Development and Peer Review of Time-to-Effect Models for the Analysis of Neurotoxicity and Other Time Dependent Data

Objective/Intended Use

The objective of this project is to provide Agency risk assessors with software tools that will facilitate the analysis of time dependent dose-response data. Data generated from neurobehavioral studies was used as example data during development of this software. The resultant statistical tools and models will be used by risk assessors both inside and outside of EPA. Anticipated products include new tools and models for incorporation into the Agency's benchmark dose software (BMDS) system, guidelines for the use of these models, and updates to BMDS web pages and user manuals.

Abstract

Neurobehavioral studies pose unique challenges for dose-response modeling, including small sample size and relatively large intra-subject variation, repeated measurements over time, multiple endpoints with both continuous and ordinal scales, and time dependence of risk characterization. Other studies in which the time required to induce an effect is a critical factor (e.g., high morbidity cancer studies) share a number of these challenges. The Agency has developed draft (prototype) statistical models (polynomial and PBPK-based) to derive benchmark doses for these types of data and is in the process of writing draft guidelines for the use of these models for the evaluation of neurotoxicity test battery results. The purpose of this task is to provide support for the completion of the two prototype models, including the development of user-friendly (Windows based) model interfaces and incorporation into the Agencys BMDS software, and for the peer review of both the models and the aforementioned draft guidelines.

Project Status

Two neurotoxicity "time-to-effect" models have been developed in S-Plus and converted to R (both S-Plus and R are programs that facilitate the running of statistical models) and are undergoing testing in the Agency. Draft guidelines for applying these models to assess dose-response data from neurotoxicity test batteries are complete. The models and the supporting application guidelines were subjected to an external scientific peer review on December 15, 2003, conducted by Oak Ridge Institute for Science and Education (ORISE). After revision based on comments received at the peer review, the models and guidance will be made available to the public through NCEAs BMDS website.

Project Start Date

04/01/2001