Distinguishing Pharmacokinetic and Pharmacodynamic Components When Animal-to-Human Extrapolation Uncertainty Is Characterized by an Empirical Distribution. L. R. Rhomberg, Gradient Corporation, MA
Since the advent of PBPK modeling in health risk assessment, the question of separately characterizing pharmacokinetic (PK) and pharmacodynamic (PD) components to dose extrapolation has been problematic. In recent "distributional" approaches to noncancer risk assessment, the traditional fixed Uncertainty Factor for animal-to-human extrapolation is replaced by a statistical distribution of possible equitoxic doses, often based on empirical results of comparative toxicity studies. In this talk, I explore issues regarding how to divide such distributions into PK and PD components for use when PBPK modeling makes available species-specific estimates of target-organ doses in experimental animals and humans, thereby reducing the PK component of uncertainty. A proposed method hinges on the additivity of variances of added independent random variables, which can be applied to parse the geometric standard deviation (GSD) of an empirically based lognormal distribution into two GSDs representing multiplied PK and PD components. Chemical- and species-specific PBPK models have uncertainty themselves, however, and this uncertainty needs to be "added back in." This construct provides an objective means to address the perennial question, How certain does new mechanistic information (in this case, PBPK modeling) need to be in order to be used in regulatory risk assessment? It also addresses the "overstuffing" problem, in which the uncertainties of separate components of an extrapolation seem to be larger in aggregate than the overall uncertainty. The implications for current methods using fixed uncertainty factors is discussed.
Go to . . .
2001 SRA Annual Meeting Table of Contents
2001 SRA Annual Meeting Author Index
Main Abstracts Menu Page
RiskWorld Home Page