Quantitative assessment of primary skin irritants in vitro in a cytotoxicity model: comparison with in vivo human irritation tests

Background While great efforts have been made in recent years to develop in vitro methods for assessing skin irritation potential, there are relativel y few data that correlate in vitro data with in vivo data. Objectives To expand our previously reported investigations on in vitro vs. in vivo correlation of a series of homologous N-alkyl sulphates of differe nt alkyl chain length to include primary skin irritants of different chemic al classes. Methods Anionic surfactants (three different sodium alkyl sulphonates and s odium lauryl sulphate), cationic surfactants (three alkyl trimethyl ammoniu m bromides), non-ionic surfactants (polyoxyethylene-20-cetyl ether and Twee n 20), benzoic, acid, dimethyl sulphoxide and phenol were chosen as model i rritants. A spontaneously immortalized human keratinocyte line, HaCaT, was used as an in vitro model to predict the cutaneous irritation. The end-poin t used to assess toxicity was uptake of the vital dye neutral red (NR) 24 h after dosing. The cytotoxicity data from these assays were compared with t he irritant responses (as evaluated by measurement of erythema and transepi dermal water loss) obtained after 24-h application of the same compounds (1 00 muL of 20 mmol L-1 aqueous solution) to the volar forearm of human volun teers. Results All tested irritants had cytotoxic effects as demonstrated by a dec reased NR uptake, which showed a clear dose-response relationship. Concentr ations resulting in 50% inhibition of NR uptake (IC50) ranged from 8 Lmol L -1 (hexadecyl trimethyl ammonium bromide) to 328 mmol L-1 (dimethyl sulphox ide). We found a good overall correlation between in vitro cytotoxicity (NR uptake IC50 values) and in vivo irritation potential in humans. Only the h igh molecular weight compounds Tween 20 and polyoxyethylene-20-cethyl ether were problematic, as their irritation potential was overestimated by the i n vitro assay. This non-conformity of these high molecular weight (>1000) c ompounds was expected, and can be largely attributed to the epidermal perme ability barrier. The epidermal barrier, which greatly limits the percutaneo us penetration of xenobiotics in vivo, does not exist in cell culture model s. Conclusions The in vitro cytotoxicity model is a useful screening tool, but data should be interpreted critically and require confirmation by appropri ate in vivo studies.