The role of mechanistic data in dose-response modeling.

The results from employing metabolized dose in the case of vinyl chloride demonstrate that the delivered dose concept can radically alter the shape of the dose-response curve in the observable response range. Simultaneously, it can eliminate the obvious bias in low-dose risk estimates that arises from use of an inappropriate dose measure, namely, the administered airborne vinyl chloride concentration. For formaldehyde, the data regarding covalent binding to respiratory mucosal DNA provide the best measure of target tissue exposure that is currently available. The results obtained with this dose measure demonstrate that incorporation of the delivered dose concept into any of the commonly used low-dose extrapolation procedures leads to a unilateral reduction in estimated cancer risk associated with the exposure of rats to low airborne formaldehyde concentrations. These findings strongly suggest that low-dose risk estimates obtained with the administered formaldehyde dose as the measure of exposure are biased conservatively, and that regulatory assessments of formaldehyde should consider the delivered formaldehyde dose to be the valid and relevant measure of exposure. Additional research is, of course, required to refine and elaborate the delivered dose concept, especially for human exposures. Nevertheless, this concept provides a convenient vehicle for incorporating mechanistic information directly into the quantitative risk assessment process in a meaningful and relevant manner. Clearly, risk estimates based on delivered dose reflect what is known of the underlying biological reality more faithfully than estimates that are based solely on unverified assumptions and the findings in chronic bioassays. Use of the delivered dose concept should be strongly encouraged, since it can help place quantitative risk assessment on a sound and scientifically defensible footing.