Title:	Development of models for dose-response and dose-effect analyses in relation to design of experiment
DNr:	SNIC 2015/1-298
Project Type:	SNIC Medium Compute
Principal Investigator:	Gunnar Johanson <gunnar.johanson@ki.se>
Affiliation:	Karolinska Institutet
Duration:	2015-08-31 – 2016-09-01
Classification:	30102 30303
Homepage:	http://ki.se/en/imm/mathematical-dose-effect-analysis
Keywords:

Abstract

In the area of health risk assessment, toxicological effect data is used for hazard identification and to describe the relation between dose and effect/response. A central aspect of quantitative health risk assessment of chemicals concerns the development of so-called Point of Departures (PODs). A POD is an exposure level of a chemical that serves as a starting point in the establishment of guidance values or reference doses, like Acceptable Daily Intake (ADI). The so-called Benchmark dose (BMD) concept has been recommended by us and others as a statistical method to define the POD as an alternative. The BMD has recently been included in regulatory guidelines, e.g. the European chemical legislation (REACH), as a preferred method for dose-response/effect analysis. The BMD method involves fitting a dose-response model to all the available data. A benchmark dose is determined from the model as the dose that causes a predetermined level of effect/response. A lower confidence bound is also placed on the BMD, commonly referred to as the BMDL, to account for uncertainties in the data material that is used. BMD softwares are available and BMD approaches are described and recommended in official guidelines. However, most generated toxicity data are still based on study designs suitable for traditional evaluation methods (i.e. four equally sized groups of animals). More emphasis is needed on the selection of study designs appropriate for dose-response modeling and cost-benefit analysis (including economical and ethical aspects as well as regulatory usefulness) should be included during the selection process. Aim To further develop the concept of BMD modeling to improve the methodology for design of toxicological experiments in relation to the use of experimental animals. Method We will create (Monte Carlo simulation) and evaluate different simulated scenarios with variable number of dose groups, intervals, number of animals, symmetric and asymmetric distribution of animals. For this task we will also identify relevant set of toxicological data. The present project aims to investigate how much the number of animals can be reduced, and in what way they can be reduced, without resulting in a high uncertainty in quantities of interest (e.g. the BMD). In a later stage we will develop a measure of animal distress and use that measure as a base for cost-benefit optimization where the cost is measured in animal distress.