Our earliest knowledge concerning chemical carcinogenesis came from clinical observations in humans. In 1775, Percivall Pott, an eminent English physician and surgeon, described the occur- rence of cancer of the scrotum in a number of his patients. The common history given by these individuals was their employment as chimney sweeps when they were young. On the basis of this observation, Pott, with remarkable insight, concluded (1) that the occupation of these men as young boys was directly and causally related to their malignant disease and (2) that the large amounts of soot to which they were exposed was the causative agent of the cancer. Strangely enough, Pott did not suggest avoidance of contact with soot as a means of prevention, although his report in 1775 apparently inspired the Danish Chimney Sweepers’ Guild to rule 3 years later that its members should bathe daily. While the publication of Pott soon led other observers to attribute cancer of various sites to soot exposure, there was a relative lack of effective impact of his work on British public health practice during the succeeding century (Lawley, 1994). It was not until more than a century later that Butlin (1892) reported the relative rarity of scrotal cancer in chimney sweeps on the European continent compared with those in England. It appeared that the lower incidence of the disease on the continent was the result of frequent bathing and protec- tive clothing.
The lesson from Pott’s report has been a long time in the learning. One hundred years after his publication, the high incidence of skin cancer among certain German workers was traced to their exposure to coal tar, the chief constituent of the chimney sweeps’ soot (cf. Miller, 1978). It was another 40 years before the disease was reproduced experimentally, and even today—more than 200 years after Pott’s original scientific report of the association of soot and smoke prod- ucts with cancer—many still disregard the obvious hazards of the carcinogenic products that result from the combustion of tobacco in cigarettes and of many of the organic fuels of the in- dustrialized world.
Before we embark on a discussion of the causation of cancer, it again becomes important to add to our vocabulary. The term carcinogen has generally been used by oncologists to indicate an agent that causes cancer. However, as our knowledge of oncology increases, this simplistic definition is not sufficient. We have proposed the following definition to include most examples of agents that are carcinogens, at the same time excluding those agents that do not have a direct action on cells undergoing neoplastic transformation.
A carcinogen is an agent whose administration to previously untreated animals leads to a sta- tistically significant increased incidence of neoplasms of one or more histogenetic types as compared with that in appropriate untreated animals, whether the control animals have low or high spontaneous incidences of the neoplasms in question.
This definition includes the induction of neoplasms that are not usually observed, the ear- lier induction of neoplasms that are usually observed, and/or the induction of more neoplasms than are usually found. Although it would be important to distinguish between agents that in- duce neoplasms by direct action on the cells that become neoplastic and those that produce neo- plasia by indirect actions in the animal as a whole, at present it is not always possible to do so. Some agents, such as immune suppressants, can increase the incidence of neoplasms in tissues previously exposed to carcinogens by indirect effects on the host. Where the action of a chemi- cal in causing an increase in neoplasms is known to be indirect—mediated by its effect on the host—the agent should not be designated a carcinogen.
At this stage of our discussion, the above definition and a prolonged explanation may be confusing. Later in the text (Chapters 7 and 8) we consider the stages and modifying factors of the process of carcinogenesis. As we learn more of the process of carcinogenesis, the term carcino- gen will be further refined in relation to general usage as well as in terms of specific chemicals.
One hundred and forty years after Dr. Pott’s report of the association of soot from the combus- tion of coal with epidermal cancer of the scrotum, an experimental basis for Pott’s clinical obser- vation was reported. In 1915, the Japanese pathologists Yamagawa and Ichikawa described the first production of skin tumors in animals by the application of coal tar to the skin. These inves- tigators repeatedly applied crude coal tar to the ears of rabbits for a number of months, finally producing both benign and, later, malignant epidermal neoplasms. Later studies demonstrated that the skin of mice was also susceptible to the carcinogenic action of such organic tars. During the next 15 years, extensive attempts were made to determine the nature of the material in the crude tars that caused malignancy. In 1932, Kennaway and associates reported the production of carcinogenic tars by pyrolysis of simple organic compounds consisting only of carbon and hy- drogen (cf. Kennaway, 1955). In the early 1930s, several polycyclic aromatic hydrocarbons were isolated from active crude tar fractions. In 1930, the first synthetic carcinogenic chemical was produced. This compound, dibenz(a,h)anthracene (Figure 3.1), was tested for carcinogenic ac- tivity by painting it on the skin of mice and found to be a potent carcinogen. The isolation from coal tar and the synthesis of benzo(a)pyrene (3,4-benzpyrene) were achieved in 1932. The struc- tures of several polycyclic aromatic hydrocarbons are given in Figure 3.1. Polycyclic hydrocar- bons vary in their carcinogenic potencies; for example, the compound dibenz(a,c)anthracene has very little carcinogenic activity while, as noted above, the a,h isomer is carcinogenic (cf. Heidel- berger, 1970). Among the more potent polycyclic aromatic hydrocarbon carcinogens yet de- scribed are 3-methylcholanthrene and 7,12-dimethylbenz(a)anthracene. The carcinogenic dibenzo(c,q)carbazole is also considered in this class of compounds, possessing a nitrogen in its central ring. Further on, other polycyclic aromatic hydrocarbons containing nitrogen, sulfur, and halogens are considered (Chapters 7 and 9). Benzo(e)pyrene is reportedly inactive in inducing skin cancer in mice but can serve to “initiate” the carcinogenic process (Chapter 7). Perylene is inactive as a chemical carcinogen, while chrysene may have slight carcinogenic activity in this system.
Polycyclic aromatic hydrocarbons, both carcinogenic and noncarcinogenic, are found ubiquitously in air, soil, and other places in our environment. Reportedly as much as 894 tons of benzo(a)pyrene are emitted into the air per year in the United States. Much of this is due to industrial and home combustion of carbon-containing materials, whereas natural fires contribute only about 1% (Zedeck, 1980). As expected, soil contents of carcinogenic polycyclic aromatic
hydrocarbons are much greater near cities and industrial areas than in rural and wilderness areas, with concentrations differing by as much as a thousandfold.
Because of the relatedness of the chemical structures of many carcinogenic polycyclic hy- drocarbons, numerous studies have reported attempts to determine the molecular configura- tion(s) of these molecules that is (are) responsible for their carcinogenic activity. One of the earlier attempts was that of the Pullmans (cf. Pullman and Pullman, 1955), who utilized quan- tum mechanics and molecular orbital theory to determine reactivity of specific regions of carci- nogenic polycyclic aromatic hydrocarbons. Reactivity indices were calculated for regions of the molecules termed the K and L regions (Figure 3.2). A K region is defined as the external corner of a phenanthrenic moiety in a polycyclic aromatic hydrocarbon, whereas an L region consists of a pair of opposed, open anthracenic “point” atoms (cf. Lowe and Silverman, 1984). If the calcu- lated reactivity of the K region exceeds a certain limit, the chemical is expected to be carcino- genic unless the more active L region also exceeds its reactivity limit. In the latter case, the molecule will not be carcinogenic. The K and L regions for several polycyclic aromatic hydro-
carbons are noted in Figure 3.2. Although this early attempt at discovering a relationship be- tween the structure of a chemical and its carcinogenic activity has not found general application because of numerous exceptions, much more extensive methods of relating structure and carci- nogenic activity of polycyclic aromatic hydrocarbons have recently been developed through the use of computerized databases (Richard and Woo, 1990). In addition, as outlined below, the me- tabolism of polycyclic aromatic hydrocarbons is important for their carcinogenicity, and certain structural components of the molecule have been utilized in predicting carcinogenic activity (Jerina et al., 1982).
In 1935, Sasaki and Yoshida opened another field of chemical carcinogenesis by demon- strating that the feeding of the azo dye o-aminoazotoluene (2′,3-dimethyl-4-aminoazobenzene) (Figure 3.3) to rats resulted in the development of liver neoplasms. Kinosita (1936) later demon- strated that administration of 4-dimethylaminoazobenzene in the diet also caused neoplasms in the liver. A number of analogs of this compound were also prepared and tested for carcinogenic potential. An interesting correlation arising from all of these studies was the fact that the amino
group of carcinogenic dyes usually had at least one methyl substituent, although o-aminoazotol- uene does not. Unlike the polycyclic aromatic hydrocarbons, the azo dyes generally did not act at the site of first contact of the compound with the organism but rather in a remote area, the liver. Painting of the skin with most azo dyes resulted in few or no tumors, and the oral adminis- tration of polycyclic aromatic hydrocarbons to rodents except in the neonatal period has gener- ally resulted in no hepatomas. Another important carcinogen that acts at remote sites is 2- acetylaminofluorene (Figure 3.3). In addition, the aromatic amines, 2-naphthylamine and benzi- dine, are carcinogenic for the urinary bladder in humans. The carcinogenic chemical ethyl car- bamate also appears to be a general “initiating agent” in the mouse (Chapter 6). Ethyl carbamate was in use in Japan from 1950 to 1975 as a cosolvent for dissolving water-insoluble analgesic drugs (Miller, 1991), but this practice was stopped after 1975. No systematic study of the inci- dence of cancer in this cohort has been made as yet. In addition, certain cytocidal drugs, such as the nitrogen mustards (Figure 3.3), which have been used to treat cancer in humans, are also known to be potent carcinogens (Chapter 11). The other three agents depicted on the bottom line of Figure 3.2 are also alkylating agents that are used industrially. Bis(chloromethyl)ether, a pop- ular intermediate in organic synthetic reactions, has been classified as carcinogenic to the human based on epidemiological as well as animal studies (Vainio et al., 1991).
Members of a number of other classes of compounds have been shown to be strong carcin- ogens and of potential importance in the genesis of neoplasia in the human being. The dialkylni- trosamines have the following general structure:
in which R1 and R2 can be alkyl substituents or can be fused to yield a cyclic aliphatic substitu- ent. One of the structurally simplest nitrosamines, dimethylnitrosamine (Figure 3.3), is highly carcinogenic for the liver and kidney in rodents and for these and/or other tissues in all other mammals tested. Hepatic toxicity due to dimethylnitrosamine exposure occurred in humans working with this chemical at the time of its earliest industrial use. Subsequently, such exposure was eliminated by cessation of the industrial use of nitrosamines as solvents. Several investiga- tors (cf. Lijinsky, 1977; Magee and Swann, 1969; Mirvish et al., 1983) have shown in experi- mental animals that certain dietary components, especially in the presence of high levels of nitrite, may give rise to low levels
of nitrosamines or nitrosamides in the diet or in the stomach and induce neoplasia of the gas- trointestinal tract. The action of bacterial flora within the intestine may enhance the formation of these compounds. Furthermore, there is increasing evidence of an etiological role for endoge- nously formed N-nitroso compounds in the development of certain human cancers (Bartsch et al., 1990). The nitrosamine NNK (Figure 3.3) is produced in tobacco smoke from nicotine, a normal tobacco alkaloid (cf. Hecht, 1985). This is an extremely potent carcinogen to which all tobacco smokers are exposed and may play a role in the induction of tobacco-related cancers in the human. Methapyrilene was developed as an antihistamine but is a potent carcinogen in the rat (Mirsalis, 1987).
Another important environmental as well as experimental hepatocarcinogenic agent is aflatoxin B1. This toxic substance is produced by certain strains of the mold Aspergillus flavus. Aflatoxin B1 is one of the most potent hepatocarcinogenic agents known and has produced neo- plasms in rodents, fish, birds, and primates. This agent is a potential contaminant of many farm products (for example, grain, peanuts) that are stored under warm and humid conditions for some time. Aflatoxin B1 and related compounds may cause some of the toxic hepatitis and he- patic neoplasia seen in various parts of Africa and the Far East (Chapter 11).
Ethionine is an antimetabolite of the normal amino acid methionine. Farber (cf. 1963) was the first to show definitively that administration of ethionine in the diet for extended periods resulted in the development of liver cancer in rats. This was the first example of the direct inter- ference with the metabolism of a normal metabolic constituent resulting in the development of cancer.
In addition to organic compounds such as those illustrated in Figures 3.1 and 3.3, a num- ber of inorganic elements and their compounds have been shown to be carcinogenic in both ani- mals and humans (IARC, 1973). Figure 3.4 shows a periodic classification of the elements with an indication of those elements that, in their elemental form or compounded with other ele- ments, have been shown to be carcinogenic or possibly carcinogenic in humans, other animals, or both (Martell, 1981). This is not to say that all the elements of the table that have not been specifically shown to be carcinogenic should be considered noncarcinogenic. Many elements and their compounds have not yet been adequately tested for carcinogenicity in animals, and at this time there is no evidence that such elements exhibit effects in humans on the basis of epide- miological studies (Chapter 11). On the basis of epidemiological studies (Chapter 11), chro- mium and nickel are carcinogenic in both humans and experimental animals. However, several other elements have so far demonstrated carcinogenicity only in experimental animals. Exposure to several of these, including lead (Verschaeve et al., 1979) and beryllium (Kuschner, 1981), has been implicated as causes of cancer in humans, but as yet the data are not sufficient to demon- strate such an association unequivocally. On the other hand, arsenic and its derivatives present an
Figure 3.4 Periodic classification of the elements forming compounds having definite carcinogenic ef- fects in the human (squares) or lower animals (circles). Dashed squares or circles around the symbols of some elements indicate that the compounds formed from them are suggestive of but not proven as to their carcinogenicity for humans or lower animals. (Adapted from Martell, 1981, with permission of the author and publisher.)
interesting paradox (Landrigan, 1981) in that there is essentially no experimental evidence to substantiate the carcinogenicity of this element and its compounds in lower animals, but the evi- dence for its carcinogenicity in humans is quite clear (Chapter 11).
One class of chemical carcinogens is different from those described thus far—the group of inert plastic and metal films or similar forms that cause sarcomas at the implantation site in some rodents (Brand et al., 1975). The implantation site is usually subcutaneous. At least one study indicated that the implantation of metal “films” did not induce neoplasms when they were im- planted within the central nervous system (Bischoff and Bryson, 1976), although this may have been related to the relatively small size of the implants. Rats and mice are highly susceptible to this form of carcinogenesis, but guinea pigs appear to be resistant (Stinson, 1964). The carcino- genic properties of the implant are, to a large extent, dependent on its physical characteristics and surface area. Multiple perforations—each greater than a certain diameter (for example, 0.4
µm)—pulverization, or roughening of the surface of the implant (Ferguson, 1977) markedly re- duced the incidence of neoplasms. Plastic sponge implants may also induce sarcomas subcuta- neously, and in this instance the yield of tumors is dependent on the thickness of the sponge implant (Roe et al., 1967). The age of the animal upon implantation also affects the time from implantation until tumor development. Young rats first developed sarcomas at the site of implan-
tation after nearly 1¹⁄₂ years, whereas older animals developed such neoplasms within 6 months
after implantation of the material (Paulini et al., 1975).
The chemical nature of the implant is clearly not the critical factor in its ability to trans- form normal cells to neoplastic cells. Brand and associates (cf. Johnson et al., 1970) have stud- ied this phenomenon for many years and have demonstrated a variety of kinetic and morphological characteristics of the process of “foreign-body tumorigenesis” in mice. These studies have shown that DNA synthesis occurs in the film-attached cell population throughout the preneoplastic phase and that preneoplastic cells may be identified well before neoplasms de- velop (Thomassen et al., 1978). More recently, Kirkpatrick and associates (2000), using a vari- ety of plastics and metals as the subcutaneous implant, found that about one-third of the animals developed sarcomas in association with the implant by 2 years of age. Examination of the im- plants between 8 and 24 months after implantation revealed the presence of nonneoplastic pro- liferative lesions as well as preneoplastic lesions (Chapter 7), which may have served as precursors to the sarcomas. Brand has suggested that such “preneoplastic” cells may already be present in the normal tissue prior to implantation and that the implant appears to “create the conditions” required for carcinogenesis of these cells (Brand et al., 1975). Other possible mech- anisms for this unique type of carcinogenesis are discussed later (Chapter 7), as well as its rela- tion to specific human carcinogens (Chapter 11). An interesting but possibly unrelated observation is the demonstration by Hirono and associates (1983) of the carcinogenicity of so- dium dextran sulfate in relation to its molecular weight. Only the polymer with a molecular weight of approximately 54,000, compared with polymers of 520,000 and 9500, induced colon cancer in rats when given orally.
While the epidemiological evidence that implants of prostheses in the human—such as seen with the repair of hernias, joint replacements, etc.—induce the formation of sarcomas is not substantial, there have been a number of isolated reports of neoplasms arising in association with such foreign bodies (Sunderman, 1989). A study in the rat of the carcinogenic potential of a number of materials used in such prostheses demonstrated a small increase in sarcomas in ani- mals with certain metal alloy implants that contained significant amounts of cobalt, chromium, or nickel (Memoli et al., 1986). However, it is likely that in the human “foreign-body tumorigen- esis” may be induced by certain types of asbestos fibers, wherein the dimensions of the fiber are directly related to its effectiveness as a carcinogenic agent (Chapter 11).