Data considerations for regulation of water contaminants.

There are several pieces of legislation based on human health assessment that set the framework for U.S. EPA's regulation of water contaminants, such as bromate. The Safe Drinking Water Act, for example, specifies that the best available science be used in support of regulation of drinking water contaminants, and highlights that regulations must provide protection to sensitive human populations. Recent EPA guidance, including the 2005 Cancer Guidelines, emphasize analyzing data, and using defaults only in the absence of adequate data. This represents a major shift from the former practice of invoking default methodologies or values unless it was judged that there were sufficient data to depart from them. The Guidelines further present a framework for assessing data in order to determine if a mode of action (MOA) can be established, based on a modification of the Bradford-Hill criteria for causality. A similar approach is used by the International Programme on Chemical Safety (IPCS). To illustrate the application of the framework for evaluating animal tumors, three case studies are considered here. In the first example (chloroform carcinogenicity), sufficient data exist to identify the MOA in animals, and the data are used to illustrate the evaluation of the plausibility of the animal MOA in humans, taking into account toxicokinetics and toxicodynamics. In this case, the MOA was judged to be relevant to humans, and was used to determine the approach for the cancer quantitation. In the second example (naphthalene inhalation carcinogenicity), the key question is whether the weight of evidence (WOE) is sufficient to establish the MOA in animals. Atrazine-induced mammary tumors form the final example, illustrating the reasoning used to determine that the tumor MOA in animals was not considered relevant to humans; atrazine is therefore considered not likely to be a human carcinogen.