INTRODUCTION

There is a great deal of interest in developing more biologically insightful approaches to dose-response analysis, and also in achieving harmonization of the rationale and methods for analysis of cancer and noncancer endpoints (Beck et al., 1993; Conolly, 1995; Barton, Andersen, and Clewell, 1998; Conolly, Beck, and Goodman, 1999). These goals require one to think about how events at underlying levels of molecular and physiological activity are modulated by the action of a toxic agent, and how such low-level effects then propagate to produce the overt toxic effects seen in the living organism. In essence, we need to delve into how and why it is that varying levels of a toxicant lead to varying levels of response, dissecting the chains of cause-and-effect relationships involved. This paper explores some of these issues.

BASIC DOSE-RESPONSE APPROACHES

There are two broad approaches to analysis of dose-response relationships, and they correspond to two alternative rationales for why responses are observed to vary as a function of dose (Rees and Hattis, 1994). The first is the idea that a dose-response curve describes a tolerance distribution (Figure 3.1). Under this view, the population at risk consists of individuals who vary in the amount of an agent they can experience without ill effect (Krewski et al., 1999). In essence, each individual is assumed to have an exposure threshold, but the value of this threshold differs from one individual to the next (Eaton and Klaassen, 1996).