[Prediction of human lethal concentrations by cytotoxicity data from 50 MEIC chemicals].

Procedures for predicting human toxicity on the basis of cytotoxicity data for different chemicals are of current interest. The study was designed to clarify the possibility of predicting human toxicity by using the cytotoxicity values IC(50x) of the 50 MEIC chemicals listed in the Registry of Cytotoxicity (RC). All calculations with the data of cytotoxicity and in vivo toxicity were carried out uniformly by using standardised methods with the aim of comparing the results in the literature with each other. The following results were achieved: Firstly, the IC(50x) values in the RC are suited better for predicting human toxicity than IC(50) values determined in cell culture experiments with one cell type and one cytotoxic endpoint. This result is correct for the values of linear regression parameters as well as for the values of the prediction error (PE) defined by Ponsoda et al. (1997). Secondly, by using the geometrical mean of four lethal concentrations (LCx) - calculated from the tabulated lethal concentration (LC), lethal plasma concentration (LPC), clinical lethal concentration (CLC) and forensic lethal concentration (FLC) - the parameters for predicting the human toxicity were slightly improved. Moreover, these comparative examinations demonstrate that in all cases the cytotoxicity data are suited better for predicting acute animal toxicity than for predicting acute human toxicity. The results confirm, once more, the reliability and general validity of RC cytotoxicity data for examinations of different problems in cytotoxicology.