9.10. KEY ASSUMPTIONS AND INFERENCES

One of the primary functions of the risk characterization is to present key assumptions and inferences that are used to reach conclusions in the absence of definitive information. Not all scientists may agree with the use of these specific assumptions and inferences. The degree to which there is disagreement will have profound effects on the acceptance of this analysis. While many of these assumptions and inferences are discussed in previous sections, it is important that they be recognized in order to put our overall conclusions in a proper perspective. Key assumptions and inferences are listed below.

The limited information on sources, fate, and transport in the environment provides a reasonable basis for predicting human exposure. While data are limited and, therefore, uncertain, information from a variety of studies in industrialized countries coupled with our detailed knowledge of physicochemical properties for this class of compounds allows reasonable assumptions to be made regarding relative ranking of sources with regard to their contribution to environmental loading, the persistence of this class of compounds under specific environmental conditions, and the likelihood that the chemical will be transferred from the environment to biological systems. Nonetheless, these are assumptions that are arguable and that will be refined as more data become available. Additional data will be required to validate the numerous hypotheses that go into assembling models for environmental release, fate, and transport for this complex mixture of individual chemical congeners.

The air to food hypothesis is plausible and is supported by enough data to warrant its use in the absence of more complete information. The air-to-food hypothesis is founded on data evaluating deposition, environmental transport, bioaccumulation, and consumption patterns. It is supported by studies from Europe and Canada. While individual measurement data are still quite limited, the consistency of the evidence supporting the validity of the hypothesis is compelling. The hypothesis has been accepted by a large segment of the knowledgeable scientific community. Because airborne dioxin may come from direct releases to air or from recycling of dioxin-like compounds released into various environmental media from a number of sources, this hypothesis provides a perspective on how dioxin-like compounds move through the environment to humans but does not allow attribution of exposure to particular sources.

Toxicity equivalence is a valid, interim method for assessing exposure to a complex mixture of dioxin and related compounds and predicting likely health outcomes. The EPA and the international scientific community have agreed that the use of toxicity factors to predict relative toxicities of mixtures of this class of compounds has an empirical basis, is theoretically sound, and, in the absence of more complete data sets on the toxicity of individual members of this class, is a useful procedure. This is not to say that the use of TEFs is a certain procedure. Since 1986 when the first Agency-wide consensus on the use of TEFs was published, additional refinements to the data bases and to the use of TEFs have occurred. Published revisions in accord with international agreement appeared in 1989. In the course of this reassessment, critical data were collected, and agreement was reached regarding the contribution of dioxin-like PCBs to overall TEQs. Additional validation of the TEQ concept in predicting effects of this class of compounds on wildlife species lends further support to the use of this approach. It must be recognized that this relatively simple, additive approach does not take into account interactions between dioxin-like compounds and other chemical exposures. These interactions may result in either an overestimate or an underestimate of likely effects of the complex mixture. While generally accepted as useful for evaluating intakes of various dioxin-like compounds, the application of this approach to the evaluation of measured body burdens remains even more uncertain.

Use of one-half the nondetect level for estimating low levels of exposure is a reasonable but conservative approach to evaluating limited blood and tissue level data. For some data sets, use of zero values for nondetects could result in significantly lower estimates and, therefore, use of the current procedure may be overestimating blood or tissue levels. However, it is widely held that use of zero values for nondetects would most likely underestimate true levels of exposure, particularly where nondetects do not dominate measured values. Similar estimates of TEQs derived from different data sets, developed by different investigators in several countries, strengthen the probability that this inference represents the exposure of the general population in industrialized countries to dioxin and related compounds.

The limited data available from studies of levels of dioxin and related compounds in humans provide an adequate basis to infer general population body burdens. Although there are still limited measurements of general population body burdens, the data provide a consistent picture of background body burdens for industrialized countries. While additional data will help refine the range of general population body burdens as a function of location, human activity, age, and the like, there are adequate data to estimate current body burdens in the general population for the purposes of this assessment. It is highly unlikely that these estimates would represent a sensitive parameter in estimating margins of exposure within an order of magnitude.

Laboratory animal studies provide useful information in evaluating potential human responses to dioxin and related compounds. Based on our knowledge of the biochemical and biological similarities between laboratory animals and humans, our understanding of some of the fundamental impacts of this class of compounds on biological systems, and comparable responses from animal and human studies both in vitro and in vivo, our decision to use laboratory animal data to contribute to weight-of-the-evidence conclusions on human hazard and risk is reasonable. Humans do not appear to be an unusual responder for dioxin effects, that is, we do not, on average, appear to be either refractory to or exquisitely sensitive to the effects of dioxin-like compounds. While positive human data are preferable for ascribing hazard or risk, the lack of adequate human data to demonstrate causality for many suspected dioxin effects is assumed not to negate the findings from laboratory animal and in vitro studies. Although some scientists may disagree, in our estimation, the data base on dioxin and related compounds is one of the most comprehensive among all environmental chemicals. The fundamental understanding of mechanisms of dioxin action provides a unifying theory for the mechanisms for observed effects in laboratory animals and humans and for using a weight-of-the-evidence approach considering all relevant data to infer the human health impacts of dioxin and related compounds.