Prevalidation of the ex-vivo model PCLS for prediction of respiratory toxicity

- Prevalidation of PCLS was successfully conducted in three laboratories using 20 chemicals.
- Intra- and inter-laboratory variabilities showed best consistency and low variability for WST-1 and BCA assays.
- Two prediction models were developed for translation of results from in-vitro to in-vivo.
- The results can potentially be used to predict starting concentrations of acute inhalation toxicity studies.
- Three-dimensional tissue models of the lung provide best correlation to in-vivo data.

In acute inhalation toxicity studies, animals inhale substances at given concentrations. Without additional information, however, appropriate starting concentrations for in-vivo inhalation studies are difficult to estimate. The goal of this project was the prevalidation of precision-cut lung slices (PCLS) as an ex-vivo alternative to reduce the number of animals used in inhalation toxicity studies. According to internationally agreed principles for Prevalidation Studies, the project was conducted in three independent laboratories. The German BfR provided consultancy in validation principles and independent support with biostatistics.In all laboratories, rat PCLS were prepared and exposed to 5 concentrations of 20 industrial chemicals under submerged culture conditions for 1 h. After 23 h post-incubation, toxicity was assessed by measurement of released lactate dehydrogenase and mitochondrial activity. In addition, protein content and pro-inflammatory cytokine IL-1α were measured. For all endpoints IC50 values were calculated if feasible. For each endpoint test acceptance criteria were established.This report provides the final results for all 20 chemicals. More than 900 concentration-response curves were analyzed. Log10[IC50 (μM)], obtained for all assay endpoints, showed best intra- and inter-laboratory consistency for the data obtained by WST-1 and BCA assays. While WST-1 and LDH indicated toxic effects for the majority of substances, only some of the substances induced an increase in extracellular IL-1α. Two prediction models (two-group classification model, prediction of LC50 by IC50) were developed and showed promising results.