© 1990 Oxford University Press
other |
Interpreting Rodent Clinical Laboratory Data in Safety Assessment Studies: Biological and Analytical Components of Variation1
*Quality Management and Technology Center, E 1. du Pont de Nemours and Company, Inc. Newark, Delaware 19714 Haskell Laboratory for Toxicology and Industrial Medicine, E 1. du Pont de Nemours and Company, Inc. Newark, Delaware 19714
Received April 5, 1990; accepted July 12, 1990
Interpreting Rodent Clinical Laboratory Data in Safety Assessment Studies: Biological and Analytical Components of Variation. CARAKOSTAS, M. C., AND BANERJEE, A. K. (1990). Fundam. Appl. Toxicol 15, 744–753. Evaluation of the biological and toxicological significance of clinical laboratory results obtained in safety assessment studies requires an understanding of factors unrelated to the treatment that may affect test results. Since the magnitude of the components of variation is usually unknown, the toxicologic significance of small, but statistically significant, differences between control and treatment group results can be difficult to assess. Over a 12-week period, components of variation were determined for a wide range of hemato-logic and clinical biochemical assays in clinically normal Sprague-Dawley-derived rats. Estimates of variance components and ratios were obtained for each test. While the intra-animal/inter-animal ratios (r ratios) revealed some important tests with a high or low degree of individuality, many hematologic and clinical biochemical tests had r ratios within the 95% confidence interval for equal variances. Analytical variance ratios revealed tests that are sensitive to the effects of experimental error and experimental design. Due to the diversity of the variance component patterns among clinical laboratory tests, complex experimental designs may be required to reduce the effects of analytical and biological variation on the statistical analysis of clinical laboratory data. The results of this study suggest that statistically significant clinical laboratory findings that are not biologically or toxicologically important will be present in many rodent safety assessment studies with a standard design. Overreliance on the result of standard prepackaged statistical analyses for determining the presence of toxicologically significant findings can lead to misinterpretation of clinical laboratory data. Sound medical judgment must be applied to clinical laboratory findings using appropriate statistical analyses as a tool for pattern recognition.