TERA

ATSDR's qualitative conclusions regarding carcinogenicity are presented in the Toxicological Profiles using a weight of evidence approach. This approach relies upon NTP's Annual Report on Carcinogens. Conclusions from IARC, U.S. EPA and OSHA are also presented as appropriate.

Health Canada classifies chemicals into six categories with regard to carcinogenicity based on a modification of the scheme used by the International Agency for Research on Cancer. The following is excerpted from Human Health Risk Assessment for Priority Substances (Health Canada, 1994):

Data from adequate epidemiological studies indicate that there is a causal relationship between exposure to a substance and an increased incidence of cancer in humans.

Data from epidemiological studies are inadequate to assess carcinogenicity. However, there is sufficient evidence of carcinogenicity in animal species (i.e., there is an increased incidence of malignant tumours in multiple species or strains, in multiple experiments with different routes of exposure or dose levels, or the incidence, site or type of tumour or age of onset is unusual). Exceptionally, a compound for which the evidence of carcinogenicity is limited but for which there is a strong supporting dataset (on genotoxicity, for example) which indicates that the compound is likely to be carcinogenic would be included in this category.

Health Canada has three subgroups in Group III which describe the data in humans and laboratory animals that would result in a classification in this Group. In summary, this includes chemicals for which data from epidemiological studies are inadequate, or which indicate an association between exposure and human cancer but alternative explanations such as chance, bias or confounding cannot be excluded. There is some evidence of increased tumour incidence in animals but the data are limited because the studies involve a single species, strain or experiment; study design (i.e., dose levels, duration of exposure and follow-up, survival, number of animals) or reporting is inadequate; the neoplasms produced often occur spontaneously and have been difficult to classify as malignant by histological criteria alone (e.g., lung and liver tumours in mice). The weight of limited evidence indicates that the compound is genotoxic or results are mixed. Chemicals believed to have an epigenetic mechanism of cancer induction may also be classified in Group III if there are positive cancer studies in long-term animal experiments.

Health Canada has four subgroups in Group IV which describe the data in humans and laboratory animals that would result in a classification in this Group. In summary, this includes chemicals for which there is no evidence of carcinogenicity in adequate epidemiological studies or data are inadequate. There is some evidence of carcinogenicity in well -designed and well-conducted carcinogenicity bioassays in animals, but the results are limited or can be confidently ascribed to species-specific mechanisms of toxicity and/or metabolism which do not appear to be operative in humans.

Health Canada has three subgroups in Group V which describe the data in humans and laboratory animals that would result in a classification in this Group. In summary, this includes chemicals for which there is no evidence of carcinogenicity in sufficiently powerful and well-designed epidemiological studies; there is no evidence or inadequate data on carcinogenicity in laboratory animals.

Health Canada has three subgroups in Group VI which describe the data in humans and laboratory animals that would result in a classification in this Group. In summary, this includes chemicals for which data from epidemiological and/or animal studies are inadequate or not available.

An explanation of IARC's methods is available in the Preamble to the IARC Monographs at http://monographs.iarc.fr/monoeval/preamble.html. The Preamble to the Monographs sets out the objective and scope of the evaluation programme, the procedures used when making assessments, and the types of evidence considered and criteria used in reaching the final evaluations.

NSF International currently uses the U.S. EPA (2005) weight of evidence narrative approach to cancer classification. The conclusion reached by NSF is included as part of the hazard assessment in a weight of evidence evaluation and cancer characterization section of the oral risk assessment document. If the U.S. EPA or another internationally recognized organization such as the NTP (National Toxicology Program), ATSDR, Health Canada, IARC or other members of the World Health Organization has also classified the chemical, that classification will be included in the risk comparisons and conclusions section of the NSF document, with discussion if the classifications differ. The U.S. EPA and NSF International classifications may occasionally differ if new data have been evaluated by one of the organizations.

An explanation of RIVM's risk assessment methods is available in the following report:

Janssen, PJCM and GJA Speijers. 1997. Guidance on the Derivation of Maximum Permissible Risk Levels for Human Intake of Soil Contaminants. National Institute of Public Health and the Environment. Bilthoven, The Netherlands. January. Available at http://www.rivm.nl/bibliotheek/rapporten/711701006.pdf or at http://www.rivm.nl/index_en.html (click on Search, type "711701006", then click on document).

In 1986, the U.S. EPA published general guidelines to be used by Agency scientists in developing and evaluating risk assessments for carcinogens (U.S. EPA, 1986). Almost all of the carcinogen assessments on IRIS were based on the 1986 guidelines. Assessments developed between 1996 and approximately 1999 may have used the 1996 proposed guidelines; and assessments developed between approximately 1999 and early 2005 may have used the 1999 draft guidelines. Both the 1996 and 1999 versions were similar to the 2005 (final) version, and used comparable quantitative approaches. However, the 1996 version included two fewer categories, and both 1996 and 1999 versions differed in some other details from the 2005 guidelines. For more details about the evolution of U.S. EPA’s cancer guidelines, please see http://cfpub.epa.gov/ncea/raf/recordisplay.cfm?deid=116283.

Below is a brief description of the weight of evidence and cancer classification guidelines from the 1986 guidelines. This is followed by a brief description of the 2005 guidelines.

The 1986 guidelines specify that information be categorized into one of three types: human data, animal data, and supporting data. The human and animal data are used to make a preliminary judgment as to the likelihood that the agent in question may produce tumors in humans. The supporting data (e.g., genotoxicity, mechanistic data, and pharmacokinetic information) are then used to elevate or downgrade the classification. For a description of the amount and type of data required for a chemical to be assigned to any one of these groups, the reader is referred to the 1986 guidelines (U.S. EPA, 1986). In brief, the categories from the 1986 guidelines, as defined by U.S. EPA, are as follows:

Classification in Group A requires the observation of a statistically significant association between exposure to an agent and malignant or life-threatening benign tumors in humans.

EPA divides this group into the categories B1 and B2. Limited human evidence of carcinogenicity in humans is necessary for placement of a chemical in Group B1. Group B2 includes chemicals with sufficient animal evidence, but inadequate human evidence for carcinogenicity.

An agent is classified in Group C when human data are inadequate and animal data demonstrate limited evidence of carcinogenicity (e.g., an increased incidence of benign tumors only; a positive finding of carcinogenicity in one species only; an increased incidence of neoplasms that occur with high spontaneous background incidence)

An agent is classified in Group D when insufficient data are available to make a determination as to carcinogenicity.

An agent is classified in Group E if there is no increased incidence of neoplasms in at least two well-designed and well-conducted animal studies of adequate power and dose in different species.

The 2005 guidelines (U.S. EPA, 2005) significantly change the way hazard evidence is weighed in reaching conclusions about an agent's potential for human carcinogenicity. Tumor findings in animals or humans dominated the 1986 classification scheme. Under the 2005 guidelines, decisions are based on all of the evidence, particularly information regarding mode of action at cellular and subcellular levels, as well as toxicokinetics and metabolic processes. Weighing of the evidence includes considering the likelihood of human carcinogenic effects of the agent and the conditions under which such effects may be expressed, as these are revealed in the toxicological and other biologically important features of the agent. This more complete characterization of the expression of carcinogenic potential might include findings that an agent is observed to be carcinogenic by one route, but not another. Alternatively, the agent's carcinogenic activity might be secondary to another toxic effect.

The 2005 guidelines use standard descriptors of conclusions rather than letter designations. The descriptors are incorporated into a brief narrative that explains an agent’s human carcinogenic potential and the conditions that characterize its expression. Significant issues, strengths, and limitations of the data and conclusions are included. The narrative also summarizes the mode of action information that underlies the approach to dose-response assessment. Five categories of descriptors are used, with additional text further defining the conclusion. In brief, the descriptors from the 2005 guidelines are:

This descriptor is appropriate when there is convincing epidemiologic evidence demonstrating causality between human exposure and cancer. EPA also considers this descriptor to be appropriate when there is an absence of conclusive epidemiologic evidence to clearly establish a cause and effect relationship between human exposure and cancer, but a number of other criteria are met. The criteria are (1) strong evidence of an association between human exposure and either cancer or key precursor events, (2) extensive evidence of carcinogenicity in animals, (3) the mode(s) of action and key precursor events have been identified in animals, and (4) there is strong evidence that the key precursor events are anticipated to occur in humans and progress to tumors.

This descriptor is appropriate when the available tumor effects and other key data are adequate to demonstrate carcinogenic potential to humans. Adequate data are within a spectrum. At one end is evidence for a plausible (but not definitively causal) association between human exposure to the agent and cancer, usually with some supporting evidence (not necessarily carcinogenicity data) in animals. At the other end of the spectrum is an agent with no human data, but a positive tumor study in animals and the weight of experimental evidence shows that in experimental animals the agent causes events generally known to be associated with tumor formation.

This descriptor is appropriate when the weight of evidence from human or animal data is suggestive of carcinogenicity; a concern for carcinogenic effects in humans is raised, but is judged not sufficient for a stronger conclusion. Examples of such evidence may include: (1) a small and possibly not statistically significant increase in tumors in a single study that is not contradicted by other studies of equal quality in the same system, or (2) a small increase in a tumor with a high background rate in that sex and strain, when there is some evidence that the observed tumors may be due to intrinsic factors. Dose-response assessment is generally not indicated for these agents.

“Data are Inadequate for an Assessment of Human Carcinogenic Potential”

This descriptor is used when available data are judged inadequate to perform an assessment. This includes a case when there is a lack of pertinent or useful data or when existing evidence is conflicting, e.g., some evidence is suggestive of carcinogenic effects, but other studies of equal quality in the same sex and strain are negative.

ATSDR does not currently perform dose-response assessments for carcinogens. This Agency does, however, report values established by other Agencies (e.g., U.S. EPA, IARC).

ATSDR does not currently engage in low-dose modeling efforts or in the development of cancer potency factors (ATSDR 1993).

For substances considered by Health Canada to have no threshold (i.e., mutagens and genotoxic carcinogens), it is assumed that there is some probability of harm to human health at any level of exposure. For these chemicals, Health Canada considers it inappropriate to specify a concentration or dose associated with a negligible or de minimis level of risk (e.g., the1 in a million risk often used by U.S. EPA) by low-dose extrapolation procedures. Rather, potency is expressed as the dose or concentration which induces a 5% increase in the incidence of, or deaths due to, tumours or heritable mutations considered to be associated with exposure. The TD05/TC05 is then compared with exposure levels. If the ratio between exposure and the TD05/TC05 is less than 2 x 10-6, there is little need for further action. If the ratio is 2 x 10-4 or greater, there is a high priority for further action. Values in between are of moderate priority.

In order to compare cancer potencies estimated by different Agencies, TERA chose to express each Agency's potency value as the equivalent of a 1 in a 100,000 risk level. For Health Canada, this required dividing the TD05/TC05 (i.e., a 1 in 20 risk level) by a factor of 5,000 to represent a 1 in a 100,000 risk level. It is noted, however, that unlike the methodology used by U.S. EPA, Health Canada's TD05/TC05 is not based on a confidence limit, but is computed directly from the dose-response curve within or close to the experimental range. Health Canada considered this to be appropriate in view of the stability of the data in the experimental range and to avoid unnecessarily conservative assumptions.

NSF International currently uses U.S. EPA (2005) dose-response assessment methodology. Earlier documents have used U.S. EPA (1999) draft, U.S. EPA (1996) proposed or U.S. EPA (1986) final guidelines. Specific implementation of this methodology is described in Annex A of NSF International/American National Standard 60 “Drinking water treatment chemicals – Health effects,” and of NSF International/American National Standard 61 “Drinking water system components – Health effects”. For a tumor endpoint, human equivalent doses are first calculated by scaling the applied daily doses to body weight raised to the 0.75 power. The dose-response data are then subject to benchmark dose modeling (U.S. EPA, 1995) to determine the point of departure, which is generally the 95% confidence limit on a dose associated with an estimated 10% increased tumor or related non-tumor response (the LED10). If the data cannot be modeled, a NOAEL or LOAEL may be used as the point of departure. If the weight of evidence indicates that the compound is genotoxic, the dose-response assessment is performed by linear extrapolation from the point of departure to a specific risk level. If there is a plausible mode of action that indicates the tumor or tumor precursor is not produced by a genotoxic mechanism, a margin-of-exposure approach may be used.

An explanation of RIVM's risk assessment methods is available in the following report:

U.S. EPA published guidelines for carcinogen risk assessment in 1986 (U.S. EPA, 1986). These guidelines outline procedures for estimating cancer potency. Almost all of the carcinogen assessments on IRIS were based on these 1986 guidelines. In 1996, EPA proposed revisions to the cancer guidelines (U.S. EPA, 1996), and these were further modified in the draft 1999 guidelines (U.S. EPA, 1999), and were then finalized in the 2005 guidelines (U.S. EPA, 2005). Assessments developed between 1996 and approximately 1999 may have used the 1996 proposed guidelines; and assessments developed between approximately 1999 and early 2005 may have used the 1999 draft guidelines. For more details about the evolution of U.S. EPA’s cancer guidelines, please see http://cfpub.epa.gov/ncea/raf/recordisplay.cfm?deid=116283.

Below is a brief description of EPA dose-response procedures based on the 1986 guidelines and an explanation of how TERA expresses the results on ITER for comparison purposes. This is followed by a brief description of the 2005 guidelines.

Two extrapolations are generally necessary when using animal data. The first step is extrapolation from animals to humans. According to the 1986 guidelines, this extrapolation is done by estimating a human equivalent oral dose, by scaling the daily applied doses to body weight raised to the 0.66 power. Second, one needs to extrapolate from the high doses used in animal studies to the generally lower doses of interest for environmental exposure. Risk at low exposure levels generally cannot be measured directly (either by animal experiments or by epidemiologic studies). Therefore, a number of mathematical models and procedures have been developed for use in extrapolating from high to low doses. Under EPA's 1986 cancer risk assessment guidelines, the linearized multistage model was generally chosen as the default model for extrapolation to low doses. Multistage models are exponential models approaching 100% risk at high doses, with a shape at low doses described by a polynomial function. The multistage model is fit to the tumor dose-response data, and an upper bound for the risk is estimated by incorporating an appropriate linear term into the statistical bound for the polynomial. At sufficiently small exposures, any higher-order terms in the polynomial will contribute negligibly, and the graph of the upper bound will appear to be a straight line. The slope of this line (formerly called the potency) is called the slope factor. Its units are (proportion of individuals with tumors)/mg/kg-day.

For the oral route, EPA calculates both a slope factor and a unit risk. As described above, the oral slope factor expresses the risk per mg/kg-day. The unit risk is a numerically equivalent term that is expressed as the risk associated with a drinking water concentration of 1 ug/L (with assumptions being made that an adult weighs 70 kg and drinks 2 L/day). For the route of inhalation, EPA does not provide a slope factor, but rather expresses the risk only in terms of a unit risk. The units for the inhalation unit risk are risk per 1 ug/m³. In other words, it is the risk associated with an air concentration of 1 ug/m³ (assuming a 70 kg adult breathes 20 cubic meters/day).

In order to compare cancer potencies estimated by different Agencies, TERA chose to express each Agency's potency value as the equivalent of a 1 in a 100,000 risk level. Thus, TERA calculates risk specific doses (RSDs) from EPA's oral slope factors and risk specific concentrations (RSCs) from EPA’s inhalation unit risks. Specifically, for oral slope factors, TERA converts the EPA risk estimate to a concentration at the 1 in 100,000 (E-5) risk level by dividing 1E-5 by the unit risk [in units of “per (ug/m³)”] and then by another 1000 to convert to mg/cu.m to determine a risk specific concentration (RSC) (in units of “mg/ m³”). Similarly, TERA converts the EPA oral slope factors to a dose at the 1 in 100,000 (E-5) risk level by dividing 1E-5 by the slope factor [in units of “per (mg/kg-day”] to determine a risk specific dose (RSD) (in units of “mg/kg-day”).

In setting standards for carcinogens, EPA generally considers a de minimis (e.g., less than or equal to 1 in a million) risk to be an acceptable goal. Using the output from the linearized multistage model, EPA often determines the oral intake or inhalation concentration that is associated with a risk of 1 in a million as a goal for setting limits on exposure. Risk management issues may lead to the setting of intakes/concentrations that are higher or lower.

The 2005 cancer guidelines (U.S. EPA, 2005) differ significantly from the 1986 guidelines. When animal studies are used, the estimation of a human equivalent dose utilizes toxicokinetic models when available, and if not, the default for oral doses is to scale the daily applied doses to body weight raised to the 0.75 power. The default dose scaling methodology for inhalation follows that developed for derivation of reference concentrations (RfCs), estimating the relative animal and human respiratory deposition of particles, and the relative internal dose or dose to the respiratory region of gases, depending on the chemical and physical properties of the gas.

Response data from effects of the agent on carcinogenic processes (i.e., nontumor data) are analyzed along with tumor incidence data. Tumor incidences and precursor effects may be combined to extend the dose-response curve below the tumor data. A biologically based or case-specific dose response model to relate dose and response data in the range of empirical observation may be used when data are sufficient. When this is not the case, standard default procedures are used to fit a curve to the data and to calculate the lower 95% confidence limit on a dose associated with an estimated 10% increased tumor or relevant nontumor response (LED10). The LED10 then serves as a point of departure for extrapolating outside the observable range. Depending on the mode(s) of action of the agent, low-dose extrapolation from the LED10 is done using a linear approach, a nonlinear approach, or both. Linear extrapolation to low doses is used when the mode of action data indicates that the agent is DNA-reactive and has direct mutagenic activity, or if the human exposure or body burden is high and near doses associated with key precursor events. Linear extrapolation is also used as a default when there are insufficient data to evaluate mode of action. For linear extrapolation, a straight line is drawn from the point of departure to zero dose, zero response, corrected for background. The slope of the line expresses the extra risk per unit dose. This risk can be converted to the risk specific dose or risk specific concentration, as described for the 1986 guidelines. A nonlinear extrapolation is used when there a tumor mode of action supporting nonlinearity applies, and the chemical does not demonstrate mutagenic effects consistent with linearity. Alternatively, a nonlinear extrapolation may also be used when the data support a nonlinear mode of action, and there is a suggestion of mutagenicity, but the data justifies the conclusion that mutagenicity is not operative at low doses. The guidelines present criteria (based on a modification of the Hill criteria for evaluation of epidemiology data) for evaluation of potential modes of action. When the nonlinear extrapolation is used, an RfD- or RfC-like value is derived using standard methods. Mode of action analysis is critical to the 2005 draft guidelines. This emphasis will bring new research on carcinogenic processes to bear in assessments.

The 2005 guidelines also include supplemental guidance for assessing susceptibility from early-life exposure to carcinogens. This guidance states that particular attention should be paid to the potential for higher potency from early-life exposure. Mode of action data should also be evaluated for age-specific differences. If chemical-specific data are available to evaluate the age-specific potency, those data should be used. If chemical-specific data are not identified, but the chemical acts via a mutagenic mode of action, age-dependent adjustment factors are used.

Health Canada. 1994. Human Health Risk Assessment for Priority Substances. Environmental Health Directorate. Canadian Environmental Protection Act. Health Canada, Ottawa, 1994.

U.S. EPA. (Environmental Protection Agency). 2005. Guidelines for Carcinogen Risk Assessment. Washington, DC, National Center for Environmental Assessment. EPA/630/P-03/001b. NCEA-F-0644b. Available online at http://www.epa.gov/cancerguidelines.

U.S. EPA (Environmental Protection Agency). 1999. Guidelines for Carcinogen Risk Assessment. Risk Assessment Forum. NCEA-F-0644. July 1999. Available at http://www.epa.gov/ncea/raf/crasab.htm.

U.S. EPA (Environmental Protection Agency). 1996. Proposed Guidelines for Carcinogen Risk Assessment. 61 Federal Register pp17960-18011. April 23. Available at http://cfpub.epa.gov/ncea/raf/cra_prop.cfm.

U.S. EPA. 1986. The risk assessment guidelines of 1986. Office of Health and Environmental Assessment, Washington, DC. EPA 600/S9-85/001F.

Cancer Risk Assessment Methods