Unit 6: Risk, Exposure, and Health // Section 5: Cancer Risk
Cancer is a major focus of environmental risk analysis for several reasons. First, it is a leading cause of death in developed countries that have passed through the demographic transition and brought other threats such as infectious disease and malnutrition under control (for more details, see Unit 5, "Human Population Dynamics"). Various types of cancer account for 25 percent or more of yearly deaths in the United States and other industrialized nations. Cancer rates are also increasing in the developing world.
Second, environmental exposures broadly defined account for a substantial fraction of cancers—at least two-thirds of all cases in the United States, according to the National Institutes of Health (footnote 11). This estimate includes all influences outside the body, including many lifestyle choices such as smoking and eating a high-fat diet. Tobacco use alone causes about one-third of all annual U.S. cancer deaths, while inactivity and obesity together cause an estimated 25 to 30 percent of several major types of cancer (footnote 12).
In contrast, the narrower category of exposure to environmental pollutants causes about 5 percent of annual U.S. cancer deaths (footnote 13). However, these risks are not spread equally across the population. They have higher impacts on heavily-exposed groups—for example, workers in industries that use known or possibly carcinogenic substances or communities that draw their drinking water from a contaminated source. Environmental exposures also can cause gene alterations that may lead to cancer over time.
Risk analyses have led to bans or use restrictions on carcinogens such as benzene (a solvent), asbestos (an insulating fiber), and a number of pesticides, and have contributed to the development of guidelines and workplace standards that minimize exposure to other known or suspected carcinogens. Figure 11 shows one example, an illustration from an EPA brochure on reducing radon gas levels in houses. Exposure to radon, a natural byproduct of radioactive elements decaying in surrounding soil, causes an estimated 20,000 lung cancer deaths in the United States annually.
Figure 11. Techniques for reducing home radon gas levels
See larger image
Source: Courtesy United States Environmental Protection Agency.
The Environmental Protection Agency and other regulators quantify cancer risks as probabilities—the number of excess individual lifetime cases of cancer (beyond those that could be expected to occur on average in the population) that will occur in response to a specific exposure. For example, in 1999 EPA estimated that the added cancer risk from polychlorinated biphenyl (PCB) pollution in the upper Hudson River was one additional case of cancer for every 1,000 people who ate one meal per week of fish caught in that section of the river (footnote 14). As this approach suggests, not everyone exposed to a hazard becomes ill, but exposure increases the likelihood of suffering harmful effects.
EPA's traditional classification system for carcinogens combines human data, animal data, and other supporting evidence to characterize the weight of evidence regarding whether a substance may cause cancer in humans (Table 2). However, these rankings are based on levels of certainty that agents may cause cancer, not on relative levels of risk from one substance versus another, so other materials not currently classified as carcinogens may be equally hazardous. Some materials are classified as possible or probable carcinogens because they have not been studied thoroughly enough yet to make a determination about whether they cause cancer in humans (footnote 15).
|Human Evidence||Animal Evidence|
|Sufficient||Limited||Inadequate||No Data||No Evidence|
Group A Human carcinogen
Group B Probable carcinogen
B1 Limited evidence of carcinogenicity from epidemiology studies
B2 Inadequate human evidence but positive animal evidence
Group C Possible human carcinogen
Group D Not classifiable as to human carcinogenicity
Group E Evidence of noncarcinogenicity for humans
One of the most controversial issues in cancer risk assessment is whether the dose-response relationship for all carcinogens is linear. Most risk analyses assume that the answer is yes—in other words, that exposure to any amount of a carcinogen produces some risk of cancer, with risk increasing in proportion to the size of the dose. Under this approach, risk is estimated using the equation
Risk = LADD x CSF
where risk is the unitless probability of an individual developing cancer, LADD is the lifetime average daily dose per unit of body weight (milligrams per kilogram of body weight per day), and CSF is the cancer slope factor, or the risk associated with a unit dose of a carcinogen, also called the cancer potency factor (mg/kg-day)-1. The CSF usually represents an upper bound estimate of the likelihood of developing cancer, based on animal data (footnote 16).
Assuming a linear dose-response relationship has major implications for regulating carcinogens because it indicates that even very low exposure levels can be hazardous and thus may need to be controlled. However, cancer research findings over the past several decades indicate that some carcinogens may act in non-linear ways. For example, radon damages the DNA and RNA of lung cells, but the long-term risk associated with exposure to radon is much higher for smokers than for non-smokers, even if their exposures are the same. Another chemical, formaldehyde CSF, is under review by EPA because it has been shown that before animals exposed to high doses developed cancer, they developed ulcerations in their mucous membranes. This observation suggests that lower concentrations of formaldehyde CSF, a water soluble compound, had a different potency factor than higher concentrations.
Further complicating the issue, juvenile test animals are more susceptible to some cancer causing compounds than adult animals of the same species. EPA's cancer risk guidelines now reflect this difference. On the other hand, it is understood that the human body's ability to repair damaged DNA diminishes with age. Age-dependent cancer slope factors are not available for the hundreds of suspected cancer causing compounds, so the unit risk factors are assumed to apply uniformly over a lifetime, except where observations support a different risk for infants and children.
These questions can influence what type of model scientists use to calculate dose-response relationships for carcinogens, or even whether carcinogens are treated similarly to non-cancer endpoints with presumed population thresholds (as described below). A common model for dose-response for carcinogens is the so-called one-hit model, which corresponds to the simplest mechanistic explanation of cancer—that a single exposure to a dose as small as a molecule would have a non-zero probability of changing a normal cell into a cancer cell. Researchers typically use this model to analyze pollutants that are hypothesized to operate under this mode of action or as a default model in the absence of mechanistic evidence.
In contrast, multi-stage models (of which the one-hit model is a special case) assume that a cell passes through several distinct phases that occur in a certain order as it becomes cancerous. It is hard to determine empirically which model is more appropriate, so this choice relies on understanding the mode of action of the compound. Because CSF values are sensitive to these assumptions, EPA's newest carcinogen risk guidelines (issued in 2005) focus on finding a point in the range of observed data, called a point of departure, which is less sensitive to model choice. For compounds that are direct mutagens or with substantial background processes, linearity is assumed below the point of departure, while non-linear approaches are used if suggested by the mode of action.