Interpretation of cell toxicity data for the estimation of potential irritation.

Three cytotoxicity assays were evaluated using 57 chemicals of various classes (inorganic and organic metal salts, solvents, detergents, reagents, drugs) which have widely different mechanisms of cytotoxicity. Baby hamster kidney fibroblasts (BHK-21/C13) and early (Keller) and late (MRC-5) passage human fibroblasts were used to measure cell detachment, cloning efficiency, and growth inhibition under subconfluent culture conditions. For the majority of chemicals, for which comparisons were made, the ranking order was roughly the same in all three tests and with all three cell types. However, for some chemicals specific growth effects could either be detected or excluded because the relationship between the data from the detachment assay and that from one of the growth assays was characteristically altered. The ranking order resulting from our in vitro data correlated better with threshold limit values for human workroom air (TLV/TWA) than with LD50 values (rat, oral). Correlations with data from Draize skin and eye irritation tests were not determined since the available in vivo values were derived using various different scoring systems. However, when our in vitro data were used to divide the chemicals into three crude classes, (i) non-irritant, (ii) mild to moderate irritant, or (iii) strong irritant or corrosive, and the results were compared with the known irritation potential for skin and mucous membranes derived from human exposure data, the in vitro data were more than 80% predictive of the in vivo classifications.