P-22 A Comparative Cytotoxicity Analysis of the Results from Tests of the First 30 MEIC Reference Chemicals in 68 Different In Vitro Toxicity Systems

MEIC (Multicenter Evaluation of In vitro Cytotoxicity) is an ongoing international project to evaluate the relevance for human toxicity of various in vitro toxicity and toxicity and toxicokinetics tests. It started in 1989 and will be finalized in the end of 1995. Interested laboratories are invited to test a set of 50 reference chemicals in their respective in vitro systems and submit results to us for evaluation. Presently, 31 MEIC laboratories have tested the first 30 reference checmials in 68 different in vitro toxicity tests, including 15 with human cell lines, 3 with human primary cultures of hepatocytes and keratinocytes, 18 with animal cell lines, 21 with animal primary cultures and 11 ecotoxicological systems (bacteria, pollen tubes, onion roots, and small aquatic multicellular organisms). This communication presents and analyses the in vitro data, as a first necessary step of an ongoing comparison of these data to human and animal acute toxicity. The present results have been compared with use of principal component analysis (PCA), variance analysis (ANOVA) and linear regression. Firstly, the similarity of all methods was determined. PCA as well as pairwise variance analysis (ANOVA and Tukey's method) indicated an 80% general concordance of all methods taken together. Secondly, the influence of some key methodological factors on the general variability of the data was studied by linear regression as well as by ANOVA plus an a priori analysis of linear con-trasts of means：1) Exposure time was found to increase toxicity for many checmials. 2) Human cytotoxicity was well predicated by test with animal cells (except for digoxin and malathion), and relatively well by ecotoxicological tests. However, most ecotoxicological tests predicted human cytotoxicity well, provided that a few chemicals were excluded, indicating a close connection between ecotoxicity and human toxicity of chemicals. 3) One organotypic toxicity creterion used, i.e. inhibition of contractility of muscle cells, gave different results compared with the viability/growth criteria. 4) All twelve comparisons, that could be made in this study, between identical test systems using different cell types (including highly differentiated cells) showed a similar toxicity regardless of cell type. 5) Nine of the ten comparisons of test systems with identical cell types and exposure times revealed a similar toxicity regardless of the endpoint used. Here, factors 1-3, including other factors not ana