You are here:
STATISTICAL APPROACH TO BRAIN MORPHOMETRY DATA REQUIRED IN DEVELOPMENTAL NEUROTOXICITY (DNT) TESTING GUIDELINES: PROFILE ANALYSIS.
Citation:
Geller, A M. AND A. M. Jarabek. STATISTICAL APPROACH TO BRAIN MORPHOMETRY DATA REQUIRED IN DEVELOPMENTAL NEUROTOXICITY (DNT) TESTING GUIDELINES: PROFILE ANALYSIS. Presented at Society of Toxicology, Nashville, TN, March 17-21, 2002.
Description:
Brain morphometry measurements are required in test guidelines proposed by the USEPA to screen chemicals for developmental neurotoxicity. Because the DNT is a screening battery, the analysis of this data should be sensitive to dose-related changes in the pattern of brain growth without requiring prior expectations about specific regions or the magnitude and direction of these changes. In addition, analysis should take into account the lack of independence between multiple brain region measurements taken within the same animal, since the sizes of different regions are likely related. Analysis with t-tests or univariate analyses of variance is inappropriate because it assumes independence and runs the risk of Type 1 error due to repeated sampling using the same error term. A multivariate analysis of variance, profile analysis, was adopted to examine brain morphometry data submitted for the risk assessment of perchlorate (ClO4-), an endocrine-disrupting compound shown to reduce thyroid hormone levels and alter brain development. Profile analysis reduces the number of main effects tested by making between-groups (dose group) comparisons using vectors that comprise all of the within-subject morphometric data from each animal, then comparing these vectors. It also allows gender to be nested within litter, which is necessary to address DNT study designs. The primary test of profile analysis examines whether the pattern made by the various measurements is similar for all treatment groups. This is a test of the interaction between the treatment factor and the within-subjects scores. If the profiles are not parallel, an interaction exists, and further analyses of the interaction factors may be made to elucidate the effects. Results of this analysis showed that ClO4- affected the pattern of brain growth in developing pups at all doses tested. We conclude that this analysis and approach are readily applicable to future DNT data. This is an abstract of a proposed presentation and does not necessarily reflect EPA policy.