Zhu Yiliang, Wessel Michael R, Liu Tiebin, Moser Virginia C
Department of Epidemiology and Biostatistics, College of Public Health, University of South Florida, Tampa, FL 33612, USA.
Regul Toxicol Pharmacol. 2005 Apr;41(3):240-55. doi: 10.1016/j.yrtph.2004.12.005.
Neurotoxic effects are a non-cancer endpoint for health risk, and neurobehavioral screening tests can serve as a first tier investigation of neurotoxicity [US EPA, Federal Register 63 (1998) 26926]. Analysis of neurobehavioral screening data such as those of the functional observational battery (FOB) traditionally relies on analysis of variance (ANOVA). ANOVA is designed to detect whether there are dose-effects, but does not model the underlying dose-response relationship and subsequent risk assessment fails to utilize the shape of the underlying dose-response. In contrast, dose-response modeling interpolates toxic effects between experimental points, and permits prediction of toxic effects within the experimental range. Additionally it is also a prerequisite for estimating a benchmark dose. This paper discusses dose-time-response modeling of longitudinal neurotoxicity data and illustrates the methods using three continuous FOB outcomes from an EPA study involving acute exposure to triethyltin (TET). Several mathematical functions are presented as candidate dose-time-response models. The use of random effects is discussed to characterize inter-subject variation. The results indicate that it is feasible to use simple mathematical functions to model empirical dose-time-response observed in existing longitudinal neurotoxicological data. Further research is needed on the types of design and data required to reliably approximate the true underlying dose-time-response.
神经毒性效应是健康风险的非癌症终点，神经行为筛查测试可作为神经毒性的一级调查手段[美国环境保护局，《联邦公报》63 (1998) 26926]。传统上，对诸如功能性观察组合（FOB）等神经行为筛查数据的分析依赖于方差分析（ANOVA）。方差分析旨在检测是否存在剂量效应，但并未对潜在的剂量反应关系进行建模，因此后续的风险评估无法利用潜在剂量反应的形态。相比之下，剂量反应建模可在实验点之间内插毒性效应，并允许在实验范围内预测毒性效应。此外，它也是估计基准剂量的先决条件。本文讨论了纵向神经毒性数据的剂量 - 时间反应建模，并使用美国环境保护局一项涉及急性暴露于三乙基锡（TET）的研究中的三个连续FOB结果来说明这些方法。提出了几种数学函数作为候选剂量 - 时间反应模型。讨论了使用随机效应来表征个体间差异。结果表明，使用简单的数学函数对现有纵向神经毒理学数据中观察到的经验性剂量 - 时间反应进行建模是可行的。对于可靠地近似真实潜在剂量 - 时间反应所需的设计和数据类型，还需要进一步研究。
