Principles for the selection of doses in chronic rodent bioassays. ILSI Risk Science Working Group on Dose Selection.

Dose selection in chronic rodent bioassays has been one of the most debated issues in risk assessment. The Committee on Risk Assessment Methods of the National Research Council attempted, but failed, in 1993 to reach consensus on how to select doses for chronic rodent bioassays. However, a more recent effort conducted by the ILSI Risk Science Institute has resulted in a consensus set of principles for dose selection, including selection of the highest dose for chronic rodent bioassays. The principles encourage a move away from sole reliance on a maximum tolerated dose (MTD), as it has been traditionally defined (primarily by body weight and histopathology), and toward the use of sound scientific and toxicologic principles for the selection of all doses in the chronic bioassay. Specifically, the principles recommend that dose selection for chronic studies must be based on sound toxicologic principles; dose selection should consider human exposure; dose selection should be based on a variety of endpoints and effects derived from prechronic studies; and dose selection should consider physicochemical and other factors. Implementation of the principles internationally will have two important benefits; improvement in the quality and consistency of the rodent bioassay and international harmonization of dose selection procedures.

Dose selection in chronic rodent bioassays has been the focus of a great deal of attention and debate since the 1960s (1)(2)(3)(4)(5). A particularly contentious issue has been selection of the highest dose (typically the maximum tolerated dose; MTD) in carcinogenicity assays (6). In the 1960s, when chronic animal bioassays were first used routinely to evaluate chemicals, the primary objective was to assess the qualitative potential for toxicity and carcinogenicity. For this purpose, the highest dose was selected to elicit a toxic but not life-threatening effect in a relatively small number of animals.
With the advent of formal risk assessment methods in the late 1970s and early 1980s, studies that were originally intended to assess the potential for toxicity or carcinogenicity qualitatively were used to quantify adverse impacts; that is, to estimate the shape and slope of the dose-response curve (7). Despite changes in interpretation and use of data from chronic bioassays as well as improvements in our understanding of the mechanisms by which chemicals may induce cancer, criteria for dose selection (particularly the highest dose) have remained remarkably constant (2,(7)(8)(9)(10)(11).
Differences in dose selection procedures that exist between many developed countries have also been problematic, particularly where differences have led to increased potential for trade barriers as a result of rejection of data from chronic bioassays designed to meet the requirements of one country and submitted for regulatory purposes in another (12). The International Conference on Harmonization (5) has agreed on several options for selecting the highest dose in chronic bioassays for pharmaceuticals; however, no such consensus exists for pesticides, industrial chemicals, food additives, and other nonpharmaceutical chemicals.
In 1993, the ILSI Risk Science Institute formed a working group on dose selection for chronic animal bioassays as one of the activities under its cooperative agreement with the EPA Office of Pesticide Programs.
Recognizing the dissatisfaction with current approaches to dose selection, the goal of the working group was to improve the testing of chemicals for long-term toxicity and carcinogenicity through appropriate dose selection and to work toward international harmonization of dose selection procedures for chronic bioassays. The working group included scientists from academia, industry, and government representing the United States, Canada, Europe, and Japan. This article sets forth a set of principles for dose selection in chronic rodent bioassays. The principles, which represent the consensus of working group members, encourage a move away from sole reliance on an MTD as it has been traditionally defined (primarily by body weight and histopathology) and toward the use of sound scientific and toxicologic principles for the selection of the highest as well as lower doses in the chronic bioassay. The principles do not, however, address the number of doses (typically three plus a control) in a chronic bioassay. Increasing the number of doses beyond three plus a control may shed light on the shape and slope of the dose-response curve; however, there is general resistance to increasing the number of doses because of cost con-straints. Further, doses selected for chronic studies may depend on or be influenced by species, strain, route and mode of administration, diet, and other factors, and there are critically important issues associated with the interpretation of results of chronic bioassays in the context of risk assessment. Despite the importance of these issues, the principles developed by the working group and presented in this paper focus solely on dose selection for chronic rodent bioassays.

Principles for Dose Selection
Principle #1 Dose selection for chronic studies must be based on sound toxicologic principles, e.g., see Klaassen et al. (13). Within a reasonable dose range, increasing the dose can increase the ability to detect an effect; therefore, doses for chronic rodent bioassays should be selected within this range to maximize the sensitivity of a chronic bioassay. However, trying to increase study sensitivity by increasing doses into ranges that do not reflect application of sound toxicologic principles could lead to results that are inappropriate for human risk assessment.
Increasing the highest dose in a chronic bioassay may increase sensitivity within some defined dose range, but the potential exists that different mechanisms of toxicity or chemical modes of action are active at higher doses, which may not be relevant to humans exposed to lower doses. The working group encourages an approach to dose selection that incorporates all relevant information from prechronic studies and other sources, uses a wide range of toxicologic Address correspondence to J.A. Foran, ILSI Risk Science Institute, 1126 16th. St., NW, Washington, D.C. 20036 USA. The principles for dose selection are a distillation of 3 years of work by many people, most prominently the members of the ILSI Risk Science Institute working group on dose selection. We thank A. Gasper for providing secretarial support throughout the project, S. Carter and D. Dalisera for administrative support, and G. Scarano for technical support. This project was conducted under a cooperative agreement with the U.S. Environmental Protection Agency, Office of Pesticide Programs. The financial support of the EPA and the American Crop Protection Association is gratefully acknowledged. While the principles represent the consensus of the working group members, they do not necessarily represent the policies, positions, or opinions of their respective agencies and organizations. tools to understand the mechanisms or modes of action by which a chemical produces an effect (e.g., genotoxicity, cell proliferation, etc.), and uses good scientific principles to enhance the accuracy of judgments of potential human risks.

Principle #2
Scientists who conduct chronic bioassays and those who use data from bioassays, induding regulatory agencies, should encourage innovative approaches to dose selection by considering appropriate study designs, mechanistic data, and other information in the design and interpretation of studies. Use of additional endpoints and other information must be based on sound scientific rationale, and such designs should be evaluated based on their individual merits.
A goal of high dose selection in carcinogenicity bioassays is, in the context of hazard identification, to reduce the likelihood of a false negative result; however, it is recognized that the qualitative nature of the hazard (e.g., carcinogenic response) may itself be dose dependent. This principle encourages approaches to dose selection that incorporate consideration of mechanistic and other toxicologic information. Such approaches should improve the scientific basis for dose selection and aid in interpretation of data generated from chronic bioassays.

Principle #3
Selection of the middle and lower doses should take into account factors, such as the mechanism or mode of action, toxicokinetics, and others listed in Principles #4 and #5, and should not be based solely on a fraction of the highest dose. Further, the middle and lowest doses should be selected to characterize the shape of the dose-response curve as much as possible. Human exposure should also be considered in dose selection, particularly for the selection of the middle and lowest doses.
For substances expected to exhibit a toxicity threshold, or if the evaluation of carcinogenic potential is being combined with an evaluation of chronic toxicity, the study should be designed to include one dose that does not elicit adverse effects, i.e., one dose should be a no observed adverse effect level (NOAEL). Of course, caution must be exercised to ensure that the NOAEL is not simply an artifact of small sample size or poor study design.
Where human exposure influences dose selection, issues that should be considered include the exposure route and mode, the dose range in the chronic bioassay in relation to human exposure, and the duration and frequency of human exposure, if known. Subpopulations that may be more highly exposed than the general population, or that are genetically more susceptible, should also be considered.

Principle #4
The working group recommends the use of innovative approaches and endpoints and other information in the selection of doses for chronic rodent bioassays. The following endpoints, which can be assessed in prechronic studies, should be considered in dose selection for chronic rodent bioassays.
Histopathology. The site, morphology, and severity of the treatment-related effects observed in prechronic studies are critical in dose selection. Histopathologic examination of tissues, especially the liver, gastrointestinal tract, urinary tract, respiratory tract, skin, spleen/bone marrow/blood, and endocrine tissues is recommended.
Toxicokinetics. Consideration of the effect of dose (or exposure concentration) on absorption, tissue distribution, metabolism, and clearance of a compound is recommended. Cellular growth. Information on the dose-dependence of regenerative cell proliferation, induced mitogenesis, and apoptosis will be a useful adjunct to histological observations in determining the shapes of organ-specific toxic response curves; such information will be of significant value in selecting high, middle, and low doses, and in interpreting the results of the study.
Physiological function. Disturbances of physiology or homeostasis that would compromise the validity of the study must be considered in the dose selection process. Examples include hypotension, inhibition of blood clotting, overwhelming normal pulmonary clearance mechanisms, immune system effects, and, in some cases, hormonal imbalance.
Clinical chemistry, hematology, and urinalysis. These endpoints are best used to support dose selection decisions based on other criteria/parameters. Changes in serum clinical chemistry in the absence of histopathologic observations may not affect high dose selection, but may complement dose selection decisions based on toxicokinetics, cell proliferation, and other parameters. However, when hematological tissues are determined to be a target organ in prechronic studies, hematology results may be an appropriate basis for dose selection.
Organ weights. This endpoint is not often the critical factor in the selection of doses for chronic rodent bioassays. Chemically induced changes in organ weights should, however, be considered in conjunction with other data in the dose selection process. Body weight. If body weight changes are the primary factor in the selection of the highest dose group (that is, when no other toxic effects are observed in prechronic studies), a decrement in body weight gain of no more than 5-10% in prechronic studies should be used in the selection of the highest dose for chronic assays of carcinogenicity.
While many of these endpoints are not presently assessed in typical prechronic studies, it will be necessary to begin to gather such data so that dose selection may draw from a greater base of toxicologic data. It should also be recognized that not all of these endpoints may be useful or necessary for every compound and that other endpoints, where they are based on sound toxicologic principles, may provide important information for dose selection.

Principle #5
Physicochemical factors (e.g., solubility, vapor pressure), the bioavailability of the compound, the palatability of the compound in food or drinking water, and other factors such as the potential for the substance to cause adverse effects at the site of administration (e.g., irritation, erosion, and ulceration) will influence the selection of the highest dose for chronic rodent bioassays. It is recommended that doses for chronic rodent bioassays be selected to minimize or avoid adverse nutritional, physical, organoleptic, and irritant effects.

Conclusions
These principles do not provide detailed instructions for dose selection. Rather, they encourage the selection of doses that maximize study sensitivity and concurrendy produce results that are biologically and toxicologically credible. The working group agreed that, at the completion of the bioassay, one should look retrospectively at the doses that were selected and the data generated at those doses. Data obtained from doses that are inconsistent with the principles presented above may not be appropriate for use in human health risk assessment. Implementation of these principles will have two important benefits. First, implementation will improve the quality and consistency as well as the interpretation and utilization of data from the rodent bioassay, our most important tool to assess the longterm toxicity and carcinogenicity of chemicals, leading to a better assessment of human risk. Second, international harmonization of dose selection procedures will reduce the potential for trade barriers that may occur as a result of rejection of data designed to meet the requirements of one country and submitted for regulatory purposes in another.