MEIC Evaluation of acute systemic toxicity. Part VII. Prediction of human toxicity by results from testing of the first 30 reference chemicals with 27 further In Vitro assays

The Multicenter Evaluation of In Vitro Cytotoxicity (MEIC) programme was set up to evaluate the relevance for human acute toxicity of in vitro cytotoxicity tests. By the end of the programme in 1996, 39 laboratories had tested the first 30 reference chemicals in 82 in vitro toxicity assays, and the last 20 chemicals had been tested in 67 assays. All in vitro results and the human database have been presented in five previous papers (Parts I-V). Part VI evaluated the in vitro results from the 61 assays used to test all 50 chemicals by comparisons of in vitro IC50 values (concentrations causing 50% inhibition) with LC50 values (blood concentration causing approximately 50% lethality) from human single-dose acute poisonings by the chemicals. These comparisons demonstrated a good prediction of human peak LC50 values (peaks from the LC50 curves over time) by most of the 61 assays, notably the human cell line assays (R 2 = 0.74). The present paper is supplementary to Part VI, and presents a similar evaluation of the 27 new assays used to test the first 30 reference chemicals, as well as a less-detailed analysis of the six new assays used to test the last 20 chemicals. Comparisons between the IC50 values from the 27 assays used to test the first 30 chemicals and peak LC50 values demonstrated that human cell line tests gave the best mean results (R 2 = 0.76), followed by human primary culture assays (R 2 = 0.75), animal cell line assays (R 2 = 0.68), animal primary culture assays (R 2 = 0.65), and bacterial assays (R 2 = 0.43), and confirm the findings of Part VI. Some assays were evaluated separately. Firstly, mean IC50 values from three human keratinocyte assays were compared with human peak LC50 values, resulting in a very good correlation (R 2 = 0.84). When the IC50 values for 19 chemicals which freely pass the blood-brain barrier were divided by a factor of ten (to compensate for a hypothetical extra sensitivity of the central nervous system to cytotoxicity), the correlation improved (R 2 = 0.87). However, the pattern of outlier chemicals was found to be different to that of typical outliers for human cell line assays, which makes it difficult to use keratinocytes for acute systemic toxicity testing. Secondly, mean IC50 values from two human hepatocyte assays were compared with human peak LC50 values, resulting in a good direct correlation (R 2 = 0. 78) and an even better blood-brain barrier-compensated correlation (R 2 = 0.81). The hepatocyte assays also had typical outlier chemicals without relevance for acute systemic toxicity. To find out whether typical hepatocyte responses could be used to predict lethal liver injury, a differential cytotoxicity study was conducted. The relatively high sensitivity of hepatocytes compared with human cell lines, as judged by the differences between the mean IC50 values for the two groups of assays, was compared with reported lethal liver injuries and the basal cytotoxicities of the human lethal blood concentrations of the 30 compounds. For three chemicals, i.e. methanol, phenol and arsenic trioxide, high sensitivity of the hepatocytes was indeed correlated with some liver injury. However, there was a much better correlation between a high basal cytotoxicity of the human blood concentration and liver injury.