A 3d reconstructed hemi-cornea model for predicting the eyeirritating potential of chemicals: Results of an inter-laboratory evaluation study

Currently, no single in vitro test system or respective test battery are able to completely replace the Draize Rabbit Eye Irritation Test. For example, by using validated assays, only severe irritants can be identified. Based on a previously developed 3-dimensional hemi-cornea consisting of both an epithelium and collagen-embedded stromal cells, we intended to establish a test method to reliably distinguish between eye-irritating and non-irritating substances. In an inter-laboratory trial test we aimed to: a) demonstrate the quality controlled production of the hemi-cornea model in two independent laboratories; and b) assess the predictive capacity and intra-laboratory and inter-laboratory reliability of the test system. Both laboratories produced their own hemi-cornea tissues according to the standard operation procedure (SOP). Only tissue models matching the predefined quality criteria based on negative (NC) and positive controls (PC), qualified for subsequent use in the eye irritation study. A set of 20 chemicals was tested under blinded conditions, to assess the performance and limitations of the test system comparing three different prediction models. The test chemicals comprised different chemical classes with different eye-irritating properties according to the Draize Rabbit Eye Irritation Test, covering all three GHS categories. Test chemicals, as well as negative and positive controls, were applied topically onto the surface of the hemi-corneal tissue and the viability of the entire tissue after different exposure intervals was monitored by using the MTT assay. The most suitable prediction model of the test system revealed in vitro-in vivo concordance of 80% and 70% in the participating laboratories, respectively, and an inter-laboratory concordance of 80%. Sensitivity of the test was 77% and specificity was between 57% and 86%. Although the results are promising, the test system has to be further optimised in order to enhance the predictive capacity. We conclude that additional physiologically-relevant endpoints in both the epithelium and stroma have to be developed for a satisfactory selectivity of the test system, and hence for the prediction of all GHS classes of eye irritation in one standalone test system.