CARCINOGENESIS: Synopsis for Pesticide Toxicology Cancer is a disease caused by the excessive division of cells, forming an overgrowth of tissue in the organ where the excessive cell division occurs. Any process that triggers excessive cell division is likely to lead to cancer, but not every cell overgrowth is a cancer. Benign tumors are noncancerous overgrowths. They occur when excessive cell division causes an overgrowth of cells within a tissue, but the cells within the overgrowth still retain normal patterns of growth, insofar as they do not invade adjacent tissues, and all the cells in the overgrowth remain in place. For an overgrowth to be cancerous, it must have several attributes in addition to excessive cell division. These extra characteristics are invasiveness and metastasis. Invasiveness is the movement of the abnormal cells into adjacent normal tissue. Metastasis is the movement of tumor cells through the bloodstream or lymph fluid to other parts of the body. 1. Cancer is a genetic disease. The evidence is overwhelming that cancer is in all cases a genetic disease. This is easily demonstrated for genotoxic carcinogens, since these (whether they are chemicals or physical agents) interact directly with DNA. Moreover, the weight of evidence strongly implicates indirect effects on DNA as the mechanism by which nongenotoxic chemicals induce cancer. The genetic origins of cancer are most easily demonstrated with mutagens -- genotoxic carcinogens. Genotoxic carcinogens may cause point mutations, in which a single base pair is altered; they may cause deletions, in which one or more bases is lost; and/or they may form DNA adducts, with subsequent mispairing of the adducted base during DNA replication. Each of these events changes the sequence of bases in the double helix: directly, in the case of deletions or point mutations; indirectly, by subsequent mispairing, in the case of adduct formation. Genotoxic carcinogens may also interfere with DNA repair or DNA methylation, which do not directly alter the sequence of bases in DNA, but permit spontaneously occurring errors to persist. If one or more DNA repair systems are defective, errors of DNA replication (whether spontaneous or chemically induced) are not corrected. Repair is essential to normal DNA replication, as is shown by inherited diseases in which a repair system is defective. People with such syndromes usually develop cancer at young ages. For example, children with xeroderma pigmentosum are unable to repair UV-induced DNA damage, and often die of skin cancer in their early teens. Chemicals that alter DNA methylation alter gene expression, because demethylation tends to turn a gene on. DNA methylation patterns are (like DNA itself) self-replicating, so altered patterns of DNA methylation change gene expression. If demethylation turns on a gene that enhances (or turns off a gene that inhibits) cell division, increased rates of cell division will lead to increasing numbers of the altered cell. Other mechanisms by which agents interact with DNA include chromosome breakage or interference with the metaphase spindle apparatus, either of which can lead to unequal loss of parts of chromosomes and/or unequal distribution of DNA between daughter cells during division. Obviously, many of the possible instances of damage to DNA would lead to cell death; these would not be recognized if cancer is the end point. Only the subset of effects on DNA which lead to increased cell division would be recognized as carcinogenic. Nongenotoxic carcinogens do not interact directly with DNA or its replicative processes, but cause changes in a tissue that result in the subsequent production of mutations relevant to carcinogenesis. Implicit in this definition is the possibility that the agent is gone from the cell by the time mutation occurs. The best known examples of nongenotoxic carcinogens are growth-promoting hormones. Such hormones (notably the sex hormones, estrogen and testosterone) stimulate increased cell division of their target organs and, by stimulating cell division, increase the probability that other (mutational) events leading to cancer will occur or be propagated. Another example is the large group of chemicals that induce subsets of cytochromes P450. Because this family of enzymes also activates many promutagens to active mutagens, chemicals that induce P450s are quite likely to increase the level of carcinogenic chemicals in the organism as a whole (or at the least, in its liver), and so increase the probability of a mutagenic event. Because the chemistry and structure of DNA is essentially identical across species from bacteria to humans, it must be assumed that genotoxicity is a property of the agent, and that the activity of genotoxic agents can be extrapolated across species. For nongenotoxic carcinogens, the tissue response leading to mutagenesis is the critical event, and this response can be highly species-specific. For example, bacteria do not have hormoneresponsive elements; therefore hormones are negative in bacterial assays of genotoxicity, even though they are proven mammalian (and human) carcinogens. Species-specificity may also be seen with genotoxic agents if 1) access to DNA is blocked in one species or 2) if the environmental agent must be activated before it can interact with DNA (see above). For example: metals that are carcinogenic in mammals are not genotoxic in bacteria, simply because bacteria are better able to exclude metals from the cell than are mammals. This is an example of lack of access. Many promutagens - chemicals that must be metabolically activated in order to interact with DNA - are also negative in simple microbial assays, because bacteria lack the cytochrome P450 enzymes that metabolize so many xenobiotics in animals1. Obviously there are also cases where mammalian species differ in metabolism of a promutagen, which may then be carcinogenic in one species but not in another -- but it would be foolhardy to assume that this is the case for any given chemical, unless there is good evidence for the metabolic difference. 2. Carcinogenesis Observed It has long been noted that the probability of cancer increases with increasing age. This is true in all species; however, the incidence of cancer rises in proportion to the life span, so that rats (old at 2 years) begin to exhibit spontaneous cancers with some frequency by 1.5 years of age, whereas in humans the steep increases come after the 4th or 5th decade. There is, however, already a low incidence of cancer in children (e.g., 1 child per 1,000 develops leukemia before age 12). If we model the observed cancer incidence in adults using basic probability theory, the incidence pattern corresponds to a pattern which requires that more than 2 independent mutagenic events occur in the same cell, while childhood cancers require a single event. The complete sequence of genetic alterations required for a cell to progress from fully normal to a cancer cell has been delineated for several cancers. Colon cancer, a common adult cancer that increases slowly with increasing age, requires at least 6 alterations of DNA in a single cell and its progeny before metastatic cancer is seen. In contrast, retinoblastoma, a cancer of the eye seen mostly in young children, requires only the loss of a single region on both copies of chromosome 13: that is, a maximum of 2 mutagenic events2. 3. Identifying Environmental Carcinogens It is extraordinarily difficult to prove that an environmental chemical or agent is a human carcinogen. There are several reasons. First, for most cancers there is a long interval between exposure and the appearance of symptoms (detectable tumors). For common adult cancers, this interval is typically 20 to 30 years. Secondly, cancer of a specific site is relatively rare in any given population: even smoking a pack of cigarettes per day for 20 years induces cancer in only 10% of smokers. Thirdly, for most cancers there is a background incidence (spontaneous occurrence) of the same type of cancer. A generation ago, one woman out of 8 (12.5%) developed breast cancer. If a specific chemical caused breast cancer in 10% of exposed women, and if 10% of the population were exposed to this (very potent carcinogenic) chemical, it would only increase the incidence of breast cancer from 12.5% to 13.5%. In laboratory rats, one might identify a 1% increase with a sample of 300 to 400 rats, all of which are kept in the same room under uniform conditions of heat, light, disease and diet. How big a sample of women, with their varied health histories, differing lifestyles, and exposure to multitudinous 1 It is possible to add metabolic enzymes to the bacterial culture medium , in which case promutagens can also be identified. 2 In the inherited form of retinoblastoma, a child is missing the critical segment on one copy of chromosome 13 throughout all cells. Such children typically develop tumors in both eyes, suggesting that the loss of the critical segment in one copy of chromosome 13 is a relatively common event. other chemicals would one need to identify this increase? Moreover, how many of the women would remember their exposure at all, much less be able to quantify it3? Because of the long latency period and the confounding effects of spontaneously occurring cancer, far more "animal carcinogens" than "human carcinogens" have been identified. As for other forms of toxicity, occupational exposure has usually provided the best evidence for chemical carcinogenicity in humans. Occupational cohorts are often well-defined and exposures tend to be both heavy and prolonged. Thus the carcinogenicity of vinyl chloride was proven when doctors identified a high incidence of liver cancer among workers in vinyl chloride factories. Identification was helped by the rarity of this particular cancer, which is otherwise seen only in older heavy drinkers (who often also have cirrhosis of the liver). Of Richard Doll's "established causes of human cancer", 18 are occupational, 11 medical, and 11 social or lifestyle (Table in textbook). 3. Testing for Carcinogens In vitro assays, comprising microbial assays and cell or tissue culture assays, are rapid and inexpensive. They may test for mutagenicity or clastogenicity (chromosome breakage). Because of the difference in metabolic capacity and the absence of complex regulatory systems (such as hormone receptors), these assays are most useful as preliminary screens to weed out potent genotoxicants. Animal bioassays are expensive, require huge amounts of highly skilled labor, and take 3-5 years to complete. They remain the most reliable test of carcinogenicity, and are required not only for marketing of new pesticides, but also to maintain registration of older pesticides. Unfortunately, relatively few industrial chemicals are subject to the same testing requirements as pesticides. Human epidemiology is an unfortunately common way of identifying human carcinogens. All epidemiology studies require that a significant number of people be exposed to the suspect agent for a long time. In practice, dependence on epidemiological studies to identify human carcinogens would mean that such agents are in use for at least 20 to 30 years before controls are instituted. The result is that cases continue to occur for 20-30 years after a carcinogen is identified. (Example: asbestos. Case history in text book.) Case-control studies compare exposures of people with the cancer of interest to exposures of people with other diseases (cancer and noncancer). Case-control studies are usually retrospective: they begin with people who are already ill. Problems with retrospective studies include the difficulty of determining exposures that may be > 20 years in the past, the confounding factors of other events (such as life-styles, medical histories, socioeconomic status). Prospective studies follow a healthy population with different exposure patterns (e.g., smokers vs nonsmokers) to see what diseases occur with the passage of time. Such studies take a long time. Moreover, they are extremely expensive, because the number of people who must be included in the initial sample is very large in order to have enough cases for statistical significance at study's end. Ecological studies are retrospective studies using surrogate measurements (typically, occupation) for exposure: an example would be a study comparing the incidence of soft tissue sarcoma among farmers using herbicides and feedlot operators. Although less expensive than other epidemiology studies, ecological studies cannot control for confounding factors such as smoking, alcohol consumption, or quality of medical care. Moreover, they do not really measure exposure, so people may easily be misclassified (e.g., an organic farmer in the ‘exposed’ category). 3 A clear example of the difficulties of ascertaining exposures is provided by the example of clear-cell adenocarcinoma of the vagina, resulting from prenatal exposure to the synthetic estrogen DES. In many cases, women whose daughters developed this rare cancer either did not remember or had never known that they took this hormone. Several women insisted they had never been told. Since this cancer is extremely rare in the general population – and unheard of in young women – the cause can be presumed even when exposure could not be documented from prescription records. The same would not be true for more common cancers.