Chapter 8: Chemical Carcinogenesis

Cancer is a disease characterized by mutation, modified gene expression, cell proliferation, and aberrant cell growth. It ranks as one of the leading causes of death in the world. In the United States, cancer ranks as the second leading cause of death, with over one million new cases of cancer diagnosed and more than 1.5 million Americans dying from cancer annually. Multiple causes of cancer have been established including infectious agents, radiation, and chemicals. Estimates suggest that 70% to 90% of all human cancers have a linkage to environmental, dietary, and behavioral factors (Fig. 8-1). Although our understanding of the biology of the progression from a normal cell to a malignant one has advanced considerably in the past several decades, many aspects of the cause, prevention, and treatment of human cancer in particular the influence of lifestyle remain unresolved.

A strong historical foundation for the linkage of the induction of cancer by chemicals has been documented (Table 8-1). Several excellent reviews of the historical background of carcinogenesis and cancer research have been published (Creech, 2000; Diamandopoulos, 1996; Shimkin, 2008). Studies over the last three centuries on chemically induced cancer are marked initially by epidemiological observations followed by experimental studies involving animal carcinogenesis models.

In 1775, Percival Pott described a linkage between the increased occurrence of scrotal and nasal cancer among chimney sweeps and their occupation. Pott concluded that chimney soot was the causative agent for cancer induction in these individuals (Pott, 1775). Other investigators also recognized an association between exposure to chemicals and the induction of human cancer. Thiersch (1875) described a relationship between sunlight exposure and skin cancer in humans. A linkage between an increased incidence of lung cancer and uranium mining was detected (Harting and Hesse, 1879). Butlin, in a follow-up of the Potts’ observations, noted that scrotal cancer in chimney sweeps in the European continent was relatively rare on the continent compared to that seen in England (Butlin, 1892). He attributed this difference to better hygiene practices by the European sweeps as well as to the use of younger boys in England, suggesting that age of exposure and the duration of exposure influence the formation of the cancer. In 1895, Rehn reported a linkage between the manufacturer of aniline dyes and the induction of bladder cancer in dye workers (Rehn, 1895). Due to the increase in demand for synthetic dyes in the 19th century, production of the aniline-based dyes increased, as did the development of skin and bladder cancer in exposed workers. The specific chemicals in the dyes related to the cause of skin and bladder cancer were later determined to be the aromatic compounds 2-napthylamine and benzidine (Hueper et al., 1938). Exposure of workers to metals, such as chromium (Alwens, 1938) and asbestos (Wood and Gloyne, 1934), was shown to be associated with an increased incidence of lung cancers in workers. Tobacco use and its linkage to lung cancer provide an excellent case study on the linkage between epidemiology and experimental carcinogenesis. As early as 1912, Adler noted an association between smokers and lung cancer (Spiro and Silvestri, 2005). Similarly, Lickint reported on the linkage between tobacco smoking and lung cancer in the 1920s and also associated the higher rate of lung cancer seen in males with tobacco smoking (Proctor, 2001). Subsequent investigations lead to the observation by Doll and colleagues, in a longitudinal epidemiology study, a direct cause and effect of smoking to lung cancer (Doll et al., 2004).

Based on these human epidemiological observations, experimental investigations starting in the first half of the 20th century were performed to understand the biological mechanisms, nature of these chemical mixtures, and individual chemical compounds in the mixtures using animal models. Initial studies by Yamagiwa and co-workers examined the linkage between exposure to coal tar and its derivatives on cancer induction in humans by painting with coal tar on rabbit ears, which resulted in the formation of skin cancer (Yamagiwa et al., 1915). Subsequently, Kennaway and Hieger purified dibenz(a,h)anthracene from this coal tar extract, and determined that dibenz(a,h)anthracene was at least one component of the coal tar mixture that was responsible for inducing skin cancer (Kennaway and Hieger, 1930). Cook and co-workers (1933) similarly extracted another polyaromatic hydrocarbon (PAH), benzo(a)pyrene, from the coal tar and showed this to also function as a carcinogen in rodents. In similar approaches with aromatic amines, Hueper and colleagues reported that 2-naphthylamine induced bladder tumors in dogs (Hueper et al., 1938) and Yoshida (1933) and Kinosita (1936) found that feeding aminoazotoulene or aminoazobenzene to rodents induced liver cancer. James and Elizabeth Miller (and their students) in the last half of the 20th century established the role of chemical structure and metabolism in the “activation” of carcinogens to a macromolecular interactive form (Miller et al., 1951; Miller and Miller, 1981; Miller et al., 1983). Studies with the aromatic amine compounds in animal models correlated with the human epidemiological studies that had revealed an association between bladder cancer in humans and exposure to aniline dyes in the occupational setting and helped to define the specific chemicals in the dyes responsible for the cancer induction. Studies by investigators such as Harris and colleagues over the last 30 years have utilized the laboratory to examine the environmentally induced cancers reported in epidemiological finding, especially in the area of lung cancer and tobacco smoking (Greenblatt et al., 1994). These epidemiological and experimental studies have shown a clear correlation between the induction of cancer in humans and rodents by chemical exposure.

An important historical event in the development of carcinogenesis testing was the development of a short-term bacterial-based mutagenesis bioassay in the early 1970s by Ames and colleagues. This Salmonella-based assay has become a stalwart in the early detection of mutagenic compounds that have the potential to be carcinogenic. A recent review of the historical aspects of the development and use of the Salmonella assay has been published (Claxton et al., 2010). The success of the Ames Salmonella Assay spurred the use of in vitro assays for mutagenicity in bacteria and mammalian cells as predictors of potential rodent and human carcinogens (Tennant et al., 1987), culminating in the current genetic toxicity test battery (Eastmond et al., 2009). As we move into the second decade of the 21st century, the onus to further develop high-throughput and shortened bioassays for carcinogen detection (cell or molecular biology based) has been identified, is ongoing (NAS, Tox 21 National Toxicology Program [NTP/NIEHS]; Tox cast [USEPA]), and represents new approaches based on our current understanding of the molecular steps in cancer-dependent pathways.

Definitions

An understanding of the cellular and molecular aspects of the cancer process requires an understanding of the scientific terms involved in defining and describing the pathology of neoplasia (Table 8-2). Neoplasia is defined as new growth or autonomous growth of tissue. A neoplastic lesion is referred to as a neoplasm. A neoplasm can be either benign or malignant. Both types of lesions are induced by chemical carcinogens. Benign neoplasms (eg, adenomas) are lesions characterized by expansive growth, frequently exhibiting slow rates of proliferation that do not invade surrounding tissue or other organs. Benign neoplasms can impair and damage the normal function of an organ through its growth impeding of blood flow. A malignant neoplasm (eg, a carcinoma) demonstrates invasive growth characteristics, capable of spreading not only through the organ of origin but via metastasis to other organs. Metastases are secondary growths derived from the cells of the primary malignant neoplasm.

The term tumor is a general term that is frequently used synonymous with a neoplastic lesion and describes a grossly visible lesion prior to a histopathological examination can confirm the exact type, grade, and stage of the neoplasm. In classifying neoplasms, the nomenclature reflects both the tissue or cell of origin, and the characteristics of the type of tissue involved (Table 8-3). For benign neoplasms, the tissue of origin is frequently followed by the suffix “oma”; for example, a benign fibrous neoplasm would be termed fibroma, and a benign glandular epithelium termed an adenoma. Malignant neoplasms from epithelial origin are called carcinomas (from with the term cancer has evolved) while those derived from mesenchymal origin are referred to as sarcoma. Thus, a malignant neoplasm of fibrous tissue would be a fibrosarcoma while that derived from bone would be an ostoesarcoma. Similarly, a malignant neoplasm from the liver would be a hepatocellular carcinoma while that derived from skin squamous epithelium is referred to as a squamous cell carcinoma. Preneoplastic lesions have also been observed in a number of target organs both animal models and humans, and reflect an early reversible lesion in neoplasm progression. The characterization and study of preneoplastic cells have led to further understanding of the process of cancer formation.

The term cancer describes a subset of neoplasia that represents malignant neoplasms. A carcinogen is an agent, chemical or physical, that causes or induces a cancer, although it is often also used to describe an agent that produces benign neoplastic lesions in rodent bioassays. In regulatory science this has been further defined as an agent whose administration to previously untreated animals leads to a statistically significant increased incidence of neoplasia of one of more histogenetic types as compared with the incidence seen in untreated control animals (Pitot, 1986). Thus, the induction of either or both benign and malignant neoplasms by a chemical or physical agent is included in the definition of a carcinogen. Carcinogens can be chemicals, viruses, hormones, radiation, or solid materials. Carcinogens either produce new neoplastic growth in a tissue or organ or increase the incidence and/or multiplicity of background spontaneous neoplastic formation in the target tissue.

The process of carcinogenesis is a multistage and multistep process involving modification and mutation to a number of normal cellular processes. An extensive literature base exists to describe the changes that cells undergo during the formation of neoplasia. Characteristics of cancers and the biological processes involved have been extensively reviewed (Hanahan and Weinberg, 2011). For regulatory classification purposes carcinogens have frequently been divided simplistically into two major categories based on their general mode of action: genotoxic and nongenotoxic. Genotoxic carcinogens are those agents that interact with DNA to damage or change its structure. They are frequently mutagenic in a dose-responsive manner, and for regulatory purposes, they are classified as exhibiting low-dose linear patterns, which approximates a straight line at very low doses without a threshold and a lack of reversibility of effect upon removal of the agent.

Nongenotoxic carcinogens are the agents that do not directly interact with nuclear DNA. Nongenotoxic carcinogens may change gene expression, modify normal cell function, bind to or modify cellular receptors, and increase cell growth. These agents work through epigenetic mechanisms and change DNA expression without modifying or directly damaging its structure. Nongenotoxic chemicals create a situation in a cell or tissue that makes it more susceptible to DNA damage from other sources (natural or xenobiotic). Nongenotoxic agents for regulatory purposes exhibit a nonlinear, threshold, and reversible dose response pattern in their carcinogenicity. Common features of genotoxic and nongenotoxic carcinogens are shown in Table 8-4.

This knowledge has led to continued efforts, using both epidemiological information and experimental animal models, to assess chemical carcinogenicity of occupational, industrial, and environmental agents, to gain an understanding of the mechanisms of action of these agents, and to determine the relevance of human exposure to cancer risk.

Extensive experimental studies with animal models and pathological evaluation of human cancers, have demonstrated that the carcinogenesis process involves a series of definable stages and steps (Shih et al., 2001; Osada and Takahashi, 2002). Operationally, three defined stages, initiation, promotion, and progression have been identified in rodent studies (initially in skin and liver tumors models) (Fig. 8-2). These steps follow a temporal sequence of events demonstrable by histopathology and observed in a wide variety of target tissues. The defining characteristics of each of these stages are outlined in Table 8-5.

Initiation

The first stage of the cancer process involves initiation, a process that is defined as a stable, heritable change. This stage is a rapid, irreversible process that results in a carcinogen-induced mutational event. Chemical and physical agents that function at this stage are referred to as initiators or initiating agents. Initiating agents lead to genetic changes including mutations and deletions. Chemical carcinogens that covalently bind to DNA and form adducts that result in mutations are initiating agents. Included among chemicals classified as initiating carcinogens are compounds such as polycyclic hydrocarbons and nitrosamines, biological agents, certain viruses, and physical agents such as x-rays and ultraviolet (UV) light. Most chemical carcinogens that function at the initiation stage of the cancer process are indirect-acting genotoxic compounds that require metabolic activation in the target cell to produce the DNA-damaging event. For these compounds, the chemical must be taken into the target site and metabolized (in the case of an indirect genotoxic carcinogen). The ultimate form of the carcinogen is then able to bind to nuclear DNA, resulting in adduct formation. The initiating event becomes “fixed” when the DNA adducts or other damage to DNA is not correctly repaired or is incompletely repaired prior to DNA synthesis. This event can lead to inappropriate base pairing and formation of a mutation.

Initiation by itself does not appear to be sufficient for neoplastic formation. Once initiated cells are formed, their fate has multiple potential outcomes: (1) the initiated cell can remain in a static nondividing state through influences by growth control either via normal surrounding cells or through endocrine influence; (2) the initiated cell may possess mutations incompatible with viability or normal function and be deleted through apoptotic mechanisms; or (3) the cell, through stimuli such as intrinsic factors or from chemical exposure, may undergo cell division resulting in the growth in the proliferation of the initiated cell. Besides the production of an initiated cell through carcinogen binding and misrepair, additional evidence has come forth showing that induction of continual stress, resulting in continual cell proliferation, can also produce new mutated, initiated cells (Matthews et al., 2009; Johnstone and Baylin, 2010) Therefore, modification of cell signaling and gene expression (while usually ascribed to the tumor promotion process and as an epigenic pathway [see below]) may indirectly through sustained cell proliferation also result in the formation of mutations (Lewandowska and Bartoszek, 2011). In some instances, typically following relatively high doses and/or repeated exposure to the genotoxic carcinogen, a chemical carcinogen may function as a complete carcinogen, that is, it is capable of progressing through all stages of the cancer process.

Promotion

The second stage of the carcinogenesis process, derived from either endogenous or exogenous stimuli of cell growth, involves the selective clonal expansion of initiated cells to produce a preneoplastic lesion. This is referred to as the promotion stage of the carcinogenesis process. Both exogenous and endogenous agents that function at this stage are referred to as tumor promoters. Tumor promoters are not mutagenic and generally are not able to induce tumors by themselves; rather they act though several mechanisms involving gene expression changes that result in sustained cell proliferation either through increases in cell proliferation and/or the inhibition of apoptosis. This stage involves the modulation of gene expression through receptor or non–receptor-mediated processes. Nongenotoxic carcinogens frequently function at the tumor promotion stage. The growth of preneoplastic lesions requires repeated applications or continuous exposure to tumor-promoting compounds. Although initial exposure to tumor promoters may result in an increase in cell proliferation and/or DNA synthesis in all tissues of the organ, this is usually a transient effect on the normal cells and with repeated applications of the tumor promoter only the initiated cells continue to clonally expand and divide (Fig. 8-2). Promotion is a reversible phenomenon whereby upon removal of the promotional agent, the focal cells may return back to the initiated cell. In addition, these agents demonstrate a well-documented threshold for their effects—below a certain dose or frequency of application; tumor promoters are unable to induce cell proliferation. Multiple chemical compounds as well as physical agents have been linked to the tumor promotion stage of the cancer process. Tumor promoters in general show organ-specific effects, for example, a tumor promoter of the liver, such as phenobarbital, will not function as a tumor promoter in the skin or other tissues.

Progression

The final stage of the carcinogenesis process, progression, involves the conversion of the benign preneoplastic lesions into a neoplastic cancer. In this stage, due to increase in DNA synthesis in cell proliferation in the preneoplastic lesions, additional genotoxic events may occur resulting in additional DNA damage including chromosomal damage such as aberrations and translocations. These events result in the transfer from preneoplastic, clonally derived cell populations into neoplastic cell populations. Cells accumulate a variety of mutations and epigenetic changes that cause them to outgrow the surrounding cells and to attract all of the necessary nutrients via angiogenesis. The tumor microenvironment has been also shown to be an important component of this process and the presence of “normal” cells and stroma within the lesion is critical for the neoplastic cells to survive and propagate. During the progression phase of the process the clonal nature of the neoplastic lesion is typically lost with a polyclonal appearance of cells within the lesion. Agents that impact on the progression stage are usually genotoxic agents although the acquisition of additional mutagenic events may occur through the continual stimulation of cell proliferation. By definition, the progression stage is an irreversible stage in that neoplasm formation, whether benign or malignant, occurs. With the formation of neoplasia, an autonomous growth and/or lack of growth control is achieved. Spontaneous progression can occur from spontaneous karyotypic changes that occur in mitotically active initiated cells during promotion. An accumulation of nonrandom chromosomal aberrations and karyotypic instability are hallmarks of progression. As such, chemicals that function as progressor agents are many times clastogenic and are capable of causing chromosomal abnormalities. Complete carcinogens have the ability to produce initiation, promotion and progression and hence by definition have genotoxic properties. In addition, a number of model systems are available to test for carcinogenicity and/or to study the multistep mechanisms involved in chemical carcinogenesis. The models are described in a later section of this chapter.

The formation of a neoplasm is a multistage, multistep process that involves the ultimate release of the neoplastic cells from normal growth control processes and creating a tumor microenvironment. In a highly regarded review, Hanahan and Weinberg (2000) initially defined the major characteristics of a neoplasm as it progresses into a malignant state. This review has recently been updated (Hanahan and Weinberg, 2011). These authors have proposed eight “hallmarks of cancer” that involve modifications to the homeostasis of normal cells imparting the growth and metastatic properties seen in the development of a neoplastic cell. These eight properties include (1) sustaining cell proliferation, (2) resisting cell death (apoptosis), (3) inducing angiogenesis, (4) enabling replicative immortality, (5) activating invasion and metastasis, (6) evading growth suppressors, (7) reprogramming of energy metabolism, and (8) evading immune destruction (Table 8-6). An important concept from this review is the fact that tumors are not just a collection of clonal neoplastic cells but a complex tissue with multiple cell populations and interact with one another and function as a unique tissue. This tumor microenvironment involves the recruitment of normal stromal and inflammatory cells that contribute to the growth the development of the neoplasm.

Although our understanding of the cancer process has expanded exponentially in the past several decades as is illustrated by the “hallmarks-of-cancer” concept, in its minimalist form two major processes are needed with regard to the induction of neoplasia by chemical agents: a mutational event and a selective proliferation of the mutated cell to form a neoplasm. Based on this supposition, chemicals that induce cancer have been classified into one of two broad categories (as noted above)—genotoxic or DNA-reactive agents, and nongenotoxic or epigenetic agents. This is admittedly a very simplistic approach to the complexities of the cancer process (as illustrated by the Hanahan and Weinberg’s review [2011]); however, for the purposes of regulatory processes and risk assessment this categorization has for the most part continued.

Genotoxic/DNA-Reactive Compounds

Genotoxic compounds interact with nuclear DNA of a target cell producing unrepaired DNA damage that is inherited in subsequent daughter cells. Experimental and epidemiological observations made in the middle of the 20th century identified a number of chemicals that could cause cancer in humans or experimental animals. Coal tar carcinogens including benzo(a)pyrene, pesticides such as 2-acetylaminofluorine and azo dyes such as diaminobenzamide were among the first chemical carcinogens to be studied. DNA-reactive carcinogens can be further subdivided according to whether they are active in their parent form (ie, direct-acting: agents that can directly bind to DNA without being metabolized) and those that require metabolic activation (ie, indirect-acting carcinogens: compounds that require metabolism in order to react with DNA). Examples of direct-acting and indirect-acting carcinogens are listed in Table 8-7.

Direct-Acting (Activation-Independent) Carcinogens

A class of carcinogens that do not require metabolic activation or modification to induce cancer, are termed direct-acting or activation- independent carcinogens. These chemicals are also defined as ultimate carcinogens. Examination of the chemical structure of these chemicals reveals that they are highly reactive electrophilic molecules that can interact with and bind to nucleophiles such as cellular macromolecules including DNA. Some common electrophilic species in these groups are shown in Fig. 8-3. Generally these chemicals containing these moieties frequently cause tumor formation at the site of chemical exposure.

Direct-acting carcinogens include epoxides, imines, alkyl and sulfate esters, and mustard gasses (Fox and Scott, 1980; Sontag, 1981). Direct-acting electrophilic carcinogenic chemicals typically test positive in the Ames test without additional bioactivation with a liver metabolic fraction. The relative carcinogenic strength of direct-acting carcinogens depends in part on the relative rates of interaction between the chemical and genomic DNA, as well as competing reactions with the chemical and other cellular nucleophiles. The relative carcinogenic activity of direct-acting carcinogens is dependent upon such competing reactions and also on detoxification reactions. Chemical stability, transport, and membrane permeability determine the carcinogenic activity of the chemical. Direct-acting carcinogens are typically carcinogenic at multiple sites and in all species examined. A number of direct-acting alkylating agents, including a number of chemotherapeutic chemicals, are carcinogenic in humans (Marselos and Vainio, 1991).

Indirect-Acting Genotoxic Carcinogens

An important discovery in the understanding of chemical carcinogenesis came from the investigations of the Millers who established that many carcinogens are not intrinsically carcinogenic, but require metabolic activation to be carcinogenic. They demonstrated that azo dyes covalently bind to proteins in liver, leading to the conclusion that carcinogens may bind to proteins that are critical for cell growth control (Miller and Miller, 1947). Work with benzo(a)pyrene showed covalent binding of benzo(a)pyrene or the metabolites of benzo(a)pyrene in rodents (Miller et al., 1951). Subsequent investigations with other indirect-acting genotoxic carcinogens by other investigators confirmed that metabolism of the parent compound was necessary to produce a metabolite (activation) that was able to interact with DNA (Miller, 1970).

It has since been shown that the majority of DNA-reactive carcinogens are found as parent compounds, or procarcinogens. Procarcinogens are stable chemicals that require subsequent metabolism to be carcinogenic (Conney, 1982; Miller and Miller, 1981; Miller et al., 1983; Weisburger and Williams, 1981 in Becker, 1982). The terms procarcinogen, proximate carcinogen, and ultimate carcinogen have been coined to define the parent compound (procarcinogen) and its metabolite form, either intermediate (proximate carcinogen) or final (ultimate carcinogen) that reacts with DNA (Fig. 8-4). The ultimate form of the carcinogen is most likely the chemical species that results in mutation and neoplastic transformation. The ultimate form of certain carcinogenic chemicals is not known, whereas for other chemicals there may be more than one ultimate carcinogenic metabolite depending on the metabolic pathway followed. It is important to note that besides activation of the procarcinogen to a DNA-reactive form, detoxification pathways may also occur inactivating the carcinogen.

Indirect-acting genotoxic carcinogens usually produce their neoplastic effects, not at the site of exposure (as seen with direct-acting genotoxic carcinogens) but at the target tissue where the metabolic activation of the chemical occurs. Indirect-acting genotoxic carcinogens include the polycyclic PAHs, nitrosamines, aromatic amines, and aflatoxin B1. Fig. 8-4 shows the parent (procarcinogen) and metabolites for several representative indirect-acting genotoxic carcinogens.

Mutagenesis

The reaction of a carcinogen with genomic DNA either directly or indirectly may result in DNA adduct formation or DNA damage, and frequently produces a mutation. Several mechanisms of mutagenesis are known to occur. Modification of DNA by electrophilic carcinogens can lead to a number of products. These events are dependent upon when in the cell cycle the adducts are formed, where the adducts are formed, and, the type of repair process used in response to the damage.

Transitions are a substitution of one pyrimidine by the other or one purine by the other (changes within a chemical class), whereas a transversion occurs when a purine is replaced by a pyrimidine or a pyrimidine is replaced by a purine (changes across a chemical class). Carcinogens can induce transitions and transversions several ways. In one scenario, when adducts (or apurinic or apyrmidinic sites) are encountered by the DNA replication processes, they may be misread. The polymerase may preferentially insert an adenine (A) in response to a noninformative site. Thus, the daughter strand of an A, C, G, or T adduct will have an adenine (A) and this change is fixed (mutation) and resistant to subsequent DNA repair. A second outcome, a shift in the reading frame (resulting in a frame shift mutation) may also result from carcinogen–DNA adducts formation. Most frame shift mutations are deletions and occur more frequently when the carcinogen–DNA adduct is formed on a nucleotide. In a third scenario, DNA strand breaks can also result from carcinogen–DNA adducts. These may arise either as a result of excision repair mechanisms that are incomplete during DNA replication or via direct alkylation of the phosphodiester backbone leading to backbone cleavage. Strand scission can lead to double-strand breaks, recombination, or loss of heterozyogosity.

Damage by Alkylating Electrophiles

As noted above, most chemical carcinogens require metabolic activation to exert a carcinogenic effect. The ultimate carcinogenic forms of these chemicals are frequently strong electrophiles (Fig. 8-3) that can readily form covalent adducts with nucleophilic targets. Because of their unpaired electrons, S:, O:, and N: atoms are nucleophilic targets of many electrophiles. The extent of adducts formed is limited by the structure of DNA where bulky electrophilic chemicals can bind, and size of the ultimate carcinogenic form. In general, the stronger electrophiles display a greater range of nucleophilic targets (ie, they can attack weak and strong nucleophiles), whereas weak electrophiles are only capable of alkylating strong nucleophiles (eg, S: atoms in amino acids). In addition, the metabolic capability of a target cell will dictate the extent and types of electrophiles generated from the procarcinogenic parent.

An important and abundant source of nucleophiles is contained not only in the DNA bases, but also in the phosphodiester backbone (Fig. 8-5). Although carcinogen–DNA adducts may be formed at all sites in DNA, the most common sites of alkylation include the N⁷ of guanine, the N³ of adenine, the N¹ of adenine, the N³ of guanine, and the O⁶ of guanine. Alkylations of phosphate may also occur at a high frequency. Selective examples of carcinogen interactions with proteins and nucleic acids are shown in Fig. 8-6. Different electrophilic carcinogens will often display different preferences for nucleophilic sites in DNA and different spectra of damage. Dimethylnitrosamine and diethylnitrosamine, for example, are metabolized by P450 oxidation to yield a methyl carbonium ion (CH₃⁺) and an ethyl carbonium ion (CH₃CH₃⁺), respectively. Despite the structural similarities of the ultimate electrophiles, they display significant differences in alkylation profiles (Pegg, 1984). The relative proportions of methylated bases present in DNA following reaction with carcinogen-methylating agents are shown in Table 8-8 (Pegg, 1984). The predominant adduct formed following exposure to methylating agents such as methylmethane sulfonate is 7-methylguanine. In contrast, ethylating agents produce adducts predominately in the phosphate backbone of DNA. The carcinogenic potential of the type of adducts formed is often debated; some believe that O⁶-alkylguanine is the most carcinogenic adduct, whereas others report that O⁴-alkylthymine is more important in the carcinogenic process due to its persistence relative to other adducts (Pegg, 1984; Swenberg et al., 1984).

Another common modification to DNA is the hydroxylation of DNA bases. Oxidative DNA adducts have been identified in all four DNA bases (Fig. 8-7); however, 8-hydroxyguanine is among the most prevalent oxidative DNA adduct (Floyd, 1990). The source of oxidative DNA damage is typically formed from free radical reactions that occur endogenously in the cell or from exogenous sources (Klaunig and Kamendulis, 2004; Floyd, 1990; Ames and Shigenaga, 1993). Although a relatively large amount of oxidative DNA adducts have been proposed to be formed per day, repair mechanisms exist that maintain the cellular level at a low rate and keep endogenous mutations to a relatively low level. The role of oxidative damage and oxidative stress is discussed in greater detail later in this chapter.

Methylation of deoxycytidine residues is a well-studied DNA adduct. This reaction occurs by the transfer of a methyl group from S-adenosylmethionine by DNA methyltransferases (Holliday, 1989). Methylation of DNA results in heritable expression or repression of genes, with hypomethylation associated with active transcription of genes, whereas hypermethylated genes tend to be rarely transcribed. Chemical carcinogens may inhibit DNA methylation by several mechanisms including forming covalent adducts, single-strand breaks in the DNA, alteration of methionine pools, and inactivation of the DNA methyltransferase responsible for methylation (Riggs and Jones, 1983). The importance of DNA methylation in chemical carcinogenesis is discussed further later in this chapter.

Although a large number of adducts can be formed following exposure to chemicals, whether a particular DNA adduct will result in mutation and participate in the carcinogenesis process is dependent in part on the persistence of the adduct through the process of DNA replication, which is also in part dependent upon DNA repair.

DNA Repair

Following the formation of a carcinogen–DNA adduct, the persistence of the adduct is a major determinant of the outcome. This persistence depends on the ability of the cell to repair the altered DNA. The detection of unique DNA adducts has proven to be important in understanding the mechanism of action of specific carcinogens and has been correlated with carcinogenesis (Goth and Rajewsky, 1974; Swenberg et al., 1985; Kadlubar et al., 1981; Becker and Shank, 1985). However, the presence of a DNA adduct is not sufficient for the carcinogenesis process to proceed. The relative rates or persistence of particular DNA adducts may be an important determinant of carcinogenicity, for example, O⁴-ethylthymine is relatively stable in DNA whereas O⁶-ethylguanine does not persist in DNA after continuous exposure to diethylnitrosamine (Swenberg et al., 1984). The persistence of DNA adducts of trans-4-aminostilbene does not correlate with organ carcinogenicity and/or tissue susceptibility. Although the liver and kidney exhibited the greatest burden and persistence of the adduct and the Zymbals gland showed the least amount of DNA adducts, the latter tissue was more susceptible to carcinogenesis by this chemical (Neumann, 1983). As such, differences in susceptibility to carcinogenesis are likely the result of a number of factors, including DNA replication within a tissue and repair of a DNA adduct.

The quantitation of covalent DNA adducts in tissues has been used to demonstrate exposure of humans to carcinogenic chemicals and to assess the relative risk from exposure to carcinogenic chemicals. For example, DNA adducts of carcinogenic PAHs have been demonstrated at relatively high levels in tissues and blood of smokers and foundry workers compared with nonexposed individuals (Perera et al., 1991). In addition, DNA adducts of aflatoxin B1 were seen in samples of human placenta and cord blood in individuals in Taiwan, an area that demonstrates a high incidence of liver cancer (Hsieh and Hsieh, 1993). The presence of macromolecular carcinogen adducts may be important to their mechanism of carcinogenicity, the presence and persistence of the adducts is only one factor in the process of cancer development.

Experimental and epidemiological evidence indicates that the development of cancer following exposure to chemical carcinogens is a relatively rare event. This can be explained by the ability of a cell to recognize and repair DNA. The DNA region containing the adduct is removed and a new patch of DNA is synthesized, using the opposite intact strand as a template. The new DNA segment is then spliced into the DNA molecule in place of the defective one. To be effective in restoring a cell to normal, repair of DNA must occur prior to cell division; if repair is not complete prior to replication, the presence of the adducts can give rise to mispairing of bases and other genetic effects such as rearrangements and translocations of DNA segments. Thus, a chemical that alters the repair process or the rate of cell division can itself affect the frequency of mutation and neoplastic transformation.

DNA Repair Mechanisms

Although cells posses mechanisms to repair many types of DNA damage, these are not always completely effective, and residual DNA damage can lead to the insertion of an incorrect base during DNA replication, followed by transcription and translation of the mutated templates, ultimately leading to the synthesis of altered protein. Mutations in an oncogene, tumor-suppressor gene or gene that controls the cell cycle, can result in a clonal cell population with a proliferative or survival advantage. The development of a tumor requires many such events, occurring over a long period of time, and for this reason human cancer induction often takes place within the context of chronic exposure to chemical carcinogens.

A number of structural alterations may occur in DNA as a result of interaction with reactive chemicals or radiation, and the more frequent types of damage are depicted in Fig. 8-8. The reaction of chemical species with DNA can produce adducts within the bases, sugar and phosphate backbone of DNA. In addition, bifunctional alkylating agents (such as mustards) may cause DNA cross-linking between two opposing bases. Other structural changes such as pyrimidine dimmer formation are specific for exposure to UV light whereas double-stranded breaks in DNA are more commonly associated with ionizing radiation. A variety of mechanisms have evolved to effectively remediate and repair DNA damage. The more common types of DNA repair mechanisms seen in mammalians are listed in Table 8-9.

In addition to the proofreading activity of DNA polymerases that can correct miscopied bases during replication, cells have several mechanisms for repairing DNA damage. Repair of DNA damage does not always occur prior to cell replication, and, in addition, repair of DNA damage by some chemicals is relatively inefficient, as such, exposure to chemicals that cause DNA damage can increase the probability of acquiring mutations that ultimately lead to cancer development.

Mismatch Repair of Single-Base Mispairs

Spontaneous mutations may occur through normal cellular DNA replication mistakes. Many spontaneous mutations are point mutations, a change in a single base pair in the DNA sequence, or a small insertion or deletion or a frame shift mutation of some modest size. The issue for mismatch repair is determining which is the normal DNA and which is the damaged DNA strand, and, therefore, repairing the mutated strand such that the correct base pairs are restored. Depurination is a fairly common occurrence and spontaneous event in mammals, and results in the formation of apurinic sites. If these lesions are left unrepaired, mutations are generated during DNA replication as DNA synthetic machinery is unable to determine the appropriate base with which to pair. All mammalian cells possess apurinic endonucleases that function to cut DNA near apurinic sites. The cut is then extended by exonucleases, and the resulting gap repaired by DNA polymerase and ligase.

Excision Repair

DNA regions containing chemically modified bases, or DNA chemical adducts, are typically repaired by excision repair processes. DNA adducts cause a distortion in the normal shape of DNA. Proteins that slide along the surface of a double-stranded DNA molecule recognize the irregularities in the shape of the double helix, and affect the repair of the lesion. DNA lesions that are repaired by excision repair processes include thymine–thymine dimmers, produced following exposure to UV light, dimers that interfere with both replication and transcription of DNA. The repair of DNA regions containing bases altered by the attachment of large chemical adducts such as benzo(a)pyrene are also effectively repaired by excision repair processes (Fig. 8-9).

End-Joining Repair of Nonhomologous DNA

A cell that has double-strand breaks can be repaired by joining the free DNA ends. The joining of broken ends from different chromosomes, however, will lead to the translocation of DNA pieces from one chromosome to another, translocations that have the potential to enable abnormal cell growth by placing a proto-oncogene next to, and, therefore, under the control of another gene promoter. Double-strand breaks can be caused by ionizing radiation and drugs such as anticancer drugs. Double-strand breaks are correctly repaired only when the free ends of DNA rejoin exactly. The repair of double-stranded DNA is therefore confounded by the absence of single-stranded regions that can signal the correct base pairing during the rejoining process. Homologous recombination is one of two mechanisms responsible for the repair of double-strand breaks. In this process, the double-strand break on one chromosome is repaired using the information on the homologous, intact chromosome.

The predominant mechanism for double-stranded DNA repair in multicellular organisms is nonhomologous repair and involves the rejoining the ends of the two DNA molecules. Although this process yields a continuous double-stranded molecule, several base pairs are lost at the joining point. This type of deletion may produce a possibly mutagenic coding change.

Classes of Genotoxic Carcinogens

Polyaromatic Hydrocarbons

PAHs are found at high levels in charcoal-broiled foods, cigarette smoke, and in diesel exhaust. Representative chemicals belonging to this class are shown in Fig. 8-10A. A recent review of the causes of human cancer (103 human carcinogens) also lists those PAHs linked to human cancer by the International Agency for Research in Cancer (IARC monographs [Vol. 100 A–F]). Benzo(a)pyrene is a representative polycyclic hydrocarbon that has been extensively studied. The metabolism and pathways that lead to tumor formation have been characterized through the work of a number of laboratories (Conney, 1982). The ultimate carcinogen is a diol epoxide of benzo(a)pyrene, formed following three separate enzymatic reactions (Sims et al., 1974). Benzo[a]pyrene is first oxidized by cytochrome P4501A1 to form benzo[a]pyrene 7,8-oxide, further metabolized by epoxide hydrolase, yielding the 7,8-dihydrodiol. And then further metabolism by cytochrome P4501A1 to yield the ultimate carcinogen, the 7,8-dihydrodiol-9,10-epoxide (Yang et al., 1976; Lowe and Silverman, 1984). The Bay region or K region of the benzo(a)pyrene molecule is the site of metabolic targeting and DNA interaction (Fig. 8-10B). The importance of the Bay region (K region) of the benzo(a)pyrene molecule was demonstrated from the understanding of the metabolism of the benzo(a)pyrene to its ultimate DNA-reactive form. Similar regions (to the Bay region) have been identified in other carcinogenic polycyclic hydrocarbons and used to access and predict carcinogenic potential of other PAHs. Evidence has emerged that besides the P450 metabolism pathway, PAHs can also be metabolized through a peroxidase pathway and the aldo–keto reductase (AKR) pathway. The latter of which has been attributed to cancer risk in humans exposed to coal burning smoke in China (Lan et al., 2004). Oxidative damage-related genes AKR1C3 and OGG2 also appear to modulate risks for lung cancer due to exposure to PAH-rich coal combustion emissions.

Alkylating Agents

Alkylating agents represent an important class of chemical carcinogens. Although some aklylating agents are direct-acting genotoxic agents, many require metabolic activation to produce electrophilic metabolites that can react with DNA. Alkylating agents can be classified into several groups including the direct-acting alkylalkanesulfonates (methyl- and ethyl methanesulfonate) and nitrosamides (N-methyl-N-nitrosourea, N-ethyl-N-nitrosourea, N-methyl-N-nitro-N-nitrosoguanidine, and the indirect-acting nitrosamides (dimethyl- and diethylnitrosamines) (Fig. 8-11). The alkylating compounds (or in the case of diethylnitrosamine and dimethylnitrosamine and their metabolites) readily react with DNA at more than 12 sites. The N⁷ position of guanine and the N³ position of adenine are the most reactive sites for alkylating chemicals to find to DNA. DNA methylation reactions occur more readily and thus exhibit >20 more adduct formation compared with ethylation reactions. However, ethylation reactions have a greater affinity for oxygen centers, an event that appears to correlate with the mutagenicity and carcinogenicity of these compounds. Nitrosamines were initially used as solvents in chemistry. Their toxic effects were identified when workers using the nitrosamines solvents developed jaundice and liver damage. Subsequent studies in animal models revealed that dimethylnitrosamine, and eventually diethylnitrosamine were highly hepatotoxic and hepatocarcinogenic.

Other alkylating chemicals including the nitrogen mustards (eg, chlorambucil, cyclophosphamide) have been used in cancer chemotherapy. They produce DNA adducts as well as induce the formation of DNA strand breaks. The alkylation of DNA by nitrogen mustards requires the formation of highly reactive N-alkylazirdinium ions (Fig. 8-12). Nitrogen mustards can produce a wide spectrum of mutations including base pair substitutions (AT and GC) and deletions. In addition, nitrogen mustards are potent clastogens causing chromosomal aberrations and sister chromatid exchanges (SCEs), predominantly in GC-rich regions.

Ethylene oxide and propylene oxide are other examples of mutatgenic and carcinogenic alkylating agents (Fig. 8-13). Ethylene oxide is a direct-acting alkylating carcinogen in rodents and perhaps of human concern (Hogstedt et al., 1986). Ethylene oxide is mutagenic in short-term in vitro assays and produces chromosomal aberrations and SCEs in eukaryotic cells (Ehrenberg and Hussain, 1981). Propylene oxide is also a mutagenic rodent carcinogen inducing nasal tumors in rodents following inhalation exposure (NTP, 1985). Alkylation of DNA by ethylene and propylene oxide occurs predominantly at the N⁷ position of guanine, yielding 7-(2-hydroxyethyl)guanine and 7-(2-hydroxypropyl)guanine adducts, respectively. These adducts represent the major adducts formed following either in vitro or in vivo exposure (Walker et al., 1992).

Vinyl chloride is another known rodent and human carcinogen, producing angiosarcomas in the liver and tumors in the lung and hematopoetic system in humans (Doll, 1988). Limited evidence also suggests that vinyl chloride exposure results in brain tumors. Vinyl chloride is mutagenic and is metabolized by cytochrome P450 to form chloroethylene oxide, which rearranges nonenzymatically to produce chloroacetaldehyde, both of which can alkylate DNA. Vinyl chloride and metabolites form several DNA adducts including 7-(2′-oxoethyl)guanine, N², 3-etheneoguanine, 3,N⁴-ethenocytosine, and 1,N⁶-ethenoadenine (Guengerich, 1994).

Aromatic Amines and Amides

Aromatic amines and amides encompass a class of chemicals with varied structures (aromatic amines, eg, aniline dyes, 2-natphylamine, benzidine, 2-acetylaminofluofene) (Fig. 8-14). Because of their use in the dye industry and other industrial processes their carcinogen risk to humans was realized in the late 19th century. Exposure to these chemicals still occurs through cigarette smoke and environmental sources, even though proper industrial hygiene processes have considerably reduced the human exposure to aromatic amines and amides in the workplace. The aromatic amines undergo both phase I and phase II metabolism. Phase I reactions occur mainly by cytochrome P450-mediated reactions, yielding hydroxylated metabolites that are often associated with adduct formation in proteins and DNA, and produce liver and bladder carcinogenicity (Miller et al., 1964). Metabolism of 2-acetylaminofluorene (AAF) results in the formation of N-hydroxy-AAF, which is a metabolite responsible for the liver tumorigenicity. Similarly, 1-napthylamine exhibits carcinogenic activity only in test systems capable of producing the N-hydroxy metabolite of napthylamine. Aromatic amines are capable of forming adducts with several DNA bases. Both heterocyclic amines (HCAs) and PAHs are formed when muscle meat, including beef, pork, fish, and poultry, is cooked using high-temperature methods, such as pan frying or grilling directly over an open flame. HCAs are also formed from fried foods, specifically fried meats (Turesky, 2007). Mutagenic activity of cooked meats at high temperature has been reported using the Ames Salmonella mutagenesis assay. In particular, an increase in HCAs (specifically MeIQx(2-amino-3,8-dimethylimidazo[4,5-b]quinoxaline), DiMeIQx (2-amino-3,4,8-dimethylimidazo [4,5-f]quinoxaline), and IQ (2-amino-3-methylimidazo [4,5-f]quinolone)) were found with increasing frying temperatures and cooking duration (Knize et al., 1995). In support of these experimental studies epidemiology linkages between consumption of high-temperature cooked foods and human cancer have been reported (Cross and Sinha, 2004; Rohrmann et al., 2007).

Classes and Mode of Action of Nongenotoxic (Epigenetic) Carcinogens

A number of chemicals that produce tumors in experimental animals following chronic treatment appear to act via mechanisms not involving direct binding, damage, or interaction of the chemical or its metabolites with DNA (Williams and Whysner, 1996). Based on the lack of genotoxicity, yet their ability to induce tumors in rodent models, these chemicals have been labeled nongenotoxic carcinogens. The organ and tissue targets induced by nongenotoxic carcinogens are many times in tissues where a significant incidence of background, spontaneous tumors is seen in the animal model. Prolonged exposure to relatively high levels of chemicals is usually necessary for the production of tumors. In addition, with nongenotoxic carcinogens, tumors are not theoretically expected to occur at exposures below a threshold at which relevant cellular effects are not observed. In contrast to DNA-reactive genotoxic effects, non–DNA-reactive mechanisms may be unique to the rodent species used for testing. Certain chemical carcinogens have been well studied and provide examples for the use of mechanistic information in risk assessment. Further, the biochemical modes of action for non–DNA-reactive carcinogens are diverse. Examples include chemicals that function via sustained cytotoxicity, receptor-mediated (eg, CAR, peroxisome proliferator–activated receptor alpha [PPARα], aryl hydrocarbon receptor [AhR]) effects, hormonal perturbation, as well as the induction of oxidative stress and modulation of methylation status (Table 8-10). Each of these potential mechanisms is discussed in greater detail in the following sections.

Cytotoxicity

Cytotoxicity and consequent regenerative hyperplasia is well-documented mode of action for a variety of non–DNA-reactive chemical carcinogens (Dietrich and Swenberg, 1991). Chloroform-induced liver and kidney tumors and melamine-induced bladder tumors are classic examples of chemical carcinogens that are classified as functioning via a cytolethal mode of action (Andersen et al., 1998; Bull et al., 1986; Pereira, 1994; Larson et al., 1994; Butterworth et al., 1990). Chemicals that function through this mechanism produce sustained cell death, often related to metabolism of the chemical, that is accompanied by persistent regenerative growth, resulting in the potential for the acquisition of “spontaneous” DNA mutations and allowing mutated cells to accumulate and proliferate. This process then gives rise to preneoplastic focal lesions that upon further expansion can lead to tumor formation. Chloroform has been shown to induce mouse liver tumors only at doses of compound that produce liver necrosis, thus demonstrating an association between necrosis with compensatory hyperplasia and the resulting tumorigenicity, and also supports that a threshold for the induction of tumors is likely at doses that do not produce toxicity. It is important to note that the induction of cytotoxicity may be observed with many carcinogens both genotoxic and nongenotoxic when high toxic exposures occur. Thus, the induction of cytotoxicity with compensatory hyperplasia may contribute to the observed tumorigenicity of many carcinogenic agents at high-dose levels.

α2u-Globulin-Binding Drugs

The carcinogens d-limonene, 1,4-dichlorobenzene, and trimethylpentane (Fig. 8-15) induce renal tumors selectively in the male rat, and provide excellent examples of the species, sex, and tissue specificity of non–DNA-reactive carcinogens. The mechanism for the species and sex specificity is related to the ability of these compounds to bind to α2u-globulin, a protein synthesized by the male rat liver at the onset of puberty, as the mechanism of tumorigenesis. α2u-Globulin is, filtered through the glomerulus, and only partially excreted (~50%) in the urine. The reabsorbed fraction accumulates in lysosomes of the P2 segment of the proximal tubules, where it is hydrolyzed to amino acids (Melnick et al., 1996). Chemicals with the ability to bind to α2u-globulin decrease the rate of digestion of α2u-globulin and results in the accumulation in the lysosomes, dysfunction of this organelle and subsequent release of digestive enzymes and cell necrosis. The greater loss of tubule cells leads to increased cell proliferation in the P2 segment, which may be responsible for the tumor development and malignant transformation (Dietrich and Swenberg, 1991).

Receptor-Mediated

CAR Receptor-Mediated (Phenobarbital-Like Carcinogens)

Phenobarbital is a commonly studied non–DNA-reactive compound that is known to cause tumors by a nongenotoxic mechanism involving liver hyperplasia (Williams and Whysner, 1996). One feature seen following phenobarbital exposure is the induction of P450 enzymes, particularly Cyp2b (Nims and Lubet, 1996). Because a number of diverse chemicals are known to induce various members of the P450 system (eg, dieldrin, ethanol, 2,3,7,8-Tetrachlorodibenzo-p-dioxin [TCDD]), the specificity of this effect to the carcinogenesis has been questioned. Recent evidence has shown that the induction of Cyp2b is mediated by activation of the constitutive androstane receptor (CAR), a member of the nuclear receptor family (Ueda et al., 2002; Kodama et al., 2004; Honkakoski et al., 1998). CAR-null mice show no induction of Cyp2b following phenobarbital exposure (Wei et al., 2000). Other phenobarbital responses that are critical for tumor formation include increased cell proliferation, inhibition of apoptosis, inhibition of gap junctional communication, hypertrophy, and development of preneoplastic focal lesions in the liver (Whysner et al., 1996), effects that have all been shown to be CAR-dependent (Wei et al., 2000; Kodama et al., 2004) (Fig. 8-16).

Peroxisome Proliferator–Activated Receptor α

A wide array of chemicals are capable of increasing the number and volume of peroxisomes in the cytoplasm of cells. These agents, termed peroxisome proliferators, include chemicals such as herbicides, chlorinated solvents (eg, trichloroethylene and perchloroethylene) plasticizers (eg, diethylhexylphthalate and other phthalates), lipid-lowering fibrate drugs (eg, ciprofibrate, clofibrate), and natural products. In addition, many of the chemicals within this group produce liver enlargement and hepatocellular carcinoma in rats and mice through non–DNA-reactive mechanisms (Lake, 1995; Reddy and Rao, 1997). Two additional tumor types are also associated with exposure to peroxisome proliferating compounds: Leydig cell tumors and pancreatic acinar cell tumors in rats. Studies conducted either in vivo or in vitro in primary hepatocyte cultures have shown important interspecies differences in the hepatic peroxisome proliferation responses to chemicals within the class of compounds. The rat and the mouse were clearly revealed as responsive species, whereas primates and the guinea pig proved to be nonresponders. The Syrian hamster exhibits an intermediate response (Lake, 1995; Bentley et al., 1993). Due to the wide structural diversity of this chemical class, the mechanism(s) involved in peroxisome proliferation and tumorigenesis went unrecognized for years. The currently accepted mode of action for this class of chemicals involves agonist binding to a nuclear hormone receptor, the PPARα. Largely through the use of PPARα knockout mice, it has been demonstrated that the activation of PPARα by agonists is needed for these chemicals to induce peroxisome proliferation and tumorigenesis in rodents (Issemann and Green, 1990; Lee et al., 1995; Peters et al., 1998, reviewed in Klaunig et al., 2003). PPARα is highly expressed in cells that have active fatty acid oxidation capacity (eg, hepatocytes, cardiomyocytes, enterocytes). It is well documented that PPARα plays a central role in lipid metabolism and acts as a transcription factor to modulate gene expression following ligand activation. This latter effect arises through the heterodimerization of PPARα and RXRα, which results in binding to response elements (PPREs) and subsequent modulation of target gene transcription (Fig. 8-17). Following this event is the induction of cell proliferation and suppression of apoptosis (Marsman et al., 1992; Burkhardt et al., 2001; James and Roberts, 1996). Both of these events would then be expected to affect tumor development as these effects would enhance the rate of fixation of DNA damage in the genome, leading to changes in gene expression, such as the silencing of tumor-suppressor genes or increased expression of oncogenes, or suppress apoptosis that may normally remove DNA-damaged, potentially tumorigenic, cells. Because humans are exposed to a number of these chemicals, the relevance of this mode of action to humans has been evaluated (Klaunig et al., 2003). Although the same events would be expected to occur in exposed humans, several species differences have been noted including a lack of induction of cell proliferation in nonhuman primates (Pugh et al., 2000), and the finding that PPARα in human liver is at least 10-fold lower compared with the rat or mouse (Palmer et al., 1998; Tugwood et al., 1998). Based on these kinetic and dynamic differences between species, it has been concluded that tumors are not likely to occur in humans by PPARα agonists (Klaunig et al., 2003).

Aryl Hydrocarbon Receptor

Activators of the AhR receptor including TCDD and selected polychlorinated- and brominated-biphenyl (PCBs and PBBs) compounds (Fig. 8-18) have been linked to tumor promotion in rodent livers (IARC, 1997a,b; 2011; Pitot et al., 1982). The tumor response has been determined to be AhR-dependent (Knutson and Poland, 1982) (Fig. 8-19). Upon ligand binding to the AhR, the ligand bound AhR translocates to the nucleus, dimerizes with the Ah receptor nuclear translocator (ARNT) and binds to aryl hydrocarbon response elements (AREs also known as doxin response elements [DRE] and xenobiotic response elements [XRE]) (for review see Nebert et al., 2000). AhR-ARNT–dependent genes include cytochrome P450 family members, NAD(P)H:quinone oxidoreductase, a cytosolic aldehyde dehydrogenase 3, a UDP-glucuronosyltransferase, and a glutathione transferase (Nebert et al., 2000), genes that are involved in metabolic activation as well as detoxification of chemical agents. It has been hypothesized that there are additional AhR-ARNT–dependent genes (Nebert et al., 2000). AhR knockout mice have a diminished response to tumor induction by AhR ligands (Nakatsuru et al., 2004), and conversely, constitutively overexpressed AhR resulted in an increased incidence of liver tumors (Moennikes et al., 2004). As seen with other hepatic tumor-promoting chemicals, TCDD also inhibits apoptosis in hepatic foci (Stinchcombe et al., 1995).

Hormonal Mode of Action

Hormonally active chemicals include steroids and peptide hormones that produce tissue-specific changes through interaction with a receptor. A number of non–DNA-reactive chemicals induce neoplasia through receptor-mediated mechanisms, and/or perturbation of hormonal balance. Trophic hormones are known to induce cell proliferation at their target organs. This action may lead to the development of tumors when the mechanisms of hormonal control are disrupted and some or other hormone shows persistently increased levels. Several well-studied examples include the induction of ovarian neoplasms via decreased estradiol and increased luteinizing hormone (LH) levels (Capen et al., 1995), and the induction of thyroid tumors in rats by phenobarbital-type P450 inducers (which modulate T3 and T4) (McClain, 1994; Whysner et al., 1996).

Estrogenic agents can induce tumors in estrogen-dependent tissue. Selective estrogenic agents (agonists and antagonists) are shown in Fig. 8-20. Oral administration of 17β-estradiol to female mice increases the incidence of mammary tumors (Welsch et al., 1977; Highman et al., 1978) whereas subcutaneous administration of estradiol to young female mice produced tumors of the cervix and vagina (Pan and Gardner, 1948). Evidence that estrogenic chemicals are carcinogenic to humans comes from epidemiological data on breast and ovarian cancer, which indicates that individuals with higher circulating estrogen levels and those with exposure to the potent estrogenic chemical diethylstilbesterol (DES) are at increased risk for cancer development. DES was first shown to induce mammary tumors in male mice following subcutaneous administration of the hormone. DES has been causally associated to the higher incidence of adenocarcinomas of the vagina and cervix in daughters of women treated with the hormone during pregnancy (Herbst and Scully, 1970; Herbst et al., 1972; Noller et al., 1972). The mechanism of action for DES is believed to function through its ability to induce aneuploidy (Tsutsui et al., 1997; Li et al., 1997). The effects of steroidal chemicals on the cell cycle (Sutherland et al., 1995) and on microtubule assembly (Metzler et al., 1996) may be important in the aneuploidy inducing effects of some hormonal agents (Li et al., 1996).

Chronic exogenous administration of hormonally active chemicals including synthetic estrogens and anabolic steroids can increase hepatic adenoma incidence in rats (Li et al., 1992) and in humans (IARC, 1997a,b). Women of childbearing age are sensitive to hepatic adenoma formation, which can be exacerbated by oral contraceptive use. These adenomas will regress upon hormone cessation (Edmondson et al., 1977) and can progress with continued administration (Christopherson et al., 1978). Chronic administration (>8 years) is required to detect this increased liver tumor risk from oral contraceptives (Tavani et al., 1993).

Many substances in plants (phytoestrogens) have been described. These include compounds such as genestein, daidzein, glycetein, equol, and their metabolites found in soy products and various lignan derivatives (Adlercreutz and Mazur, 1997). In addition, a number of environmental nonsteroidal synthetic compounds demonstrate apparent estrogenic activity (eg, nonyl-phenol, bisphenol-A, chlorinated hydrocarbons; Soto et al., 1997). The potential for these chemicals to induce cancer in humans is an area of current investigation.

Species and tissue specificity in response to receptor- and hormone-mediated carcinogenesis is often observed. As an example, tamoxifen is antiestrogenic in the chick oviduct, estrogenic in the mouse uterus with acute administration, but antiestrogenic with chronic administration in the same tissue, whereas it is estrogenic in the rat uterus. Although this may be in part due to tissue and species differences in coactivator and corepressor levels and availability, many other pharmacokinetic and pharmacodynamic properties are likely to participate (Carthew et al., 1995). To further exemplify the complexity of the role of estrogen in cancer development, estrogens have also been shown to act as a protective agent in prostate cancer.

Induction of ovarian tumors by dietary administration of nitrofurantoin in mice is an example of a tumorigenic effect secondary to drug-induced hormonal disturbance. Treatment resulted in ovarian atrophy with absence of graafian follicles and sterility. The reduction of follicles induced a reduction of sex steroid hormone by the ovary, resulting in increased production of gonadotrophins, notably LH, presumably due to the decreased negative feedback on the hypothalamus–pituitary axis by estrogens. Persistent stimulation by LH of the ovary cells resulted in development of tumors (Capen et al., 1995). A similar effect is known to occur with dopamine agonists. The induction of hypoprolactinemia changes the number of LH receptors and leads to enhanced sensitivity to LH and a higher stimulation by LH similarly to nitrofurantoin.

A reduction of thyroid hormone concentrations (T4 and/or T3) and increased thyroid-stimulating hormone (TSH) have been shown to induce neoplasia in the rodent thyroid. TSH increases proliferative activity in the thyroid. After chronic treatment chemical-induced TSH increases leading to follicular cell hypertrophy, hyperplasia, and eventually neoplasia. Inducers of metabolic enzymes in the liver, a classic and well-studied example being phenobarbital (Hood et al., 1999), result in increased thyroid hormone metabolism and as such lead to increase in TSH levels (Fig. 8-21). It is this latter event that is associated with the development of thyroid tumors in rodents. Recent evidence suggests that the enhancement of metabolism and excretion of thyroid hormones by xenobiotics (via induction of phase II enzymes) is a consequence of CAR (and presumably PXR) activation. Qualitatively, thyroid gland function in rodents is much more susceptible to disturbances of the thyroid hormone levels than it is in humans. In contrast to humans, rodents lack the thyroid-binding globulin, which is the predominant plasma protein (at least in humans) responsible for thyroxin binding and transport (McClain, 1995). A relatively higher level of functional activity is present in rat thyroid compared to humans.

DNA Methylation and Carcinogenesis

Post-DNA synthetic methylation of the five position on cytosine (5-methylcytosine; 5mC) is a naturally occurring modification to DNA in higher eukaryotes that influences gene expression (Holliday, 1990) (Fig. 8-22). Under normal conditions, DNA is methylated symmetrically on both strands. Immediately following DNA replication, the newly synthesized double-stranded DNA contains hemimethylated sites that signal for DNA maintenance methylases to transfer methyl groups from S-adenosylmethionine to cytosine residues on the new DNA strand (Hergersberg, 1991). The degree of methylation within a gene inversely correlates with the expression of that gene; hypermethylation of genes is associated with gene silencing whereas hypomethylation results in an enhanced expression of genes. Several chemical carcinogens are known to modify DNA methylation, methyltransferase activity, and chromosomal structure. During carcinogenesis, both hypomethylation and hypermethylation of the genome have been observed (Counts and Goodman, 1995; Baylin, 1997). Increased methylation of CpG islands have been observed in bladder cancer and tumor-suppressor genes such as the retinoblastoma gene, p16^ink4a, and p14^ARF have been reported to be hypermethylated in tumors (Salem et al., 2000; Stirzaker et al., 1997; Myöhänen et al., 1998; Esteller et al., 2001; Belinsky et al., 1998). Hypomethylation has been associated with increased mutation rates. Most metastatic neoplasms in humans have significantly lower 5MeC than normal tissue (Gama-Sosa et al., 1983); furthermore, many oncogenes are hypomethylated and their expression amplified.

Choline and methionine, which can be derived from dietary sources, provide a source of methyl groups used in methylation reactions. Rats exposed to choline and/or methionine-deficient diets resulted in hepatocellular proliferation and neoplasia (Abanobi et al., 1982; Wainfan and Poirier, 1992) (Table 8-11). In rodents fed choline or methyl donor groups deficient diets, the hepatic neoplasia is thought to arise from hypomethylation of c-myc, c-fos, and c-H-ras proto-oncogenes, due to the decreased availability of S-adenosylmethionine (Wainfan and Poirier, 1992; Newberne et al., 1982). Chemicals such as diethanolamine result in hepatic neoplasia, in part, via a mechanism involving choline depletion, altered methylation and modulation of gene expression (Bachman et al., 2006; Kamendulis and Klaunig, 2005). Reactive oxygen species have also been shown to modify DNA methylation resulting in changes in DNA methylation profiles through interfering with the ability of methyltransferases to interact with DNA, leading to hypomethylation of CpG sites (Weitzman et al., 1994; Turk et al., 1995). Also, the presence of 8-OHdG in DNA can lead to hypomethylation of DNA because this adduct, if present in CpCpGpGp sequences, will inhibit the methylation of adjacent C residues. Thus, oxidative DNA damage may be an important contributor to the carcinogenesis process brought about by the loss of DNA methylation, allowing the expression of normally quiescent genes. Also, the abnormal methylation pattern observed in cells transformed by chemical oxidants may contribute to an overall aberrant gene expression and promote the tumor process.

Oxidative Stress and Chemical Carcinogenesis

Experimental evidence has shown that increases in reactive oxygen in the cell, through either physiological modification or chemical carcinogen exposure, contribute to the carcinogenesis processes (Guyton and Kensler, 1993; Trush and Kensler, 1991; Vuillaume, 1987; Witz, 1991). Reactive oxygen species encompass a series of reactive compounds including the superoxide anion (^•O₂–), hydroperoxyl radical (HO₂^•.), hydrogen peroxide (H₂O₂), and the hydroxyl radical (•OH), all derived through the reduction of molecular oxygen (Table 8-12). Oxygen radicals can be produced by both endogenous and exogenous sources and are typically counterbalanced by antioxidants (Table 8-13). Antioxidant defenses are both enzymatic (eg, superoxide dismutase, glutathione peroxidase, and catalase) and nonenzymatic (eg, vitamin E, vitamin C, β-carotene, and glutathione) (Betteridge, 2000; Abuja, 2001); importantly, many of these antioxidants are provided through dietary intake (Clarkson and Thompson, 2000).

Endogenous sources of reactive oxygen species include oxidative phosphorylation, P450 metabolism, peroxisomes, and inflammatory cell activation (Table 8-13). Within the mitochondria, a small percentage of oxygen is converted into the superoxide anion via 1-electron reduction of molecular oxygen. Superoxide can be dismutated by superoxide dismutase to yield hydrogen peroxide (Barber and Harris, 1994). In the presence of partially reduced metal ions hydrogen peroxide is converted into the highly reactive hydroxyl radical through Fenton and Haber–Weiss reactions (Betteridge, 2000). Neutrophils, eosinophils, and macrophages represent another intracellular source of reactive oxygen species. Activated macrophages, through a “respiratory burst,” elicit a rapid increase in oxygen uptake that gives rise to a variety of reactive oxygen species including superoxide anion, hydrogen peroxide, and nitric oxide. The release of cytokines and reactive oxygen intermediates from activated Kupffer cells (the resident macrophage of the liver) has been implicated in hepatotoxicological and hepatocarcinogenic events (Rose et al., 1999; Rusyn et al., 1999); in particular, recent studies show that the Kupffer cell may function at the promotion stage of carcinogenesis. Reactive oxygen species can also be produced by cytochrome P450-mediated mechanisms including (1) redox cycling in the presence of molecular oxygen, (2) peroxidase-catalyzed single electron drug oxidations, and (3) “futile cycling” of cytochromes P450 (Parke, 1994; Parke and Sapota, 1996). Ethanol, phenobarbital, and a number of chlorinated and nonchlorinated compounds such as dieldrin, TCDD, and lindane are among the xenobiotics shown to increase reactive oxygen species through P450-mediated mechanisms (Eksrom and Ingleman-Sundberg, 1989; Rice et al., 1994; Klaunig et al., 1997). Chemicals classified as PPARα agonists (eg, clofibrate, phthalate esters) represent chemicals that induce cytochrome P4504A and increase the formation of peroxisomes. As such, an increase in H₂O₂ production often accompanies exposure to chemicals that stimulate the number and activity of peroxisomes (Rao and Reddy, 1991; Wade et al., 1992). Through these or other currently unknown mechanisms, a number of chemicals that induce cancer (eg, chlorinated compounds, radiation, metal ions, barbiturates, and some PPARα agonists), induce reactive oxygen species formation and/or oxidative stress (Rice-Evans and Burdon, 1993; Klaunig et al., 1997, 2011).

Oxidative DNA Damage and Carcinogenesis

Reactive oxygen species left unbalanced by antioxidants can result in damage to cellular macromolecules. In DNA, reactive oxygen species can produce single- or double-stranded DNA breaks, purine, pyrimidine, or deoxyribose modifications, and DNA cross-links (von Sonntag, 1987; Dizdaroglu, 1992; Demple and Harrison, 1994). Persistent DNA damage can result in either arrest or induction of transcription, induction of signal transduction pathways, replication errors, and genomic instability, events that are potentially involved in carcinogenesis. Oxidation of guanine at the C8 position, results in the formation of 8-hydroxydeoxyguanosine (8-OHdG; see Fig. 8-7). This oxidative DNA adduct is mutagenic in bacterial and mammalian cells, produces dose-related increases in cellular transformation, and causes G → T transversions that are commonly observed in mutated oncogenes and tumor-suppressor genes (Zhang et al., 2000; Shibutani et al., 1991; Moriya, 1993; Hussain and Haris, 1998). During DNA replication, 8-OHdG in the nucleotide pool can incorporate into the DNA template strand opposite dC or dA resulting in A:T to C:G transversions (Demple and Harrison, 1994; Cheng et al., 1992). Oxidative damage to mitochondrial DNA and mutations in mitochondrial DNA have been identified in a number of cancers (Schumacher et al., 1973; Cavalli and Liang, 1998; Tamura et al., 1999; Horton et al., 1996). Compared to nuclear DNA, the mitochondrial genome is relatively susceptible to oxidative base damage due to (1) close proximity to the electron transport system, a major source of reactive oxygen species (Barber and Harris, 1994); (2) mitochondrial DNA is not protected by histones; and (3) DNA repair capacity is limited in the mitochondria (Sawyer and Van Houten, 1999; Bohr and Dianov, 1999). Although many pathways exist that enable the formation of oxidative DNA damage, mammalian cells also possess specific repair pathways for the remediation of oxidative DNA damage (Tchou and Grollman, 1993; Fortini et al., 1999).

Aside from oxidized nucleic acids, oxygen radicals can damage cellular biomembranes resulting in lipid peroxidation. Peroxidation of biomembranes generates a variety of products including reactive electrophiles such as epoxides and aldehydes, including malondialdehyde (MDA) (Janero, 1990). MDA is a highly reactive aldehyde and exhibits reactivity toward nucleophiles, and can form MDA–MDA dimers. Both MDA and the MDA–MDA dimers are mutagenic in bacterial assays and the mouse lymphoma assay (Riggins and Marnett, 2001).

Oxidative Stress and Cell Growth Regulation

Reactive oxygen species production and oxidative stress can affect both cell proliferation and apoptosis (Burdon, 1995; Slater et al., 1995; Cerutti, 1985). H₂O₂ and superoxide anion can induce cell proliferation in several mammalian cell types (D’Souza et al., 1993). Conversely, high concentrations of reactive oxygen species can initiate apoptosis (Dypbukt et al., 1994). It has clearly been demonstrated that low levels of reactive oxygen species influence signal transduction pathways and alter gene expression (Fiorani et al., 1995). Many xenobiotics, by increasing cellular levels of oxidants, alter gene expression through activation of signaling pathways including cAMP-mediated cascades, calcium-calmodulin pathways, and transcription factors such as AP-1 and NF-κB as well as signaling through mitogen-activated protein (MAP) kinases (extracellular signal-regulated kinases [ERK], c-Jun N-terminal kinases [JNK], and the p38 kinases) (Muller et al., 1997; Kerr, 1992; Timblin et al., 1997). Activation of these signaling cascades by reactive oxygen species induced by chemical carcinogens ultimately leads to altered gene expression for a number of genes including those affecting proliferation, differentiation, and apoptosis (Chang and Karin, 2001; Martindale and Holbrook, 2002; Brown et al., 1998; Amstad et al., 1992; Hollander and Fornace, 1989; Zawaski, 1993; Pinkus, 1993) (Fig. 8-23). Activation of NFκB, a ubiquitously expressed transcription factor, is regulated, in part, by reactive oxygen species and the cellular redox status. NFκB activation has been observed following a variety of extracellular stimuli including exposure to chemical carcinogens and physical agents that induce oxidative stress (Rusyn et al., 2003; Glauert et al., 2002; Beauerle et al., 1988; Pahl, 1999; Li and Karin, 1998; Molina et al., 2001; Schulze-Oshoff et al., 1998; Nebreda and Porrai, 2000) (Fig. 8-24).

Gap Junctional Intercellular Communication and Carcinogenesis

Cells within an organism communicate in a variety of ways including through gap junctions, which are aggregates of connexin proteins that form a conduit between two adjacent cells (Lowenstein et al., 1981) (Fig. 8-25). Gap junctional intercellular communication appears to play an important role in the regulation of cell growth and cell death, in part, through the ability to exchange small molecules (<1 kDa) between cells through gap junctions. Loss of gap junctions or decreased cell-to-cell communication is seen during liver regeneration following partial hepatectomy, a situation in which gap junction expression remains decreased until liver mass and lobular conformation are restored. Aberrant growth control is an essential feature of cancer cells, and because the absence or reduction in cell-to-cell communication has been observed between cancer cells, between cancer and normal cells, and in transformed cells, it has been speculated that altered gap junctional cell communication is involved in carcinogenesis (Trosko and Ruch, 1998). If cell communication is blocked between tumor and normal cells, the exchange of growth inhibitory signals from normal cells would be prevented from acting on initiated cells, thus allowing the potential for unregulated growth and clonal expansion of initiated cell populations. This would therefore allow for the acquisition of additional genetic changes that may lead to neoplastic transformation (Klaunig and Ruch, 1990).

Intercellular communication is also decreased by growth factor administration and following exposure to a variety of tumor-promoting compounds (eg, phenobarbital, phthalates, dieldrin, phorbol esters). Tumor-promoting chemicals inhibit gap junctional intercellular communication in a number of cell types following exposure both in vivo and in vitro (Klaunig and Ruch, 1990). The ability of a tumor-promoting compound to block cell-to-cell communication in cultured cells correlates with its ability to induce rodent tumors (Klaunig and Ruch, 1987). In addition, the inhibition of gap junctional intercellular communication by tumor promoters appears to be tissue- and species-specific; they produced inhibition in target organs and sensitive species, whereas they do not inhibit cell-to-cell communication in nontarget tissues for carcinogenicity (Klaunig and Ruch, 1987). The mechanisms by which tumor promoters inhibit intercellular communication are not clear but may be due to decreased synthesis and/or enhanced degradation of gap junction proteins in the cell, reducing the number of channels in cell membranes, or other changes that modulate gap junction function and/or protein expression (Klaunig and Ruch, 1990; Sugie et al., 1987).

Inorganic Carcinogens

Metals

Several metals exhibit carcinogenicity in experimental animals and/or exposed humans. Table 8-14 provides a listing of some common metals and their corresponding carcinogenicity in animals and humans. A recent IARC review has examined in detail inorganic compounds deemed human carcinogens (IARC, 2010). Additional details on selected compounds are provided below.

Arsenic

Arsenic compounds are poorly mutagenic in both bacterial and mammalian cell assays (Lofroth and Ames, 1978). Metallic arsenic, arsenic trioxide, sodium arsenite, sodium arsenate, potassium arsenite, lead arsenate, calcium arsenate, and pesticide mixtures containing arsenic have been tested for carcinogenicity in experimental animals (IARC, 1980, 1987). In the majority of studies in experimental animals, including oral-exposure studies in mice, rats, and dogs, dermal-exposure studies in mice, inhalation-exposure studies in mice, injection studies in mice and rats, and intramedullary-injection studies in rats and rabbits, no tumors were observed or the results were inconclusive, and thus it has previously been concluded that limited evidence exists for the carcinogenicity of inorganic arsenic compounds in experimental animals (IARC, 1987).

In contrast, inorganic arsenic compounds are known human carcinogens based on sufficient evidence of carcinogenicity in humans. Epidemiological studies of humans exposed to arsenic compounds demonstrated that exposure to inorganic arsenic compounds increases the risk of cancer in the skin, lung, digestive tract, liver, bladder, kidney, and lymphatic and hematopoietic systems (IARC, 1973, 1980). Several of the epidemiological studies have reported dose–response relationships between arsenic in drinking water and several types of cancer, including bladder, kidney, lung, and skin cancer (Cantor, 1997; Ferreccio et al., 2000). The mechanisms for cancer formation are unclear but are believed to possibly involve the induction of oxidative stress, altered cell signaling, modulation of apoptosis, and/or altered cell cycle (Hughes and Kitchin, 2006; Harris and Shi, 2003; Quian et al., 2003). The latency period in humans of arsenic-related carcinogenesis is considered to be 30 to 50 years. The first signs of chronic exposure, frequently seen in water supplies contaminated with arsenic, are skin pigmentation, depigmentation, hyperkeratosis of palms and soles, and skin lesions. A unique peripheral vascular disease associated with chronic arsenic exposure is black foot disease, starting with numbness and ulceration of extremities and ending in gangrene and spontaneous amputations (Chen et al., 1988).

Beryllium

Beryllium and its salts are not mutagenic and do not appear to induce cellular transformation (IARC, 1993). Mechanistically, beryllium salts bind to nucleoproteins and inhibit enzymes involved in DNA synthesis, resulting in infidelity of DNA synthesis and also induce gene mutations in cultured cells (Leonard and Lauwerys, 1987). Studies in animal models have consistently reported increases in lung tumors in rodents and nonhuman primates exposed to beryllium or beryllium compounds (IARC, 1993; Finch et al., 1996; NTP, 1998). Beryllium metal and several beryllium compounds (eg, beryllium-aluminum alloy, beryl ore, beryllium chloride, beryllium hydroxide, beryllium sulfate tetrahydrate, and beryllium oxide) induced lung tumors in rats. Beryllium oxide and beryllium sulfate produced lung cancer (anaplastic carcinoma) in monkeys after intrabronchial implantation or inhalation. In rabbits, osteosarcomas were reported after exposure to beryllium metal, beryllium carbonate, beryllium oxide, beryllium phosphate, beryllium silicate, or zinc beryllium silicate (IARC, 1993).

Beryllium and beryllium compounds have been classified as human carcinogens based on animal studies and evidence of carcinogenicity in humans. Epidemiological studies indicate an increased risk of lung cancer in occupational groups exposed to beryllium or beryllium compounds (Steenland and Ward, 1991; Ward et al., 1992). Furthermore, an association with lung cancer has consistently been observed in occupational populations exposed to beryllium or beryllium compounds. Acute beryllium pneumonitis, a marker for exposure to beryllium, has been shown to be associated with higher lung cancer rates (Steenland and Ward, 1991).

Cadmium

Animal studies have shown that cadmium and cadmium compounds induce tumor formation at various sites in multiple species of experimental animals, following multiple exposure routes, including the induction of prostate tumors in rats, testicular tumors in rats and mice, lymphoma in mice, adrenal gland tumors in hamsters and mice, and lung and liver tumors in mice (IARC, 1993; Waalkes et al., 1994; 1999). It has been suggested that ionic cadmium, or compounds that release ionic cadmium, are the cause of genetic damage, and thus the carcinogenic species. Increased frequencies of chromosomal aberrations (changes in chromosome structure or number) have been observed in lymphocytes of workers occupationally exposed to cadmium. Many studies of cultured mammalian cells have shown that cadmium compounds cause genetic damage, including gene mutations, DNA strand breaks, chromosomal damage, cell transformation, and disrupted DNA repair (IARC, 1993).

Cadmium and cadmium compounds have been classified as known human carcinogens based on evidence of carcinogenicity in humans, including epidemiological and mechanistic information that indicate a causal relationship between exposure to cadmium and cadmium compounds and human cancer (9th RoC). Epidemiological studies of cadmium workers found that exposure to various cadmium compounds increased the risk of death from lung cancer (IARC, 1993). Follow-up analysis of some of these cohorts has confirmed that cadmium exposure is associated with elevated lung cancer risk under some industrial circumstances (Sorahan et al., 1995; Sorahan and Lancashire, 1997). Some epidemiological evidence has also suggested an association between cadmium exposure and prostate (Shigematsu et al., 1982; van der Gulden et al., 1995), kidney (Mandel et al., 1995), and bladder cancer (Siemiatycki et al., 1994).

Chromium

Chromium has multiple oxidation states—from −2 to +6; however, the most common forms are the trivalent (III) and hexavalent (VI) forms. With regard to carcinogenicity, chromium III does not exhibit carcinogenicity in laboratory animals whereas chromium VI has been shown to be positive for genotoxicity and carcinogenicity in a variety of bioassays (Langard, 1988; IARC, 1990). Chromium VI compounds cause genetic damage including gene mutations and DNA damage in bacteria. Several chromium VI compounds also caused mutations in yeast and insects. Many chromium VI compounds caused genetic damage in cultured human and other animal cells and in experimental animals exposed in vivo, including SCE, chromosomal aberrations, and cell transformation. Chromosomal aberrations, SCE, and aneuploidy were observed in workers exposed to chromium VI compounds (IARC, 1990). Chromium VI (calcium chromate, chromium trioxide, sodium dichromate, lead chromates, strontium chromate, or zinc chromates) exposure of rats following inhalation, intrabronchial, intrapleural, intratracheal, intramuscular, or subcutaneous administration resulted in benign and malignant lung tumors were observed in rats in a number of studies. In mice, calcium chromate caused benign lung tumors and chromium trioxide caused malignant lung tumors. Exposure of hamsters, guinea pigs, and rabbits to chromium (VI) compounds by intratracheal instillation did not cause lung tumors (IARC, 1980, 1990). Although the mechanisms for chromium VI carcinogenicity remain unresolved, it has been speculated that the reduction of chromium VI by glutathione is involved (Connett and Whtterhahn, 1985; Kortenkamp and O’Brien, 1994).

Hexavalent chromium (chromium VI) compounds have been classified as known human carcinogens based on data from animals studies and human epidemiological studies. Human epidemiological studies have consistently reported increased risks of lung cancer among chromate workers. Chromate workers are exposed to a variety of chromium compounds, including chromium (VI) and trivalent (III) compounds. In addition, an increased risk of a rare cancer of the sinonasal cavity was observed in these workers (IARC, 1990). Some studies suggested that exposure to chromium among workers such as chromium-exposed arc welders, chromate pigment workers, chrome platers, and chromium tanning workers may be associated with leukemia and bone cancer (Costa, 1997).

Nickel

Many studies in cultured rodent and human cells have shown that a variety of nickel compounds, including both soluble and insoluble forms of nickel, exhibit genotoxicity, producing DNA strand breaks, mutations, chromosomal damage, cell transformation, and modulation of DNA repair. Soluble nickel salts can be complete carcinogens and/or initiators of carcinogenesis (Diwan et al., 1992; Kasprzak et al., 1990). In rats and mice, inhalation or intratracheal instillation of nickel subsulfide or nickel oxide produced dose-related increases of benign and malignant lung tumors (IARC, 1990; NTP, 1996). Inhalation of nickel compounds also caused malignant and benign pheochromocytoma in rats (NTP, 1996). Short-term intraperitoneal exposure during gestation to soluble nickel salt induced malignant pituitary tumors in the offspring. Additionally, exposure to nickel acetate through the placenta followed by exposure of the offspring to barbital (a known tumor promoter) produces tumors of the kidney (renal cortical and pelvic tumors) (Diwan et al., 1992). In adult rats, injection of soluble nickel salts followed by exposure to a promoting carcinogen resulted in kidney cancer (renal cortical adenocarcinomas) that frequently metastasized to the lung, liver, and spleen (Kasprzak et al., 1990). The carcinogenic properties of metallic nickel is believed to be due to ionic nickel, which can slowly dissolve in the body from nickel compounds.

Nickel compounds are classified as known human carcinogens (11th RoC) based on animal data and sufficient evidence of carcinogenicity from human studies. The IARC (1990, 2011) evaluation of nickel and nickel compounds concluded that nickel compounds are carcinogenic to humans based on sufficient evidence for the carcinogenicity of nickel compounds in the nickel refining industry and very strong evidence of carcinogenicity of a variety of nickel compounds in experimental studies in rodents. Several cohort studies of workers exposed to various nickel compounds showed an elevated risk of death from lung and nasal cancers (IARC, 1990). An excess risk of lung and nasal cancer was seen in nickel refinery workers exposed primarily to soluble nickel compounds (Andersen et al., 1996).

Lead

Lead compounds do not appear to cause genetic damage directly, but may do so through several indirect mechanisms, including inhibition of DNA synthesis and repair, oxidative damage, and interaction with DNA-binding proteins and tumor-suppressor proteins (NTP, 2003). Lead has exhibited conflicting results concerning its genotoxicity; it does not cause mutations in bacteria, but does cause chromosomal aberrations in vitro and in vivo, and causes DNA damage in vivo and in cell-free systems, whereas in mammalian systems, conflicting results were observed. Lead also inhibits the activity of DNA and RNA polymerase in cell-free systems and in mammalian cell cultures. Conflicting results were observed for SCE and micronucleus formation in mammalian test systems.

In studies with laboratory animals, carcinogenicity has been observed for soluble (lead acetate and lead subacetate) and insoluble (lead phosphate, lead chromate) inorganic lead compounds as well as for tetraethyl lead (an organic lead compound), following exposure via oral, injection, and in offspring exposed via the placenta or lactation. Although kidney tumors (including adenomas, carcinomas, and adenocarcinomas) were most frequently associated with lead exposure, tumors of the brain, hematopoietic system, and lung were reported in some studies (IARC, 1980; 1987, 2011; Waalkes et al., 1995). Lead also appears to function as a tumor promoter, leading to increased incidence in kidney tumors initiated by N-ethyl-N-hydroxyethylnitrosamine and N-(4′-fluoro-4-biphenyl)acetamide) (IARC, 1980, 1987). The mechanisms by which lead causes cancer are not understood.

Lead and lead compounds are classified as reasonably anticipated to be human carcinogens based on limited evidence from studies in humans and sufficient evidence from studies in experimental animals. Lead exposure has been associated with increased risk of lung, stomach, and bladder cancer in diverse human populations (Fu and Boffetta, 1995; Steenland and Boffetta, 2000; NTP, 2003). Epidemiological studies link lead exposure to increased risk for lung and stomach cancer. However, most studies of lead exposure and cancer reviewed had limitations, including poor exposure assessment and failure to control for confounding factors.

Modifiers of Chemical Carcinogenic Effects

Genetic and environmental factors have a significant impact on the way in which individuals and/or organisms respond to carcinogen exposure. As with most genes, enzymes that metabolize carcinogens are expressed in a tissue-specific manner. Within tissues, the enzymatic profile can vary with cell type or display differential localization within cells. Further, carcinogen metabolizing enzymes are differentially expressed among species. These differences may represent an underlying factor explaining the differential responses to chemical carcinogens across species.

The expression of many carcinogen metabolizing enzymes is highly inducible by foreign chemicals. Compounds such as benzo(a)pyrene highly induce P450 isozymes CYP1A1 and 1A2, enzymes that display catalytic activity toward a number of polycyclic aromatics. A number of other enzymes, for example, GST isozymes, Uridine Diphosphate Glucuronyltransferases (UDPGTs) are also induced by PAHs, thus the net result is enhanced metabolism of PAHs. Numerous other chemical carcinogens are potent inducers of drug metabolizing enzymes; examples include chemicals such as phenobarbital, dichlorodiphenyltrichloroethane (DDT), ethanol, 17β-estradiol, PPARα agonists, each of these results in the upregulation of a distinct set of phase I and phase II enzymes. Many chemicals produce inhibitory effects on drug metabolism enzymes, and could either increase the half-life of a chemical carcinogen or prevent or delay the formation of active metabolites of carcinogens.

Polymorphisms in Carcinogen Metabolism and DNA Repair

Genetic polymorphisms arise from human genetic variability. A genetic polymorphism is when a gene has more than one allele. In assessing variability in the human genome project it was found that base variations occurred at approximately once in every 1000 base pairs. Therefore, there may be over one million genetic variations between any two individuals. A single nucleotide polymorphism (SNP) is a variant in DNA sequence found in greater than 1% of the population (Fig. 8-26). Thus, by definition, changes in DNA sequence goes from mutation to polymorphism when a unique genotype is seen in over 1% of the population. Over three million candidate SNPs have been identified to date with up to 10 million being estimated to be present within the human genome. In carcinogenesis, genetic polymorphisms may account for the susceptibility of some individuals to certain cancers. In particular, SNPs of genes involved in carcinogen metabolism (activation and detoxification) and DNA repair, have received considerable attention. A number of polymorphisms have been described in carcinogen metabolizing enzymes, with certain alleles linked to altered risk of selective cancers (Boddy and Ratain, 1997; Naoe et al., 2000). Glutathione-S-transferases (GSTs), which are involved in detoxification of many chemical carcinogens and also in the remediation of oxidative stress, are highly polymorphic in humans. Polymorphisms of the GSTM1 phenotype have been related to bladder, gastric, and lung cancer in humans (Welfare et al., 1999; Tsuchida and Sato, 1992). Individuals expressing the GSTM1 null genotype have a 1.5-fold higher risk for bladder cancer and 1.7-fold increase risk for gastric cancer compared to individuals expressing the genotype (Risch et al., 2001; Ates et al., 2005). The GSTM1 isoform is of particular important in carcinogenesis because this isoform exhibits the highest reactivity toward epoxides. GSTT1 isoform null individuals show a three-fold increase risk in breast cancer, a 1.6-fold increase in gastric cancer, and a four-fold increase in liver cancer compared to individuals having the genotype (Bosch et al., 2006).

Similarly in women (nonsmokers) with lung cancer, a higher level of PAH DNA adducts have been detected. Higher DNA adduct levels are also found to be associated with the expression of the cytochrome P4501A1 (CYP1A1) gene, which bioactivates PAHs (Han et al., 2003). An increase in benzo(a)pyrene DNA adduct levels correlate with increased expression of the CYP1A1. DNA repair genes, in particular oxidative DNA repair, OGG1, involved in repair of 8-OHdG–induced damage, have been linked to an increase in lung cancer in individuals with lower OGG1 expression (Goode et al., 2002; Klaunig et al., 2011).

In the case of prostate cancer, studies have suggested roles for a number of candidate SNPs. Recent evidence suggests that the presence of a genetic variant that is more prevalent in African American men that is less effective of clearing reactive oxygen species. Examples of candidate genes that likely contribute to prostate cancer carcinogenesis include the androgen receptor gene, steroid synthesis, and metabolism genes (CYP17, CYP1B1, CYP3A4, CYP3A5, CYP3A43, SD5A2, HSD3B1, HSD3B2, CYP19), transmembrane drug efflux genes (ABCG2), and reactive oxygen species metabolizing genes (MnSOD, APE1, XRCC1, GSTP1; Dong, 2005). Individually, polymorphisms in each of these genes have been associated with an increased risk of prostate cancer.

The importance of SNPs, cancer susceptibility, and cancer risk are illustrated in Table 8-15. If exposure to a chemical carcinogen is low and the genetic susceptibility for genes related to the carcinogenic activity is low, then the risk for cancer will also be low. If exposure to a chemical carcinogen is high but the genetic susceptibility is low, then the risk for cancer development is likely to be low. However, if the genetic susceptibility is high, then exposure to a chemical carcinogen will result in a higher risk for cancer development.

Proto-Oncogenes and Tumor-Suppressor Genes

Proto-oncogenes and tumor-suppressor genes play a key role in cancer induction. These genes encode a wide array of proteins that function to control cell growth and proliferation. Common characteristics of oncogenes and tumor-suppressor genes are shown in Table 8-16. Mutations in both oncogenes and tumor-suppressor genes contribute to the progressive development of human cancers. Accumulated damage to multiple oncogenes and/or tumor-suppressor genes can result in altered cell proliferation, differentiation, and/or survival of cancer cells (Croce, 2008; Huff, 2011).

Retroviruses

Evidence that specific genes (oncogenes) could cause cancer first came from studies showing sarcomas could be produced in healthy chickens by injecting them with cell-free extracts derived from chicken tumors (Rous, 1911). The transforming agent in the filtrate eventually was shown to be an RNA-type tumor virus, the Rous sarcoma virus (RSV) that is capable of transforming a normal cell and produce sarcomas within weeks of injection (Temin and Rubin, 1958). The genome of RSV and other retroviruses consists of two identical copies of mRNA, which is then reverse transcribed into DNA and incorporated into the host-cell genome. Oncogenic transforming viruses such as RSV contain the v-src gene, a gene required for cancer induction. It was later demonstrated that normal cells from chickens and other species contain a gene closely related to v-src in RSV; the normal “proto-oncogene” commonly is distinguished from the viral gene by the prefix “c” (c-src; Oppermann et al., 1979). This discovery showed that cancer may be induced by the action of normal, or nearly normal, genes.

DNA Viruses

Six major classes of DNA tumor viruses have been identified: simian virus 40 (SV40), polyoma virus, hepatitis B virus, papilloma viruses, adenoviruses, herpes viruses, and poxviruses. Unlike retroviral oncogenes, which are derived from normal cellular genes and have no function for the virus, the known oncogenes of DNA viruses are integral parts of the viral genome required for viral replication. SV40 or polyoma viruses are not capable of inducing tumors in their natural host species (monkey and mice, respectively). Infection by small DNA viruses is lethal to most non-host animal cells; however, a small proportion integrates the viral DNA into the host-cell genome. The cells that survive infection become permanently transformed due to the presence of one or more oncogene in the viral DNA. For example, SV40 codes for the large T antigen, which alone is capable of inducing transformation (Fried and Privies, 1986). Both SV40 and polyoma viruses use the host-cells DNA synthetic apparatus for their own replication. Papilloma viruses can infect and cause tumors in rabbits, cows, and humans (Lancaster and Olsen, 1982). Of the human papilloma viruses, types 6, 10, and 11 are associated with genital warts whereas types 16, 18, 31, and 33 are associated with human cervical cancers (Vousden, 1989). Herpes viruses are complex and are capable of producing tumors in frogs, chickens, monkeys, and humans (Rapp, 1974). Included among this class is the Epstein–Barr virus, a causative agent in the development of Burkett B-cell lymphomas and in nasopharengyal carcinoma (de-Thé et al., 1978), and is mediated by genes encoding EBNA-1, EBNA-2, and latent membrane protein (Fahreus et al., 1990). Adenoviruses affect host and non-host cells differently; in host cells, infection causes lysis whereas in non-host cells, infection results in transformation via E1A and E1B genes (Pettersson and Roberts, 1986). Hepatitis B viruses specifically infect the liver of ducks, woodchucks, and squirrels, and are strongly associated with liver cancer development in humans; infected individuals exhibit a >100-fold risk for cancer development (Synder et al., 1982). Hepatitis B gene X is thus far the main gene implicated in the transformation process (Kim et al., 1991). Poxviruses (eg, Yaba monkey virus, Shope fibroma virus), replicate in the cytoplasm of infected cells and produce benign tumors (Niven et al., 1961).

Proto-Oncogenes

An oncogene is a gene encoding a protein that is capable of transforming cells in culture or inducing cancer in animals. Of the known oncogenes, the majority appear to have been derived from normal genes (ie, proto-oncogenes), and are involved in cell signaling cascades (Table 8-17). Because most proto-oncogenes are essential for maintaining viability, they are highly conserved evolutionarily. It has been clearly demonstrated that altered expression of these genes results in unregulated control of cell growth. Activation of proto-oncogenes arises through mutational events occurring within proto-oncogenes (Alitalo and Schwab, 1986; Bos, 1989). It has been recognized that a number of chemical carcinogens are capable of inducing mutations in proto-oncogenes (Balmain and Pragnell, 1983).

Several oncogene products are associated with growth factors. The sis oncogene encodes for platelet-derived growth factor (PDGF), which consists of type A and B chains. Active PDGF can consist of homo- or heterodimers of the A and B chains. Only cells that express the PDGF receptor are susceptible to transformation by the sis oncogene (Dooliter et al., 1983). Fibroblast growth factor (FGF), epidermal growth factor (EGF), and the wnt family also have oncogenes that encode for their production (Burgess et al., 1989; Nusse and Varmus, 1992). Enhanced expression of these growth factor genes leads to transformation of cultured cells. In addition, lung carcinomas and astrocytomas in humans express both PDGF and its receptor.

In order for the growth factors listed above to induce cell growth, the cell must also express the appropriate growth factor receptors. Ligand binding to receptors in many cases results in the activation of signaling cascades, of which a number have protein kinase activity. Protein kinases can phosphorylate at tyrosine, serine, or threonine residues. Several oncogene products have been identified that have tyrosine kinase activity (eg, src, fps, fgr, fms, kit, and ros; Collett and Erikson, 1978). Growth factor receptors including EGF, PDGF, insulin, insulin-like growth factor, and FGF have tyrosine activity (Wells, 1999). An extensive number of protein kinases are serine or threonine kinases, some of which can be activated in tumor cells and lead to transformation. The protein kinase C family consists of serine/threonine kinases (Nishizuka, 1988). This family of proteins participates in the inositol–phospholipid second messenger pathway and can be activated by calcium and diacylglycerol, and phorbol ester tumor promoters. The raf and mos oncogenes are other example of serine threonine kinases, that interact with mitogen activated kinase pathways, leading to enhance the expression of “immediate-early” genes such as Myc, and Fos/Jun heterodimers, and ultimately leads to enhanced cell growth (Su and Karin, 1996).

Guanine nucleotide binding proteins is a subset of the Guanosine-5′-triphosphate (GTP) superfamily. The primary oncogenic members of this family are the ras genes (eg, Ha-, Ki-, and n-ras), which were initially found in murine retroviruses, and are frequently activated in human tumors and in chemically induced tumors in rodent models (McCormick, 1994). The three ras forms differ by only 20 amino acids and have a conserved cytosine-186, which is a site for posttranslational modification by a farnesyl, isoprenyl group, a modification that is needed for membrane localization (Hancock et al., 1989). Mutations in ras affect the association of the Ras with GTP, and because Ras is active when GTP is bound, constitutive Ras activation is often seen with mutated forms of ras. Many mutation sites are associated with activation of ras and include codons 12, 13, 61, 116, 117, 119, and 146 (de Vos et al., 1988). Although a dominant gain-of-function mutation in the ras gene converts it into an oncogene, a recessive loss-of-function mutation in the NF1 gene also leads to constitutive Ras activation. NF1 encodes a GTPase-activating enzyme (GAP) that accelerates hydrolysis of GTP and the conversion of active GTP-bound Ras to inactive GDP-bound Ras. The loss of GAP leads to sustained Ras activation of downstream signal transduction proteins. Individuals with neurofibromatosis have inherited a single mutant NF1 allele; subsequent somatic mutation in the other allele leads to formation of neurofibromas.

In addition to acting at the receptor level, some oncogene products are nuclear transcription factors and thus cause alteration of gene expression. One of the first to be identified was the erbA oncogene product, which is an altered form of the thyroid hormone receptor (Damm et al., 1989). This altered form is unable to bind thyroid hormone and allows the oncogene protein to act in a dominant-negative manner, functioning as a constitutive repressor of normal expression of genes regulated by thyroid hormone. Another family of oncogenes encode for the AP-1 transcription factor, a complex consisting of jun and fos, as well as Jun family members with homology to c-jun (Jun B and Jun D) (Chiu et al., 1988; Vogt, 2001). Phorbol esters and several other tumor-promoting compounds are able to rapidly induce gene expression of both fos and jun. Activation of both fos and jun results in the constitutive expression of the transcription factors and sustained stimulation of cell growth signals. The myc family (c-myc, N-myc, and L-myc) encode for transcription factors that are found activated in a number of tumor types (Marcu et al., 1992).

Tumor-Suppressor Genes

Retinoblastoma (Rb) Gene

In contrast to oncogenes, the proteins encoded by most tumor-suppressor genes act as inhibitors of cell proliferation or cell survival (Table 8-18). The prototype tumor-suppressor gene, Rb, was identified by studies of inheritance of retinoblastoma. Loss or mutational inactivation of Rb contributes to the development of a wide variety of human cancers. In its unphosphorylated form, Rb binds to the E2F transcription factors preventing E2F-mediated transcriptional activation of a number of genes whose products are required for DNA synthesis. In addition, the Rb–E2F complex acts as a transcriptional repressor for many of these same genes (Chellappan et al., 1991). During the early G₁ phase of the cell cycle, Rb is an unphosphorylated state, but becomes phosphorylated during late G₁, causing dissociation from E2F—a process that allows E2F to induce synthesis of DNA replication enzymes, resulting in a commitment into the cell cycle. Rb phosphorylation is initiated by an active Cdk4–cyclin D complex and is completed by other cyclin-dependent kinases (Weinberg, 1995; Sherr, 1993). Most tumors contain an oncogenic mutation of one of the genes in this pathway such that cells enter the S phase of the cell cycle in the absence of the proper extracellular growth signals that regulate Cdk activity. In addition, Rb is bound and inactivated by the E1A protein of the adenoviruses (Whyte et al., 1988) and also by the large T antigen of SV40 and the E7 protein of the papilloma virus, suggesting that the elimination of Rb function may be an important mechanism in tumor development by viruses and/or mutations induced by chemical carcinogens.

p53 Gene

The p53 protein is a tumor-suppressor gene that is essential for the checkpoint control and arrests the cell cycle in cells with damaged DNA in G₁. Cells with functional p53 arrest in G₁ when exposed to DNA-damaging agents such as irradiation, whereas cells lacking functional p53 are unable to block the cell cycle. p53 is activated by a wide array of stressors including UV light, γ irradiation, heat, and several carcinogens. DNA damage by γ irradiation or by other stresses leads to the activation of certain kinases, including ataxia telangiectasia mutated (ATM), which is encoded by the gene mutated in ataxia telangiectasia and a DNA-dependent protein kinase. Phosphorylation of p53 by these and other kinases results in stabilization of p53 and an increase in cell content of this protein (Finlay et al., 1989). The active form of p53 is a tetramer of four identical subunits. A missense point mutation in one p53 of the two alleles in a cell can abrogate almost all p53 activity because virtually all the oligomers will contain at least one defective subunit and such oligomers cannot function as a transcription factor. Oncogenic p53 mutations thus act as “dominant negatives,” in contrast to tumor-suppressor genes such as Rb.

Virtually all p53 mutations abolish its ability to bind to specific DNA sequences and activate gene expression (El-Deiry et al., 1992). The transcription of cyclin-kinase inhibitor p21, which binds to and inhibits mammalian G₁ Cdk–cyclin complexes, is induced by p53. As a result, cells with damaged DNA are arrested in G₁ until the damage is repaired and the levels of p53 and p21 fall; the cells then can progress into the S phase. In most cells, accumulation of p53 also leads to induction of proteins that promote apoptosis, and, therefore, would prevent proliferation of cells that are likely to accumulate multiple mutations. When the p53 checkpoint control does not operate properly, damaged DNA can replicate, producing mutations and DNA rearrangements that contribute to the development of transformed cells. Proteins that interact with and regulate p53 are also altered in many human tumors. The gene encoding one such protein, MDM2, is amplified in many sarcomas and other human tumors that maintain functional p53 (Leach et al., 1993). Under normal conditions, MDM2 binds the N-terminus of p53, repressing the ability of p53 to activate transcription of p21 and genes mediating p53 degradation. Thus, MDM2 normally inhibits the ability of p53 to block at G₁ or induce apoptosis. Enhanced MDM2 levels in tumor cells therefore results in decreased functional p53 and prevents p53 from growth arrest in response to irradiation or chemical carcinogens. For example, benzo(a)pyrene produces inactivating mutations at codons 175, 248, and 273 of the p53 gene in cultured cells; these same positions are mutational hot spots in human lung cancer (Hollstein et al., 1991). Aflatoxin exposure induces a G to T transversion at codon 249 of p53, and methyl-N′-nitro-N-nitrosoguanidine (MNNG) can alter methylation of DNA leading to inactivation of p53 (Bressac et al., 1991).

BRCA1 Gene

Genetic analysis of breast tumors has revealed a hereditary predisposition for breast cancer linked to BRCA1, a tumor-suppressor gene. Mutation of a single BRCA1 allele results in a 60% probability of developing breast cancer by age 50. In families transmitting the mutant BRCA1 allele, the normal allele is often absent in tumors. A number of investigators have shown that germ line mutations lead to loss of function of the BRCA1 gene, perhaps by acting as a transcription factor through binding at a zinc finger domain. However, no mutations have been observed in sporadic breast cancers (Futreal et al., 1994), suggesting that BRCA1 gene silencing may occur through nonmutational mechanisms.

Wilms Tumor Gene (WT1)

Wilms tumor occurs in the developing kidney at a rate of approximately one per 10,000 children. The WT1 gene is believed to be responsible for tumor development and is thought to function as a transcription factor (Hastie, 1994). The DNA-binding region of the WT1 gene contains four zinc finger domains, and binds a consensus sequence shared with EGR1; however, WT1 acts as a transcriptional repressor at this site (Rauscher et al., 1990). Similar to p53, WT1 mutant alleles act in a dominant-negative manner.

p16 Gene

The group of proteins that function as cyclin-kinase inhibitors play an important role in cell cycle regulation. Mutations, especially deletions of the p16 gene, that inactivate the ability of p16 to inhibit cyclin D–dependent kinase activity are common in several human cancers including a high percentage of melanomas (Kamb et al., 1994). Loss of p16 would mimic cyclin D1 overexpression, leading to Rb hyperphosphorylation and release of active E2F transcription factor. Thus, p16 normally acts as a tumor suppressor. As with the BRCA1 gene, relatively few mutations have been found in this gene, and some researchers have speculated that epigenetic mechanisms such as gene silencing by DNA methylation may occur during tumorigenesis (Merlo et al., 1995).

Hormesis, Dose Response, and Carcinogenesis

Hormesis is defined as a dose–response curve in which a U, J, or inverted U-shaped dose–response is observed, with low-dose exposures often resulting in beneficial rather than harmful effects (Calabrese, 2002) (Fig. 8-27). One of the first reports establishing a hormetic response was observed with ionizing radiation, in which it was hypothesized to be due to adaptation to background radiation exposure, as well as enhanced metabolism and detoxification (Calabrese, 2002; Parsons, 2003). Numerous examples exist that provide evidence for U- and J-shape dose relationships to different biomarkers of carcinogenesis, both for initiating and promoting carcinogens. For example, low-dose styrene treatment resulted in a decrease of chromosomal aberrations (Camurri et al., 1983). A reduction in DNA strand breaks in human keratinocytes was seen following low-dose exposure to N-methyl-N′-nitro-N-nitrosoguanidine, whereas high doses were greatly elevated compared to controls (Kleczkowska and Althaus, 1996). Several chronic bioassays for carcinogenicity in rats and mice have demonstrated a negative correlation between proliferative hepatocellular lesions and lymphomas at low- and medium-dose levels (Young and Gries, 1984). TCDD at hepatocarcinogenic doses resulted in a dose-dependent reduction in mammary and uterine tumors (Kociba et al., 1978). TCDD also reduces tumor incidence after exposure to low doses of radiation and selenium (Pollycove et al., 2001; Nordberg et al., 1981). Evidence for a U-shaped dose–response curve has also been demonstrated for phenobarbital (Kitano et al., 1998; Ito et al., 2003; Kinoshita et al., 2003). These investigators have shown that phenobarbital promoted the growth of diethylnitrosamine-induced hepatic lesions at doses between 60 and 500 ppm, whereas doses between one and 7.5 ppm decreased lesion growth, hepatocellular carcinoma, and tumor multiplicity.

Adaptive responses have been proposed to explain the hormetic effects observed by chemical carcinogens. When experimental animals are exposed to chemicals, the initial response is an adaptive response to maintain homeostasis (Calabrese, 2002). Adaptive responses usually involve actions of the chemical on cellular signaling pathways that lead to changes in gene expression, resulting in enhanced detoxification and excretion of the chemical as well as preserving the cell cycle and programmed cell death. It has been proposed that following very low doses of chemicals, the upregulation of these mechanisms overcompensates for cell injury such that a reduction in tumor promotion and/or tumor development is seen, and would explain the U- or J-shaped response curves obtained following carcinogen exposure.

A common feature of chemical carcinogens for which hormetic effects have been proposed is the formation of reactive oxygen species and the induction of cytochrome P450 isoenzymes. Inhibition of carcinogenesis by low levels of phenobarbital has been postulated to involve the suppression of 8-OHdG generation and cellular proliferation within areas of glutathione-S-transferase pi (GST-P) positive foci, possibly related to enhanced expression of the gene encoding the enzyme oxoguanine glycosylase one (Ogg1), which is responsible for the repair of 8-OHdG lesions (Kinoshita et al., 2003). Another explanation for the suppressive effect of low-level phenobarbital on the development of preneoplastic lesions involves the stimulation of hepatic drug metabolizing enzymes, which detoxify carcinogens (Pitot et al., 1987). In a low-dose phenobarbital study, cDNA microarray analysis resulted in a suppression of mRNA expression of MAP kinase p38, JNK1, 2, and other intracellular kinases (Kinoshita et al., 2003). Thus, suppression of signal transduction modulators appears to be involved in the inhibitory effect of phenobarbital on cell proliferation.

Chemoprevention

The study of chemicals that prevent, inhibit, or slow down the process of cancer is referred to as chemoprevention. A number of chemicals, including drugs, antioxidants, foodstuffs, and vitamins have been found to inhibit or retard components of the cancer process (Table 8-19) in both in vitro and in vivo models (Stoner et al., 1997). Cancer chemopreventive chemicals may function at one or more of the steps in the carcinogenesis process. Cancer chemoprevention involves the administration or inclusion in the diet of natural or synthetic chemicals in an attempt to prevent, halt, or reverse the process of carcinogenesis. A basic assumption in chemoprevention is that treating early stages of the malignant process will halt or delay the progression to neoplasia. Chemopreventive chemicals may function via one or several mechanisms; inhibitors of carcinogen formation, blocking agents, and/or as suppressing agents (Table 8-19). Blocking agents serve to prevent the metabolic activation of genotoxic or nongenotoxic carcinogens by either inhibiting its metabolism or enhancing the detoxification mechanisms. In the case of reactive oxygen species generation by a carcinogen, antioxidants, such as, N-acetylcysteine, or vitamins E, A, and C serve as blocking agents by trapping the oxidative species before it can induce cellular damage. The drug oltipraz induces phase two enzymes; in particular, glutathione-S-transferase has been shown to reduce the carcinogenicity of aflatoxin B1 by enhancing its detoxification (Jacobson et al., 1997). Chemopreventive chemicals increase tissue resistance function on the target tissue, usually an early preneoplastic lesion by increasing tissue maturation and/or decreasing cell proliferation. Suppressing agents induce tissue differentiation, may counteract oncogenes, enhance tumor-suppressor gene activities, inhibit proliferation of premalignant cells, or modify the effect of the carcinogen on the target tissue. Retinoic acid is a classic agent that functions by inducing target tissue differentiation. Similarly, viral vectors containing wild-type p53 have been shown to negate the effects of tumor growth driven via p53 mutation. Extensive experimental evidence exists showing the inhibition or retarding of the cancer process by chemopreventive chemicals. These studies have been performed in animal models and cell systems. Although the human results of chemoprevention interventions are less convincing than the experimental models, considerable epidemiological evidence demonstrates an inverse relationship of some nutrient factors with cancer risks.

A number of in vivo and in vitro experimental systems are available to assess the potential carcinogenicity of chemicals. The types of tests available to identify chemicals with carcinogenic potential can be classified into general categories, based on the duration required to conduct the test (Table 8-20). Short-term tests are typically of the duration of days to a few weeks, intermediate-term tests last from weeks up to a year, whereas chronic long-term tests usually encompass six months to two years exposure to a chemical. These bioassays use bacterial and mammalian targets.

Short-Term Tests for Mutagenicity

Short-term tests for mutagenicity were developed to identify potentially carcinogenic chemicals based on their ability to induce mutations in DNA either in vivo or in vitro. A variety of in vivo and in vitro short-term tests are available to test the potential carcinogenicity of a chemical (Table 8-21). The majority of these tests measure the mutagenicity of chemicals as a surrogate for carcinogenicity. Therefore, although they are usually very predictive of indirect acting and direct acting (if a metabolic source is provided), these tests routinely fail to detect nongenotoxic carcinogens.

In Vitro Gene Mutation Assays

The most widely used short-term test is the Ames assay (Ames et al., 1975; Mortelmans and Zeiger, 2000). The relative simplicity and low cost of the test make it a valuable screening tool for mutagenic carcinogens. Salmonella typhimurium strains, deficient in DNA repair and unable to synthesize histidine, are used. In the presence of a mutagenic chemical, the defective histidine gene can be mutated back to a functional state (back mutation), resulting in a restoration of bacterial growth in medium lacking histidine. The mutant colonies, which can make histidine, are referred to as “revertants.” The Ames test in basic form can detect direct but not indirect genotoxic carcinogens. With the inclusion of a metabolic source, specifically the 9000g supernatant (S9) of a rat liver homogenate to promote metabolic conversion of the chemical, the Ames can also detect indirect-acting genotoxic carcinogens. Fig. 8-28 describes the standard method used for performing the Ames assay.

Genetically unique strains of the S typhimurium bacterium have been developed for determining specific mutational targets. Strains TA100 and TA1535 are able to detect of point mutations whereas strains TA98, TA1537, and TA1538 are able to detect frame shift mutations. Chemicals are typically tested at several concentrations (usually five or more) and the mutation frequency (number of revertants) is calculated. Activation-independent (eg, sodium azide, methyl methanesulfonate) and activation-dependent (eg, 2-aminoanthracene) positive controls are included in each assay.

The mouse lymphoma assay is a mutagenicity assay used to determine whether a chemical is capable of inducing mutation in eukaryotic cells. Typically, mouse lymphoma L5178Y cells are used, and the ability of the cell cultures to acquire resistance to trifluorothymidine (the result of forward mutation at the thymidine kinase [TK] locus) is quantified. Another mammalian cell mutation assay, the Chinese hamster ovary (CHO) test, is also commonly used to assess the potential mutagenicity of chemicals. This assay uses the hypoxanthine–guanine phosphoribosyltransferase (HGPRT) gene as the endpoint. The basis for these assays is shown in Fig. 8-29. Cells are treated with the test chemical and then placed into suspension with selective medium for replication and fixation of induced mutations. Cells are then plated for colony growth, and after several days, colony numbers and colony size are recorded. The number of mutant colonies is a measure of the ability of the test chemical to induce a genetic change at the TK or HGPRT loci in these transformed cells.

As with the Ames assay, these assays are frequently performed in the presence of an exogenous metabolic source (eg, irradiated epithelial cell feeder layer). Because not all carcinogens are mutagens and/or directly damage DNA, the concordance with the chronic in vivo bioassay for these mutagenicity assays is relatively low. For a historical perspective on the utilization of the Ames/Salmonella assay, a recent review by Claxton et al. (2010) provides more information on the use of the assay and the role of metabolic activation and mutagenicity testing with respect to genotoxic carcinogens.

In Vivo Gene Mutation Assays

For the assessment of the carcinogenic potential of chemicals in vivo genotoxicity tests are important complement to in vitro mutagenicity tests. The in vivo tests have advantages over the in vitro test systems in that they take into account whole animal processes such as absorption, tissue distribution, metabolism, and excretion of chemicals and their metabolites. The commonly used in vivo models include transgenic rodent mutation assay systems based on the genes of the lac operon, MutaMouse (Gossen et al., 1989) and Big Blue (Kohler et al., 1991a,b). Some details of the lac operon are shown in Fig. 8-30).

The MutaMouse is a transgenic mouse in which a vector prepared from bacteriophage λDNA (λgt10) has been stably inserted into the genome and exists within every cell. Within the λDNA sequence, the bacterial gene lacZ, which encodes for β-galactosidase, was inserted at the single EcoRI restriction site. The entire vector contains ~47,000 bp, of which the lacZ gene comprises 3126 bp. The construction and analysis of MutaMouse have been previously described (Gossen et al., 1989). The Big Blue transgenic mouse was developed using a 47.6-kb λshuttle vector containing the bacterial lacI gene (1080 bp), which encodes the repressor protein of the lacZ gene. As with the MutaMouse, this vector is stably inserted into the genome and exists in every cell. Details on the Big Blue system are described elsewhere (Kohler et al., 1991a,b).

To detect mutations, following exposure of mice to test chemicals, mutations are analyzed in high molecular DNA isolated from the tissue under investigation. The next step is to isolate a viable λphage for the analysis of mutations in the lacZ or lacI genes from the mouse. The infectious λparticles are detected as plaques that form on a bacterial (Escherichia coli strain) lawn grown on an agar surface. In the case of MutaMouse a positive selection system is used that involves the scoring of clear plaques on titration plates to determine the total number of plaque forming units and on selection plates to determine the number of mutants (Gossen and Vijg, 1993). For Big Blue, a nonselectable color screening assay is used, scoring blue mutant plaques among the nonmutant clear plaques (Kohler et al., 1991a). In each case the ratio of mutants to the total population provides a mutation frequency for each chemical and each organ tested. As with the other short-term tests that have been discussed, it is unlikely that in vivo genotoxicity test systems will identify nongenotoxic/non–DNA-reactive compounds.

Chromosomal Alterations

Chromosomal alterations are quite common in malignant neoplasms, as such, the detection of chromosomal abnormalities by test chemicals is considered an excellent test for the assessment of carcinogenic potential. Both in vivo and in vitro assays are available to assess chromosomal alterations. In mammalian cell lines, most of the test systems used the same lines as used in the mutation assay. To assess induction of chromosomal alterations, cells are harvested in their first mitotic division after the initiation of chemical exposure. Cells are stained with Giemsa, and scored for completeness of karyotype (21 ± 2 chromosomes). The classes of aberrations recorded include breaks and terminal deletions, rearrangements and translocations, as well as despiralized chromosomes, and cells containing 10 or more aberrations.

SCEs are a measure of DNA damage and increased levels of DNA damage, events that are associated with mutation induction and cancer. During metaphase, sister chromatids, each encompassing a complete copy of one chromosome, are bound together through specific protein interactions. SCEs are a reflection of an interchange of DNA between different chromatids at homologous loci within a replicating chromosome (Latt et al., 1981). Differential staining of chromatids is therefore used to assess SCEs. Typically, 5-bromodeoxyuridine (BrdU) is added along with the test chemical for one cell replication followed by replacement of the medium with medium containing BrdU and colcemid, without test chemical. Cells are then harvested, stained and second-division metaphase cells scored to determine the frequency of SCE/cell for each dose level.

In vivo, analysis of chromosomal alterations has been described since the 1980s (Sasaki, 1980). Disruption of the DNA replication process or damage to chromosomes by chemicals can alter the genetic material distributed to either of the two daughter nuclei. When this occurs, the genetic material that is not incorporated into a new nucleus may form its own “micronucleus,” which is clearly visible with a microscope. For this assay, animals are exposed to chemicals and the frequency of micronucleated cells is determined at some specified time after treatment. Micronucleus tests must be performed on cells that are dividing, most typically in cells from bone marrow samples (Heddle et al., 1983). As with other short-term tests, the micronucleus test provides complementary information to gene mutation data but neither tests enhance the ability to detect genotoxic carcinogens, nor is capable of detecting non–DNA-reactive carcinogens. However, this assay does provide information on the ability of a chemical to disrupt mammalian chromosome structure and function.

DNA Damage

Primary DNA damage represents possible premutational events that can be detected, either directly or indirectly, by a number of techniques using mammalian cells in culture or using rodent tissue. Unscheduled DNA synthesis (UDS) is a commonly used assay that measures the ability of a test article to induce DNA lesions by measuring the increase in DNA repair. Among the available techniques is the measurement of DNA strand breaks both in vivo and in vitro (Miyamae et al., 1997). Although a wide array of assays are available to assess DNA damage, the Comet assay, or single cell gel electrophoresis, has become one of the standard methods for assessing DNA damage and repair at the individual cell level. Since its introduction 1988, the alkaline comet assay has been modified to increase the utility of this technique for detecting various forms of DNA damage (eg, single- and double-strand breaks, oxidative DNA base damage, and DNA–DNA/DNA–protein/DNA–Drug cross-linking) and DNA repair in virtually any eukaryotic cell. Through the use of lesion-specific endonucleases, the levels of UV-induced pyrimidine dimers, oxidized bases, and alkylation damage can be assessed in biological samples (Collins, 2004).

Short-Term Tests: Transformation Assays

A variety of in vitro test systems have been developed to assess the carcinogenic potential of chemicals. The C3H/10T½ cell line has been widely used for the transformation assay. It was originally derived from fibroblasts taken from the prostate of a C3H mouse embryo. The cells are approximately tetraploid but the chromosome number in the cells varies widely. As such, these cells are chromosomally abnormal and have already passed through some of the stages that might be involved in the production of a cancerous cell. Upon plating these cells, they will stop growing when their density is sufficiently high (contact growth inhibition). However, the contact inhibition can fail, resulting in cell piling forming a transformed colony. Therefore, following exposure to xenobiotics, this assay assesses carcinogenic potential based on the percentage of colonies that are transformed (Reznikoff et al., 1973).

The most frequently used endpoint for cell transformation is morphological transformation of mammalian cell fibroblasts in culture. Transformation assays using BALB3T3 and Syrian hamster embryo (SHE) cells are available for the assessment of the carcinogenic potential of chemicals. BALB/3T3 transformation assay, however, has not gained full acceptance as a carcinogenic screening assay mainly due to issues regarding its relatively low sensitivity, low reproducibility and relatively long test period (Sakai and Sato, 1989; Schechtman, 1985). The SHE cell assay, a diploid cell transformation assay, measures carcinogenic potential of xenobiotics by assessing transformed colonies based on morphological criterion. The SHE cell assay has several features that make this an appealing method to evaluate the carcinogenic potential of xenobiotics: (1) The assay has a >85% concordance with the two-year rodent bioassay; (2) the stages involved in the clonal transformation of SHE cells (eg, morphological transformation, immortalization, and tumorogenicity) closely resemble those associated with the classic defined stages of carcinogenesis (eg, initiation, promotion and progression); and (3) non-genotoxic/non–DNA-reactive carcinogens elicit a positive response on morphological transformation in the SHE cell assay (Barrett and Lamb., 1985; LeBoeuf et al., 1990; Isfort et al., 1994, 1996). The SHE assay offers an alternative, regulatory approved, means to check the validity of the positive result before embarking on potentially unnecessary and expensive chronic or subchronic testing protocols.

Chronic Testing for Carcinogenicity

The majority of in vivo carcinogenicity testing is performed in rodent models. The administration of chemicals in the diet, often for extended periods, for assessment of their safety and/or toxicity began in the 1930s (Sasaki and Yoshida, 1935). Animal testing today remains a standard approach for determining the potential carcinogenic activity of xenobiotics. In addition to the lifetime exposure rodent models, organ-specific model systems, multistage models, and transgenic models are being developed and used in carcinogen testing (Table 8-22).

Chronic (Two Year) Bioassay

Two-year studies in laboratory rodents remain the primary method by which chemicals or physical agents are identified as having the potential to be hazardous to humans. The most common rodents used are the rat and mouse. Typically the bioassays are conducted over the lifespan of the rodents (two years). Historically, selective rodent strains have been used in the chronic bioassay; however, each strain has both advantages and disadvantages. For example, the NTP typically uses Fisher 344 (F344) rats and B6C3F1 mice. The F344 rat has a high incidence of testicular tumors and leukemias whereas the B6C3F1 mouse is associated with a high background of liver tumors (Table 8-23).

In the chronic bioassay, two or three dose levels of a test chemical and a vehicle control are administered to 50 males and 50 females (mice and rats), beginning at eight weeks of age, continuing throughout their lifespan. The route of administration can be via oral (gavage), dietary (mixed in feed), or inhalation (via inhalation chambers) exposure. Typically a number of short-term in vivo tests are conducted prior to the chronic bioassay to determine acute toxicity profiles, appropriate route of administration, and the maximum tolerated dose (MTD). Generally, the MTD is used to set the high dose in the chronic study. The use of the MTD as the upper dose level has been questioned by many investigators, as it is recognized that the doses selected represent those that are considered unrealistically high for human exposure. Pharmacokinetics and metabolism at high dose are frequently unrepresentative of those at lower doses; in addition, a general relationship between toxicity and carcinogenicity cannot be drawn for all classes of chemicals. During the study, food consumption and bodyweight gain should be monitored and the animals observed clinically on a regular basis; at necropsy, the tumor number, location, and diagnosis for each animal is thoroughly assessed.

Organ-Specific Bioassays and Multistage Animal Models

Many tissue-specific bioassays have been developed with the underlying goal being to produce a sensitive and reliable assay that could be conduced in a time frame shorter in duration than the two-year chronic bioassay. These assays are commonly used to detect carcinogenic activity of chemicals in various target organs (Weisburger and Williams, 1984). Of the many models available, three models, the liver, skin, and lung models are more widely used.

Carcinogenicity Testing in the Liver

The liver represents a major target organ for chemical carcinogens. It has been estimated that nearly half of the chemicals tested in the two-year chronic bioassay by the NTP showed an increased incidence of liver cancer. The rodent liver has been used as an animal model for carcinogenesis since the 1930s. Early pioneering work by Peraino and Pitot as well as Farber showed the multistaged process that occurs in the liver. The multistage nature of carcinogenesis is paralleled in the animal models; the system is characterized by well-defined changes including the formation of initiated cells by genotoxic agents that then progress to preneoplastic focal lesions, which subsequently convert into neoplasms. The use of preneoplastic lesions as endpoints in carcinogenicity testing may shorten experiment from two years to a few months. Several rodent liver focus assays have been developed to assess the ability of a chemical to induce liver cancer and study the mechanisms involved in tumor development (Bannasch, 1986a,b; Williams, 1982). Liver carcinogenesis assays have been developed to study and distinguish chemicals that affect the initiation or promotion stage of hepatocarcinogenesis. During the assay for initiating activity of a chemical, the test substance is given prior to exposure to a promoting chemical. Although a single initiating dose can result in the induction of focal lesions, exposure over a several week period is often used to increase the sensitivity of the model (Parnell et al., 1988; Williams, 1982). Phenobarbital is a commonly used tumor promoter; however, a wide range of chemicals have also been used as promoting agents (Solt and Farber, 1976; Oesterle and Deml, 1988). To assess the promoting activity of a chemical, the liver is initiated with a genotoxic chemical, often diethylnitrosamine. The test chemical is then administered for a duration of weeks to several months, and chemicals with promoting activity may stimulate the proliferation of initiated cells or may inhibit the proliferation of the surrounding putatively normal cells. The dose of the initiating carcinogen should represent a dose that will not induce liver tumors during the course of the experiment.

Another method commonly used was developed in Japan by Ito and co-workers (Ogiso et al., 1990; Shirai et al., 1999; Ito et al., 1994). The entire assay takes only eight weeks to perform. Rats are initiated with a single dose of diethylnitrosamine, followed by a two-week recovery period, after which point the animals are exposed to the test compound for eight weeks. After one week of exposure to the test substance, the animals are given a two-third partial hepatectomy. The control group receives the same initiation and partial hepatectomy, but is not exposed to the test chemical. Hepatic focal lesions, while individually are clonal in nature, express a number of phenotypic alterations in various enzyme markers. A common endpoint assessed is the formation of liver lesions that stain for GST-P, a marker that stains many focal lesions in the rat. Using this assay, these investigators demonstrated a significant correlation between the results obtained using this assay and medium- and long-term study results (Ogiso et al., 1990). This group has also modified the original procedure to enable the detection of promoting agents. In this protocol, carcinogens are given over a four-week period to initiate the formation of focal lesion, after which, test chemicals are administered for an additional 24- to 36-week period (Ito et al., 1996). In this manner, the ability of the test chemical to promote the growth of preneoplastic lesions can be assessed. The newborn mouse model originally described by Shubik and co-workers (Pietra et al., 1959) has also been used as a model for hepatocarcinogenesis (Vesselinovitch et al., 1978; Fujii, 1991). In this model, a single dose of diethylnitrosamine is administered to infant mice to initiate focal lesions. This step is then followed by exposure to test chemicals for several weeks to assess their potential to promote focal lesion development in the liver.

Identification of hepatic foci in H&E-stained sections is regarded as the most reliable approach for the diagnosis and quantification of preneoplastic liver lesions in rodents. Preneoplastic lesions are obligatory precursor lesions that can lead to liver tumors and will progress to benign and malignant liver cell tumors without further chemical exposure, and are used as endpoints in carcinogenicity testing (Ito et al., 1989; Maronpot et al., 1989; Pereira and Herren-Freund, 1988; Pitot et al., 1987). In addition to the sensitive detection of these preneoplastic lesions in conventional H&E staining, a number of histochemically detectable phenotypic alterations have been used for their quantification; however, these markers are only useful in the rat model, as focal lesions in mice to not exhibit these same phenotypic markers.

Carcinogenicity Testing in the Skin

The Mouse Skin model is one of the most extensively studied and used models for understanding multistage carcinogenesis. This model of carcinogenesis is an assay that has been used to dissect mechanisms of carcinogenesis and is also purported to be a useful intermediate-term cancer bioassay. The skin was the target organ of the first experimental induction of chemical carcinogenesis (Yamagiwa and Ichikawa, 1915). The early studies by Friedwald and Rous (1944) and Berenblub and Shubik (1947) introduced the two-stage concept of carcinogenesis in the skin (Fig. 8-31). This model exploits many of the unique properties of the mouse skin, one major being that the development of neoplasia can be followed visually. In addition, the number and relative size of papillomas and carcinomas can be quantified as the tumors progress.

Both initiating and promoting activities of chemical carcinogens can be assessed using this model. In the promotion assay, a number of chemical carcinogens have been used to initiate cells in the mouse skin including urethane, UV light, benzo(a)pyrene, and dimethlybenzanthracene, with the latter of more common usage. The requirement for all initiating agents is to induce a genotoxic event that upon failure to repair DNA damage, results in the formation of a mutated cell. Grossly, initiated cells of the skin appear identical to normal skin. Initiation in skin is frequently linked with the mutation of the CHr gene. Because the terminally differentiated cells in the skin are no longer capable of undergoing cell division, only initiated cells retain their proliferative capacity and thus represent the cell populations that give rise to tumors. To assess promotion by a chemical, an initiating chemical is applied first and is followed by the administration of a test substance for several weeks on the shaved skin of mice (Slaga, 1984). The promotion of initiated karatinocytes is commonly assessed using the phorbol ester 12-O-tetradecanoylphorbol-13-acetate (TPA), which is routinely included as a positive control in this assay. The current hypothesis is that during the initiation stage the expansion or of initiated cells occurs as a result of inflammation and hyperplasia from either TPA or through mechanical wound healing mechanisms. Upon repeated application of tumor promoters, selective clonal expansion of initiated keratynocytes occurs, resulting in skin papillomas, which over time can progress to carcinomas. In the standard 2-stage protocol for mouse skin malignant progression is relatively rare with approximately 5% of the paillomas progressing to the carcinoma stage. For the detection of initiating activity, the test substance is applied to skin prior to promotion with phorbol esters. Several mouse models are available, including hairless mice, SENCAR mice, both of which have enhanced sensitivity to the induction of skin cancer (Brown and Balmain, 1995; Sundberg et al., 1997).

Carcinogenicity Testing in the Lung

Strain A mice are genetically susceptible to the development of lung tumors, with lung tumors being observed in control animals as early as three to four weeks of age with a steady increase to nearly 100% by 24 months of age (Shimkin and Stoner, 1975). Chemically induced tumors appear to be derived from hyperplastic lesions that progress to adenoma, carcinoma within adenoma, and ultimately, to carcinomas (Stoner et al., 1993). In this model, carcinogenicity is typically assessed as an acceleration of tumor development rather than an increase in tumor incidence. The protocol currently used is that the chemical is administered for a period of eight weeks, after which the animals remain on test for four additional months without chemical exposure. The strain A mouse lung tumor assay is sensitive to particular classes of chemicals, such as PAHs, nitrosamines, nitrosoureas, carbamates, aflatoxin, certain metals, and hydrazines (Maronpot et al., 1986; Stoner, 1991; Stoner and Shimkin, 1985).

Carcinogenicity Testing in Other Organs

Test systems to examine the ability of a chemical to promote neoplastic development at organ sites other than liver, skin, and lung have also been developed. The available systems include animal models directed at examining carcinogenicity in the kidney, bladder, pancreas, stomach, colon, small intestine, and oral cavity. These models vary in the initiating carcinogen used, and frequency, duration, and site of application, as well as duration of promoting chemical exposure. Table 8-24 provides an overview of the animal models available for these target organs.

Transgenic Animals in Carcinogenicity Assessment

Due to the development of animal models with genetic alterations that invoke a susceptibility to carcinogenesis by chemical agents, the use of transgenic and knockout animals in carcinogenicity assessment is gaining more popularity. The common models that have been used include the Tg.AC and rasH2 transgenic mice and p53^+/− and XPA^−/− knockout mice (Gulezian et al., 2000). Recently, the feasibility of the use of these animal models as alternative assays for the two-year chronic bioassay was assessed in a collaborative between the Health and Environmental Sciences Institute (HESI) branch of the International Life Sciences Institute (ILSI). In this assessment, 21 chemicals were evaluated, encompassing genotoxic, nongenotoxic, and noncarcinogenic chemicals. The conclusions drawn from the scientific review suggested that these models appear to have usefulness as screening models for assessment of chemical carcinogenicity; however, they do not provide definitive proof of potential human carcinogenicity. Further the scientific panel suggested that these models could be used in place of the mouse two-year bioassay (Cohen et al., 2001; Blaauboer et al., 1998). Coupled with information on genotoxicity, particularly DNA reactivity, structure–activity relationships, results from other bioassays, and the results of other mechanistic investigations including toxicokinetics, metabolism, and mechanistic information, these alternate mouse models for carcinogenicity appear to be useful models for assessing the carcinogenicity of chemicals. In general, replacement of the two-year rodent bioassay by transgenic mouse assay have not been as successful as originally planned (Boverhof, 2011). However, as models for studying the roles of specific pathways and mechanisms in the carcinogenesis process, these models remain an excellent resource.

New Approaches

A major concern for the correct evaluation of the safety of chemicals, and as mixtures is the need to obtain reliable and pertinent scientific information on which to develop proper risk evaluation and assessment. Our current bioassays approaches have been criticized as being too time consuming and not pertinent to human health. Research during the last decade of the 20th century and the turn of the 21st century have resulted in a dramatic increase in our knowledge of the cellular and molecular pathways that contribute to the induction and prevention of cancer. This coupled with technological advances in high-throughput assays, and computational science has raised the question if the current approaches for carcinogenesis evaluation of chemicals (and mixtures) should be reevaluated. Following a NRC report in 2007 (Toxicity Testing in the 21st Century: A Vision and a Strategy), several approaches have begun utilizing high-throughput and computational approaches to evaluating the effects of chemicals on biological processes and pathways that are important in toxicity (including cancer). These approaches are directed to using cells, cell lines and components of cells. Two approaches currently well underway include the USEPA Toxcast program and the NTP/NIEHS lead Tox 21 program (tox21 is a consortium of NIH, FDA, USEPA) utilizing this approach.

Recently, the IARC (Vol 100) has reported a review of chemically induced cancer in humans. This body has identified and classified 103 compounds as carcinogenic in humans. This review is divided into four major categories of carcinogens deemed to be human carcinogens by the IARC. These include pharmaceuticals, metals, biologicals, and radiation. A number of factors have been implicated in the induction of cancer in humans. Infectious agents, lifestyle, medical treatments, environmental, and occupational exposure account either directly or indirectly to the majority of cancers seen in humans. Of these, the component that contributes the most to human cancer induction and progression is lifestyle: tobacco use, alcohol use, and poor diet (Table 8-25). Tobacco usage either through smoking tobacco, chewing tobacco, or tobacco snuff-type products is estimated be responsible for 25% to 40% of all human cancers. In particular, a strong correlation between tobacco usage and mouth, larynx, lung, esophageal, and bladder cancer exists. It has been estimated (Doll and Peto, 1981) that 85% to 90% of all lung cancer cases in the United States are a direct result of tobacco use. The induction of pancreatic cancer also appears to have a linkage to tobacco use. Alcohol consumption also contribute anywhere from 2% to 4% of cancers of the esophagus, liver, and larynx.

Poor diets whether high-fat, low-protein, high-calories or diets lacking in needed antioxidants and minerals account for anywhere from 10% to 70% of human cancers. Diet contaminated by molds such as Aspergillus flavis (which produces aflatoxin B1) have been linked epidemiologically to a higher incidence of liver cancer. It also appears that aflatoxin B1 exposure coupled with hepatitis B virus infection produces an increased incidence of liver cancer compared to aflatoxin B1 or hepatitis B exposure individually. Mold-contaminated foodstuffs have also been shown to produce nitroso compounds.

There is substantial evidence that overnutrition either through excess calories and/or high-fat diets contribute to a number of human cancers (Doll and Peto, 1981). In particular, high-fat and high-calorie diets have been linked to breast, colon, and gall bladder cancer in humans. Diets poor in antioxidants and/or vitamins such as vitamin A and vitamin E probably also contribute to the onset of cancer. The method of cooking may also influence the production of carcinogens produced in the cooking process. Carcinogenic HCAs and PAHs are formed during broiling and grilling of meat. Acrylamide, a suspected human carcinogen, has been found in fried foods at low concentrations.

A number of occupations have been associated with the development of specific cancers (Table 8-26). As noted earlier, the linkage between chimney sweeps who as young boys in England were exposed to PAHs through the constant exposure to soot developed scrotal cancer. The linkage between occupational exposure to asbestos and the development of bronchiogenic carcinoma and as well as malignant mesothelioma has been clearly established. The appearance of bronchiogenic carcinoma was much higher in shipyard workers who were exposed to both asbestos and cigarette smoking. Muscat and Wynder (1995) noted no association between cigarette smoking and mesothelioma formation. Similarly, asbestos exposure by itself (without smoking) does not seem to increase the risk of bronchiogenic carcinoma. Aromatic amines used in the chemical and dye industries have been shown to produce or induce bladder cancer in humans. Prolonged high exposure to benzene in an occupational setting has been linked to the formation of acute myelogenous leukemia in humans.

A number of drugs and medical diagnostic tools have also been linked to the induction of human cancer (Table 8-27). Anticancer drugs such as the alkylating agent cyclophosphamide have been associated with bladder tumors and leukemia in patients receiving these treatments. The administration of the synthetic estrogenic compound diethylstibestrol to pregnant women in order to improve embryo implantation and prevent spontaneous abortion has been shown to result in the formation of clear cell carcinomas of the vagina in the female offspring of mothers treated with diethylstilbestrol during pregnancy. The use of oral contraceptives containing synthetic estrogens as their major or only component has been implicated in the induction of liver cell adenomas. In addition, an association exists between prolonged use of estrogenic oral contraceptives and an increase incidence of premenopausal breast cancer. Androgenic steroids and synthetic testosterone compounds have been implicated in hepatocellular carcinoma induction. Therapeutic immunosuppression given to transplant patients or arising secondary to selective diseases such as AIDS result in an increase in a variety of different neoplasms. These results further support the role of the immune system in identifying and removing early preneoplastic cells from the body. In addition, the previously used diagnostic tracer Thorotrast has been sufficiently linked to the formation of hemangiosarcomas.

The assessment and designation of a chemical agent or a mixture of chemicals as carcinogenic in humans is evaluated by various agencies worldwide. The evaluation usually encompasses epidemiological, experimental animal, and in vitro data utilizing assays as described earlier in this chapter. One of the first schemes for the classification of an agent’s carcinogenicity was devised by the IARC (Table 8-28). The IARC approach assigns the chemical or mixture to one of five groupings based upon strength of evidence for the agent’s possible, probable, or definite carcinogenicity to humans. In Group one classification, the agent or mixture is classified as definitely carcinogenic to humans. The second grouping is Group 2A in which the agent is probably carcinogenic to humans. In Group 2B, the agent is classified as possibly carcinogenic to humans. Group 3, the agent is not classifiable. And in the last group, Group 4, the agent is not carcinogenic to humans. The IARC produces a series of monographs that describe the methodology for the evaluation of specific chemicals and the rationale for their final classification. Currently, 100 chemical agents or mixtures or exposure circumstances have been classified by IARC as falling in Group 1, which shows sufficient evidence for carcinogenicity to humans.

Similar classifications exist for the USEPA, the Food & Drug Administration, and the European Community (EC). The classification of agents with regard to human carcinogenicity can many times be very difficult in particular, when animal data and/or epidemiological data in humans are inconclusive or confounded.

New USEPA Guidelines for Cancer Risk Assessment (2005) uses descriptors for defining the relative carcinogenic risk to humans (Table 8-29). These descriptors include carcinogenic to humans, likely to be carcinogenic to humans, suggestive evidence of carcinogenic potential, inadequate information to access carcinogenic potential, and, not likely to be carcinogenic in humans. The EPA Guidelines take into account the understanding of the mode of carcinogenic action and associated key precursor events needed for the cancer to form (Table 8-30). Central to the USEPA Guidelines for Cancer Risk Assessment is the utilization of the mode of action framework to define the key events in rodents and assessment of whether those same key events and mode of action can occur in humans (Table 8-31). This approach is similar to that has been developed by the International Program on Chemical Safety and by panels in the International Life Sciences Institute.

The current listing of chemical, occupations, or behaviors deemed to be carcinogenic in humans by IARC and the NTP are noted in Tables 8-32 to 8-33, respectively.

The induction of cancer by chemicals is well established in animal models as well as in humans. Linkages between chemicals found in human lifestyle, occupational exposure, and environmental exposure provides strong evidence for the induction or contribution to environmental occupational lifestyle carcinogens to human cancer. Cancer is a multistage process in its most reductive form involves initial mutational events followed by changes in gene expression leading to the selected clonal proliferation of the precancerous cell. Neoplasia appears to exhibit multiple characteristics including increase selective lesion growth (through sustained cell proliferation and /or resistance to apoptosis), the induction of angiogenesis, enabling replicative immortality, activation of factors that influence invasion and metastasis, evasion of normal growth suppression, modulation of energy metabolism, and the avoidance of attack by the immune system. The multistage nature and characteristics of the process have been extensively examined with regard to molecular, cellular, tissue, and organ events.