet know the intricacies of the various pathways as 4 5 6 they pertain to modulating effects and distinguishing between adaptive and adverse. Fortunately, 7 8 there are many talented scientists investigating these issues and progress is being made. The 9 10 11 danger is adopting some of these screening tests before we adequately understand their relevance 12 13 and implications. As has been suggested (MacDonald and Robertson, 2009), it is important to 14 15 focus on appropriate testing based on sound science rather than accepting a test as real, just 16 17 18 another example of an Ames test approach. It is particularly easy to be seduced by the 19 20 spectacular technologies now available, losing sight of the basic biology. It is particularly 21 22 critical as we go forward to focus on the ultimate question, is it toxic to humans, animals, or the 23 24 25 environment, and if so, at what exposure level. Toxicology began with the simple premise that 26 27 all substances are poisons, but the dose makes the poison (Paracelsus, see Gallo, 2001). The 28 29 30 technologies are tools to address the science. 31 32 33 34 Conclusion 35 36 37 We have come a long way since Sir Percival Pott identified the association of scrotal cancer in 38 39 chimney sweeps, but the major questions that he raised still require continued investigation with 40 41 the development of better methods. The focus needs to be on the ultimate question of relevance 42 43 44 to human risk rather than accepting findings in model systems as sacrosanct. The issues of dose 45 46 response, linearity, threshold, and exposure, in addition to the human relevance issue remain at 47 48 the heart of our concerns. We need to remember that approximately one-fourth to one-third of 49 50 51 individuals in the United States will develop cancer some time in their lifetime. Eliminating 52 53 causes is essential for decreasing risk. However, even identifying the causes does not guarantee 54 55 56 57 58 59 60 43 Toxicological Sciences Page 44 of 59 Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 that we are able to implement preventive measures, as can be all too readily seen with cigarette 4 5 6 smoking, obesity and sexually transmitted diseases. 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 44 Page 45 of 59 Toxicological Sciences Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 4 REFERENCES 5 6 Alden, C.L., Smith, P.F., Piper, C.E., and Brej, L. (1996). A critical appraisal of the value of the 7 mouse cancer bioassay in safety assessment. Toxicol. Pathol. 24, 722-725 8 Ames BN, Durston WE, Yamasaki E, Lee FD. Carcinogens are mutagens: a simple test system 9 combining liver homogenates for activation and bacteria for detection. Proc Natl Acad Sci U S 10 11 12 A. 1973 Aug;70(8):2281-5. Andersen, M., Brusick, D., Cohen, S., Dragan, Y., Frederick, C., Goodman, J. I., Hard, G., 13 Meek, B., and O'Flaherty, E. J. (1998). U.S. Environmental Protection Agency's revised 14 cancer guidelines for carcinogen risk assessment. Toxicol. Appl. Pharmacol. 153, 133-136. 15 Armitage, P., and Doll, R. (1954). The age distribution of cancer and a multi-stage theory of 16 17 18 carcinogenesis. Br. J. Cancer. 8, 1-12. Berenblum, I., and Shubik, P. (1947). A new quantitative approach to the study of stages of 19 carcinogenesis in the mouse's skin. Br. J. Cancer 1, 383-391. 20 Boddy, A.V., and Ratain, M. J. (1997). Pharmacogenetics in cancer etiology and chemotherapy. 21 Clin. Cancer Res. 3, 1025-1030. 22 Boobis, A. R., Cohen, S. M., Dellarco, V., McGregor, D., Meek, M. E., Vickers, C., Willcocks, 23 24 25 D., and Farland, W. (2006). IPCS framework for analyzing the relevance of a cancer mode of action for humans. Crit. Rev. Toxicol. 36, 781-792. 26 Boobis, A. R., Doe, J. E., Heinrich-Hirsch, B., Meek, M. E., Munn, S., Ruchirawat, M., 27 Schlatter, J., Seed, J., and Vickers, C. (2008). IPCS framework for analyzing the relevance of a 28 noncancer mode of action for humans. Crit. Rev. Toxicol. 38, 87-96. 29 30 31 32 Boutwell, R. K. (1964). Some biological aspects of skin carcinogenesis. Progr. Exp. Tumor Res. 4, 207-250. Boveri, T. (1914). Zur Frage der Enstehung Maligner Tumoren. Jena: Verlag von Gustav 33 Fischer. 34 Bryan, G. T. (1969). Role of tryptophan metabolites in urinary bladder cancer. Am. Ind. Hyg. 35 Assoc. J. 30, 27-34. 36 37 38 Burnet, M.F. (1957). Cancer - a biological approach. IV. Practical applications. Br. Med. J. 1, 844-847. 39 Burnet, M.F. (1964). Immunologic factors in the process of carcinogenesis. Br. med. Bull. 20, 40 154-158. 41 Calabrese, E. J. (2009). The road to linearity: why linearity at low doses became the basis for 42 43 44 carcinogen risk assessment. Arch. Toxicol. 83, 203-225. Capen, C. C. (1998). Correlation of mechanistic data and histopathology in the evaluation of 45 selected toxic endpoints of the endocrine system. Toxicol. Lett. 102, 405-409. 46 Carmichael, N.G., Barton, H.A., Boobis, A.R., Cooper, R.L., Dellarco, V.L., Doerrer, N.G., 47 Fenner-Crisp, P.A., Doe, J.E., Lamb IV, J.C., and Pastoor, T.P. (2006). Agricultural chemical 48 safety assessment: A multisector approach to the modernization of human safety requirements. 49 50 51 Crit. Rev. Toxicol. 36, 1-7 Cavalieri, E. L., Li, K. M., Balu, N., Saeed, M., Devanesan, P., Higginbotham, S., Zhao, J., 52 Gross, M. L., and Rogan E. G. (2002). Catechol ortho-quinones: the electrophilic compounds 53 that form depurinating DNA adducts and could initiate cancer and other diseases. 54 Carcinogenesis. 23, 1071-1077. 55 56 57 Chapman, J. R., Webster, A. C., Kaldor, J. M., and Grulich, A. E. (2006). Cancer incidence before and after kidney transplantation. JAMA. 296, 2823-2831. 58 59 60 45 Toxicological Sciences Page 46 of 59 Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 Chuang, S. C., La Vecchia C., and Boffetta, P. (2009). Liver cancer: descriptive epidemiology 4 5 6 and risk factors other than HBV and HCV infection. Cancer Lett. 286, 9-14. Clayson, D. B., and Cooper, E. H. (1970). Cancer of the urinary tract. Advances in Cancer 7 Research. G. Klein and S. Weinhouse (eds.). New York, Academic Press, Inc. 13, 271-381. 8 Cleaver J. E., Charles W. C., McDowell M. L., Sadinski W. J., and Mitchell D. L. (1975). 9 Overexpression of the XPA repair gene increases resistance to ultraviolet radiation in human 10 11 12 cells by selective repair of DNA damage. Cancer Res. 55, 6152-6160. Cohen, S. M. (1991). Analysis of modifying factors in chemical carcinogenesis. In: N. Ito (ed.), 13 Modification of Tumor Development of Rodents, Progress in Experimental Tumor Research, 14 Vol. 33, Karger and Basel, New York, NY, 21-40. 15 Cohen, S. M. (1998a). Cell proliferation and carcinogenesis. Drug Metabolism Rev. 30, 339- 16 357. 17 18 Cohen, S. M. (1998b). Cell proliferation in the evaluation of carcinogenic risk and the 19 inadequacies of the initiation-promotion model. Int. J. Toxicol., 17, 129-142. 20 Cohen, S. M. (1999a). Calcium phosphate-containing urinary precipitate in rat urinary bladder 21 carcinogenesis. International Agency for Research on Cancer, IARC Scientific Publications. 22 147, 175-189. 23 24 25 Cohen, S. M. (1999b). Infection, cell proliferation, and malignancy. In: J. Parsonnet and S. Horning (eds.), Microbes and Malignancy: Infection as a Cause of Cancer. Oxford University 26 Press, 89-106. 27 Cohen, S. M. (2004). Human carcinogenic risk evaluation: An alternative approach to the two- 28 year rodent bioassay. Toxicol. Sci. 80, 225-229. 29 30 31 32 Cohen, S. M. (2010). Evaluation of possible carcinogenic risk to humans based on liver tumors in rodent assays: The two year bioassay is no longer necessary. Toxicol. Pathol., 38, 487-501. 33 Cohen, S. M., and Ellwein, L. B. (1990a). Cell proliferation in carcinogenesis. Science. 249, 34 1007-1011. 35 Cohen, S. M., and Ellwein, L. B. (1990b). Proliferative and genotoxic cellular effects in 2- 36 37 38 acetylaminofluorene bladder and liver carcinogenesis: Biological modeling of the ED01 study. Toxicol. Appl. Pharmacol. 104, 79-93. 39 Cohen, S. M., and Ellwein, L. B. (1991). Genetic errors, cell proliferation, and carcinogenesis. 40 Cancer Res., 51, 6493-6505. 41 Cohen, S. M., Purtilo, D. T., and Ellwein, L. B. (1991). Pivotal role of increased cell 42 43 44 proliferation in human carcinogenesis. Modern Pathol. 4, 371-382. Cohen, S. M., Arnold, L. L., Cano, M., Ito, M., Garland, E. M. and Shaw, R. A. (2000). Calcium 45 phosphate-containing precipitate and the carcinogenicity of sodium salts in rats. 46 Carcinogenesis. 21, 783-792. 47 Cohen, S. M., Robinson, D., and MacDonald, J. (2001). Alternative models for carcinogenicity 48 testing. Toxicol. Sci. 64, 14-19. 49 50 51 Cohen, S. M., Boobis, A. R., Meek, M. E., Preston, R. J., and McGregor, D. B. (2006). 4Aminobiphenyl and DNA reactivity: case study within the context of the 2006 IPCS human 52 relevance framework for analysis of a cancer mode of action for humans. Crit. Rev. Toxicol. 53 36, 803-819. 54 Cohen, S. M., Arnold, L. L., and Emerson, J. L. (2008). Safety of saccharin. Agro Food Industry 55 56 Hi-Tech., 19, 26-29. 57 58 59 60 46 Page 47 of 59 Toxicological Sciences Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 Cooper, E. H., Frank, G. L., and Wright, D. H. (1966). Cell proliferation in Burkitt tumours. Eur. 4 5 6 J. Cancer. 2, 377-84. Creech, J., Jr., and Johnson, M. (1974). Angiosarcoma of liver in the manufacture of polyvinyl 7 chloride. J. Occup. Med., 16, 150-151. 8 DeBaun, J. R., Miller, E. C., and Miller, J. A. (1970). N-Hydroxy-2-acetylaminofluorene 9 sulfotransferase: its probable role in carcinogenesis and in protein(methion-S-yl) binding in rat 10 11 12 liver. Cancer Res. 30, 577-595. Doe, J.E., Boobis, A.R., Blacker, A., Dellarco, V., Doerrer, N.G., Franklin, C., Goodman, J.I., 13 Kronenberg, J.M., Lewis, R., McConnell, E.E., Mercier, T., Moretto, A., Nolan, C., Padilla, S., 14 Phang, W., Solecki, R., Tilbury, L., Ravenzwaay, B. (2006). A tiered approach to systemic 15 toxicity testing for agricultural chemical safety assessment. Crit. Rev. Toxicol. 36, 37-69. 16 Dragan, Y. P., Sargent, L., Xu, Y. D., Xu, Y. H., and Pitot, H. C. (1993). The initiation- 17 18 promotion-progression model of rat hepatocarcinogenesis. Proc. Soc. Exp. Biol. Med. 202, 16- 19 24. 20 Ellinger-Ziegelbauer, H., Gmuender, H., Bandenburg, A., and Ahr, H. J. (2008). Prediction of a 21 carcinogenic potential of rat hepatocarcinogens using toxicogenomics analysis of short-term in 22 vivo studies. Mutat. Res. 637, 23-39. 23 24 25 Gallo, M.A. (2008). History and scope of toxicology. In:Klaassen, C.D. (ed.) Toxicology. The Basic Science of Poisons, 7th edition. McGraw Hill Medical, New York, pp. 3-10. 26 Gee, P., Maron, D. M., and Ames, B. N. (1994). Detection and classification of mutagens: a set 27 of base-specific Salmonella tester strains. Proc. Natl. Acad. of Sci. (USA). 91, 11606-11610. 28 Gonzalez, F. J., and Shah, Y. M. (2008). PPARalpha: mechanism of species differences and 29 30 31 32 hepatocarcinogenesis of peroxisome proliferators. Toxicol. 246, 2-8. Good, R. A. (1968). Experiments of nature in immunobiology. N. Engl. J. Med. 279, 1344-1345. Greenfield, R. E., Ellwein, L. B., and Cohen, S. M. (1984). A general probabilistic model of 33 carcinogenesis: Analysis of experimental urinary bladder cancer. Carcinogenesis, 5, 437-445. 34 Grulich, A. E., van Leeuwen, M. T., Falster, M. O., Vajdic, C. M. (2007). Incidence of cancers in 35 people with HIV/AIDS compared with immunosuppressed transplant recipients: a meta- 36 37 38 analysis. Lancet. 370, 59-67. Guan, N., Fan, Q., Ding, J., Zhao, Y., Lu, J., Ai, Y., Xu, G., Zhu, S., Yao, C., Jiang, L., Miao J., 39 Zhang, H., Zhao, D., Liu, X., and Yao, Y. (2009). Melamine-contaminated powdered formula 40 and urolithiasis in young children. N. Engl. J. Med. 360, 1067-74. 41 Guengerich, F. P. (2000). Metabolism of chemical carcinogens. Carcinogenesis. 21, 345-351. 42 Hahn, W. C., Counter, C. M., Lundberg, A.S., Beijersbergen, R.L., Brooks, M.W., and 43 44 Weinberg, R.A. (1999). Creation of human tumour cells with defined genetic elements. Nature. 45 400, 464-467. 46 Hard, G. C., and Whysner, J. (1994). Risk assessment of d-limonene: an example of male rat- 47 specific renal tumorigens. Crit. Rev. Toxicol. 24, 231-254. 48 Hart, R. W., Keenan K., Turturro, A., Abdo, K. M., Leakey, J., Lyn-Cook, B. (1995). Caloric 49 50 51 restriction and toxicity. Fund. Appl. Toxicol. 25, 184-195. Haseman, J.D., and Huff, J.E. (1987). Species correlation in long-term carcinogenicity studies. 52 Cancer Lett. 34, 125-132. 53 Hein, D. W. (2006). N-acetyltransferase 2 genetic polymorphism: effects of carcinogen and 54 haplotype on urinary bladder cancer risk. Oncogene. 25, 1649-1658. 55 56 57 Hennekens, C. H., Buring J. E., Manson, J. E., Stampfer, M., Rosner, B., Cook, N. R., Belanger, C., LaMotte, F., Gaziano, J. M., Ridker, P. M., Willett, W., and Peto, R. (1996). Lack of effect 58 59 60 47 Toxicological Sciences Page 48 of 59 Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 of long-term supplementation with beta carotene on the incidence of malignant neoplasms and 4 5 6 cardiovascular disease. N. Engl. J. Med. 334, 1145-1149. Hill, A. B. (1965). The environment and disease: association or causation? Proceedings of the 7 Royal Soc. of Med., 58, 295-300. 8 Hill, J. (1761). Cautions Against the Immoderate Use of Snuff. London: Baldwin, R., and 9 Jackson, J. 10 11 12 Hill, R. N., Erdreich, L. S., Paynter, O. E., Roberts, P. A., Rosenthal, S. L., and Wilkinson, C. F. (1989). Thyroid follicular cell carcinogenesis. Fundam. Appl. Toxicol. 12, 629-697. 13 Hoeijmakers, J. H. (2009). DNA damage, aging, and cancer. N. Engl. J. Med. 361, 1475-1485. 14 Hueper, W. C., Wiley, F. H., and Wolfe, H. D. (1938). Experimental production of bladder 15 tumors in dogs by administration of beta-naphthylamine. J.Ind.Hyg.Toxicol. 20, 46-84. 16 International Agency for Research on Cancer (IARC). (1999a). Saccharin and its salts. Some 17 18 Chemicals that Cause Tumours of the Kidney or Urinary Bladder in Rodents and Some Other 19 Substances. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. 73, 517- 20 624. 21 IARC Working Group. (1999b). Consensus Report. International Agency for Research on 22 Cancer, IARC Scientific Publications. 147, 1-32. 23 24 25 Ito, N., Tamano, S., and Shirai, T. (2003). A medium-term rat liver bioassay for rapid in vivo detection of carcinogenic potential of chemicals. Cancer Sci. 94, 3-8. 26 Kavlock, R. and Dix, D. (2010). Computational toxicology as implemented by the U.S. EPA: 27 Providing high throughput decision support tools for screening and assessing chemical 28 exposure, hazard and risk. J. Toxicol. Environ. Health B Crit. Rev. 13, 197-217. 29 30 31 32 Kennaway, E., and Hieger, I. (1930). Carcinogenic substances and their fluoresence spectra. Br. Med. J. 1, 1044-1046. Kinosita, R. (1936). Research on the carcinogenesis of the various chemical substances. Gann. 33 30, 423-426. 34 Kirkland, D., Pfuhler, S., Tweats, D., Aardema, M., Corvi, R., Darroudi, F., Elhajouji, A., Glatt, 35 H., Hastwell, P., Hayashi, M., Kasper, P., Kirchner, S., Lynch, A., Marzin, D., Maurici, D., 36 37 38 Meunier, J. R., Mller, L., Nohynek, G., Parry, J., Parry, E., Thybaud, V., Tice, R., van Benthem, J., Vanparys, P., and White, P. (2007). How to reduce false positive results when 39 undertaking in vitro genotoxicity testing and thus avoid unnecessary follow-up animal tests: 40 report of an ECVAM Workshop. Mutat. Res. 628, 31-55. 41 Klaunig, J. E., Babich, M. A., Baetcke, K. P., Cook, J. C., Corton, J. C., David, R. M., DeLuca, 42 J. G., Lai, D. Y., McKee, R. H., Peters, J. M., Roberts, R. A., and Fenner-Crisp, P. A. (2003). 43 44 PPARalpha agonist-induced rodent tumors: modes of action and human relevance. Crit. Rev. 45 Toxicol. 33, 655-780. 46 Knudson, A. G. (1971). Mutation and cancer: statistical study of retinoblastoma. Proc. Natl. 47 Acad. Sci. (USA). 68, 820-823. 48 Knudson, A. G. (1993). Antioncogenes and human cancer. Proc. Natl. Acad. Sci. (USA). 90, 49 50 51 10914-10921. Kodama, M., and Kodama, T. (1998). Comparative epidemiology of cancers of the testis, lung, 52 bladder and stomach with special reference to the possible implication of environmental 53 hormones in the recent risk changes of the 4 neoplasia types. Int. J. Mol. Med. 2, 705-714. 54 Kritchevsky, D. (1999). Caloric restriction and experimental carcinogenesis. Toxicol. Sci. 52(2 55 56 Suppl),13-16. 57 58 59 60 48 Page 49 of 59 Toxicological Sciences Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 Lambert, I., Singer, T., Boucher, S.E., and Douglas, G.R. (2005). Detailed review of transgenic 4 5 6 rodent mutation assays. Mutat. Res. 590, 1-280. Lower, G. M., Jr., and Bryan, G. T. (1973). Enzymatic N-acetylation of carcinogenic aromatic 7 amines by liver cytosol of species displaying different organ susceptibilities. Biochem. 8 Pharmacol. 22, 1581-1588. 9 Lu, M. H., Hinson W. G., Turturro, A., Sheldon, W. G., and Hart, R. W. (1993). Cell 10 11 12 proliferation by cell cycle analysis in young and old dietary restricted mice. Mech. Ageing Dev. 68, 151-162. 13 Lutz, W. K., and Schlatter, J. (1992). Chemical carcinogens and overnutrition in diet-related 14 cancer. Carcinogenesis. 13, 2211-2216. 15 MacDonald, J.S. and Robertson, R.T. (2009). Toxicity testing in the 21st century: A view from 16 17 18 the pharmaceutical industry. Toxicol. Sci. 110, 40-46. Maranpot, R.R., Flake, G., and Huff, J. (2004). Relevance of animal carcinogenesis findings to 19 human cancer predictions and prevention. Toxicol. Pathol. 32 (Suppl. 1), 40-48. 20 Marnett, L. J. (2009). The COXIB experience: a look in the rearview mirror. Annu. Rev. 21 Pharmacol. Toxicol. 49, 265-290. 22 McCann, J., Spingarn, N. E., Kobori, J., and Ames, B.N. (1975). Detection of carcinogens as 23 24 25 mutagens: bacterial tester strains with R factor plasmids. Proc. Natl. Acad. of Sci. (USA). 72, 979-983.Meek, M. E. (Bette), Bucher, J. R., Cohen, S. M., Dellarco, V., Hill, R. N., Lehman- 26 McKeeman, L. D., Longfellow, D. G., Pastoor, T., Seed, J., and Patton, D. E. (2003). A 27 framework for human relevance analysis of information on carcinogenic modes of action. Crit. 28 Rev. Toxicol. 33, 591-653. 29 30 31 32 Miller, J. A., and Miller, E. C. (1977). Ultimate chemical carcinogens as reactive mutagenic electrophiles. Origins of Human Cancer. H. H. Hiatt. Cold Spring Harbor, Cold Spring Harbor Laboratory. 605-627. 33 Mitchell, S. C., and Smith, R. L. (2010). A physiological role for flavin-containing 34 monooxygenase (FMO3) in humans? Xenobiotica, 40, 301-305. 35 Moolgavkar, S. H., and Knudson, A. G. J. (1981). Mutation and cancer: a model for human 36 37 38 carcinogenesis. J. Natl. Cancer Inst. 66, 1037-1052. Munro, I.C., Ford, R.A.,, Kennepohl, E., Sprenger, J.G. (1996). Correlation of structural class 39 with no-observed-effect levels: a proposal for establishing a threshold of concern. Food 40 Chem.Toxicol. 34, 829-867. 41 Murasaki, G., and Cohen, S. M. (1983). Co-carcinogenicity of sodium saccharin and N-[4-(5- 42 43 44 nitro-2-furyl)-2-thiazolyl] formamide for the urinary bladder. Carcinogenesis. 4, 97-99. National Research Council (NRC). (2007). Toxicity Testing in the 21st Century: A Vision and a 45 Strategy. National Academies Press, Washington, DC. 46 Nie, A. Y., McMillian, M., Parker, J. B., Leone, A., Bryant, S., Yieh, L., Bittner, A., Nelson, J., 47 Carmen, A., Wan, J., and Lord, P. G. (2006). Predictive toxicogenomics approaches reveal 48 underlying molecular mechanisms of nongenotoxic carcinogenicity. Mol. Carcinog. 45, 914- 49 50 51 933. Phillips, D. H., Farmer, P. B., Beland, F.A., Nath, R.G., Poirier, M.C., Reddy, M.V., and 52 Turteltaub, K.W. (2000). Methods of DNA adduct determination and their application to 53 testing compounds for genotoxicity. Environ. Mol. Mutagen. 35, 222-233. 54 Poirier, M. C., Beland F. A., Deal, F. H., and Swenberg, J. A. (1989). DNA adduct formation and 55 56 57 removal in specific liver cell populations during chronic dietary administration of 2acetylaminofluorene. Carcinogenesis. 10, 1143-1145. 58 59 60 49 Toxicological Sciences Page 50 of 59 Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 Pott, P. (1775). Chirurgical Observations Relative to the Cataract. The Polypus of the Nose, the 4 5 6 Cancer of the Scrotum, the Different Kinds of Ruptures, and the Mortification of the Toes and Feet. London: Hawes, W. Clarke, and R. Collins. Chapter III, 60-68. 7 Preston-Martin, S., Pike, M. C., Ross, R. K., Jones, P. A., and Henderson, B. E. (1990). 8 Increased cell division as a cause of human cancer. Cancer Research. 50, 7415-7421. 9 Prichard, J.B., French, J.E., Davis, B.J., and Haseman, J.K. (2003). The role of transgenic mouse 10 11 12 models in carcinogen identification. Environ. Health Perspec. 111, 444-454 Rehn, L. (1895). Blasengeschwlste be: Fuchsin-arbeitern. Arch. Klin. Chir., 50, 588-600. 13 Ross, P. K., Woods C. G., Bradford, B. U., Kosyk, O., Gatti, D. M., Cunningham, M. L., and 14 Rusyn, I. (2009). Time-course comparison of xenobiotic activators of CAR and PPARalpha in 15 mouse liver. Toxicol. Appl. Pharmacol. 235, 199-207. 16 Rotroff, D. M., Wetmore, B. A., Dix, D. J., Ferguson, S. S., Clewell, H. J. Houck, K. A., 17 18 LeCluyse, E. L., Andersen, M. E., Judson, R. S., Smith, C. M., Sochaski, M. A., Kavlock, R. 19 J.,Boellmann, F., Martin, M. T., Reif, D. M., Wambaugh, J. F., and Thomas, R. S. (2010). 20 Incorporating human dosimetry and exposure into high-throughput in vitro toxicity screening. 21 Toxicol. Sci. 117, 348-58. 22 Ruch, R. J., and Trosko, J. E. (2001). Gap-junction communication in chemical carcinogenesis. 23 24 25 Drug Metab.Rev. 33, 117-121. Seifried H. E., Seifried R. M., Clarke J. J., Junghans T. B., and San R. H. (2006). A compilation 26 of two decades of mutagenicity test results with the Ames Salmonella typhimurium and 27 L5178Y mouse lymphoma cell mutation assays. Chem. Res. Toxicol. 19, 627-644. 28 Shin, A., Shrubsole, M. J., Rice, J. M., Cai, Q., Doll, M. A., Long, J., Smalley, W. E., Shyr, Y., 29 30 31 32 Sinha, R., Ness, R. M., Hein, D. W., and Zheng, W. (2008). Meat intake, heterocyclic amine exposure, and metabolizing enzyme polymorphisms in relation to colorectal polyp risk. Cancer Epidemiol. Biomarkers Prev. 17, 320-329. 33 Soc. of Toxicology. (1981). Re-examination of the ED01 study - Overview. Fundam. Appl. 34 Toxicol. 1, 28-63. 35 Tannenbaum, A. (1940). The initiation and growth of tumors. Introduction. Effects of 36 37 38 undernutrition. Am. J. Cancer. 38, 335-350. Thomas, L. (1959). Discussion. In: Lawrence H.S. (ed.), Cellular and Humoral Aspects of 39 Hypersensitive States. New York, Hoeber-Harper, p. 529. 40 Thomas, R. S., Pluta, L., Yang, L., and Halsey, T. A. (2007). Application of genomic biomarkers 41 to predict increased lung tumor incidence in 2-year rodent cancer bioassays. Toxicol. Sci. 97, 42 55-64. 43 44 Tong, W., Welsh, W. J., Shi, L., Fang, H., and Perkins, R. (2003). Structure-activity relationship 45 approaches and applications. Environ. Toxicol. Chem. 22, 1680-1695. 46 Vajdic, C. M., McDonald, S. P., McCredie, M. R., van Leeuwen, M. T., Stewart, J. H., and Law, 47 M., Van Kaick, G., Wesch, H., Lhrs, H., and Liebermann, D. (1986). Radition-induced 48 primary liver tumors in "thorotrast patients". Rec. Results Cancer Res. 100, 16-22. 49 50 51 Vastag, B. (2009). Nutrients for prevention: negative trials send researchers back to drawing board. J. Natl. Cancer Inst. 101, 446-448, 451. 52 Watanabe, S., Brown C., and Young, J. (1986). Cumulative incidence rates for Hodgkin's disease 53 and other hematologic malignancies, with special reference to age-related carcinogenesis. Jpn. 54 J. Cancer Res. 77, 743-749. 55 56 Wattenberg, L. W. (1985). Chemoprevention of cancer. Cancer Res. 45, 1-8. 57 58 59 60 50 Page 51 of 59 Toxicological Sciences Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 Weisburger, J. H., and Williams, G. M. (1981). Carcinogen testing: current problems and new 4 5 6 approaches. Science. 214, 401-407. Whysner, J., Ross, P.M., and Williams, G. M. (1996). Phenobarbital mechanistic data and risk 7 assessment: enzyme induction, enhanced cell proliferation, and tumor promotion. Pharmacol. 8 Ther. 71, 153-191. 9 Wood, L. D., Parsons, D.W., Jones, S., Lin, J., Sjblom, T., Leary, R. J., Shen, D., Boca, S. M., 10 11 12 Barber, T., Ptak, J., Silliman, N., Szabo, S., Dezso, Z., Ustyanksky, V., Nikolskaya, T., Nikolsky, Y., Karchin, R., Wilson, P. A., Kaminker, J. S., Zhang, Z., Croshaw, R., Willis, J., 13 Dawson, D., Shipitsin, M., Willson, J. K., Sukumar, S., Polyak, K., Park, B. H., Pethiyagoda, 14 C. L., Pant, P. V., Ballinger, D. G., Sparks, A. B., Hartigan, J., Smith, D. R., Suh, E., 15 Papadopoulos, N., Buckhaults, P., Markowitz, S. D., Parmigiani, G., Kinzler, K. W., 16 Velculescu, V. E., and Whysner, J., Ross, P.M., and Williams, G. M. (1996). Phenobarbital 17 18 mechanistic data and risk assessment: enzyme induction, enhanced cell proliferation, and 19 tumor promotion. Pharmacol. Ther. 71, 153-191. 20 Wu, W. K., Sung J. J., Lee, C. W., Yu, J., and Cho, C. H. (2010). Cyclooxygenase-2 in 21 tumorigenesis of gastrointestinal cancers: an update on the molecular mechanisms. Cancer 22 Lett. 295, 7-16. 23 24 25 Yamagiwa K., and Ichikawa, K. (1915). Experimentelle studie ber die pathogenese der Epithelialgeschwlste. Mitt. Med. Fak. Kaiserl Univ. Tokio. 15, 295-344. 26 Yang, Q., Nagano, T., Shah, Y., Cheung, C., Ito, S., and Gonzalez, F. J. (2008). The PPAR 27 alpha-humanized mouse: a model to investigate species differences in liver toxicity mediated 28 by PPAR alpha. Toxicol. Sci. 101, 132-139. 29 30 31 Yoshida, T. (1933). ber die serienweise Verfolgung der Vernderungen der Leber der experimentellen Hepatomerzeugung durch o-aminoazotuol. Trans. Jpn. Path. Soc. 23, 636-638. 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 51 Toxicological Sciences Page 52 of 59 Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 Figure 1. Competing activating and inactivating processes for DNA reactive carcinogens. In 4 5 6 addition, these must occur in a pluripotential (tissue stem) cell and the cell must replicate to fix 7 8 the DNA alteration produced by a DNA adduct as a permanent mutation. 9 10 11 12 13 Figure 2. Formula derived by Armitage and Doll (1954) to describe the exponential increase in 14 15 incidences of tumors with age. I(t) = incidence at time t; N = number of normal stem cells; = 16 17 18 rate of transition between stages; n = number of stages. N and are assumed to be constant and 19 20 cell proliferation rates are also assumed to be constant. 21 22 23 24 25 Figure 3. Incidence curves for various types of tumors relative to age. The Armitage-Doll model 26 27 predicts an exponential increase of tumor incidence with age, such as seen for cancers of the 28 29 30 colon, prostate, urinary bladder, lung and others. However, Hodgkin's disease has one major 31 32 increase during the second to fourth decades of life in the United States, with another increase 33 34 later in life. Testicular germ cell tumors have peak incidences during adulthood from the third to 35 36 37 fifth decades, being rare after age 50. Childhood tumors, such as retinoblastoma, occur virtually 38 39 exclusively in childhood. The curves are drawn to represent variations in shape, and are not 40 41 drawn to scale with respect to each other. 42 43 44 45 46 Figure 4. Cellular processes of differentiation and carcinogenesis. A. Tissues under normal 47 48 49 conditions are composed of pluripotential stem cells (SN) which divide into a duplicate stem cell 50 51 and a cell committed to differentiation (CN). Committed cells may undergo cell replication 52 53 and/or undergo terminal differentiation (DN), depending on the specific tissue. Terminally 54 55 56 differentiated cells are destined to die (downward arrow) and be replaced. Stem cells and 57 58 59 60 52 Page 53 of 59 Toxicological Sciences Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 committed cells can also die (downward arrows). B. Under special circumstances, such as direct 4 5 6 mitogenesis or cell death with replacement, the stem cells divide into two stem cells rather than 7 8 one stem cell and one committed cell. C. Every time a normal stem cell (SN) replicates, a mistake 9 10 11 can occur in a critical part of a gene essential for converting the cell ultimately into a malignant 12 13 cell. The rate of a mistake occurring is indicated as the probability P1. This converts the normal 14 15 stem cell into a stem cell in an intermediate population between normal and malignancy. The 16 17 18 intermediate stem cell (SI) can undergo the same replication and differentiation processes as the 19 20 normal stem cells, forming committed (CI) and differentiated (DI) cells in the intermediate cell 21 22 population. As the intermediate stem cells replicate, they also can produce errors in the DNA, at 23 24 25 probability P2, leading to the next step in the carcinogenic process. If two steps are required, this 26 27 second error produces a malignant stem cell (SM), which also can duplicate itself or undergo 28 29 30 commitment (CM) to differentiation (DM). If more than two steps are required, there will be 31 32 additional intermediate cell populations. Chemicals can increase the risk of carcinogenesis by 33 34 affecting the probability of mistakes with each DNA replication (increasing P1 or P2), which 35 36 37 represents a DNA reactive process (DNA reactive carcinogens), or it can increase the number of 38 39 DNA replications in the normal and/or intermediate stem cell populations. Since the critical 40 41 parameter is number of DNA replications, this can be increased by increasing the number of cells 42 43 44 in these stem cell populations and/or increasing their rate of replication. (Part C is modified from 45 46 Greenfield et al., 1984, with permission.) 47 48 49 50 51 52 53 54 55 56 57 58 59 60 53 Toxicological Sciences Page 54 of 59 Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 254x190mm (300 x 300 DPI) 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 55 of 59 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Toxicological Sciences 254x190mm (300 x 300 DPI) Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 Toxicological Sciences Page 56 of 59 Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 254x190mm (300 x 300 DPI) 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 57 of 59 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Toxicological Sciences 254x190mm (300 x 300 DPI) Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 Toxicological Sciences Page 58 of 59 Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 254x190mm (300 x 300 DPI) 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 59 of 59 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Toxicological Sciences 254x190mm (300 x 300 DPI) Downloaded from toxsci.oxfordjournals.org by guest on December 14, 2010