Identification of a Critical Dose Level for Risk Assessment: Developments in Benchmark Dose Analysis of Continuous Endpoints

The benchmark dose (BMD) method has been recommended to replace the no-observed-adverse-effect-level (NOAEL) approach in health risk assessment of chemical substances. In the present article, developments in BMD analysis from continuous experi-mental data are proposed. The suggested approach defines the BMD as the dose at which the slope of the S-shaped dose–response relationship changes the most in the low-dose region. This dose resides in a region where the sensitivity to chemical exposure may start to change noticeably. It is shown that the response (defined as a percent change relative to the magnitude, or size, of response) corresponding to the dose where the slope changes the most depends on the geometrical shape of the dose–response curve; the response becomes lower as the curve becomes more asymmetrical and threshold-like in the low-dose region. Given a symmetrical case, described by the Hill function, the response associated with the critical dose level becomes 21 % (defined as a percent change relative to the magnitude, or size, of response). According to a limiting case of asymmetry and threshold-like characteristics, reflected by a Gompertz curve, the response corresponding to the dose of interest becomes as low as 7.3 % (defined as a percent change relative to the magnitude, or size, of response). Use of a response in the range of 5–10 % when estimating the BMD conservatively accounts for uncertainties associated with the proposed strategy, and may be appropriate in a risk assessment point of view. The present investigation also indicated that a BMD defined according to the suggested procedure may be estimated more precisely relative to BMDs defined under other approaches for continuous data. Key Words: benchmark dose; BMD; risk assessment; dose–response; Gompertz model; Hill function; Richards function