Inter- and intra-laboratory study to determine the reproducibility of toxicogenomics datasets.

The application of toxicogenomics as a predictive tool for chemical risk assessment has been under evaluation by the toxicology community for more than a decade. However, it predominately remains a tool for investigative research rather than for regulatory risk assessment. In this study, we assessed whether the current generation of microarray technology in combination with an in vitro experimental design was capable of generating robust, reproducible data of sufficient quality to show promise as a tool for regulatory risk assessment. To this end, we designed a prospective collaborative study to determine the level of inter- and intra-laboratory reproducibility between three independent laboratories. All test centres (TCs) adopted the same protocols for all aspects of the toxicogenomic experiment including cell culture, chemical exposure, RNA extraction, microarray data generation and analysis. As a case study, the genotoxic carcinogen benzo[a]pyrene (B[a]P) and the human hepatoma cell line HepG2 were used to generate three comparable toxicogenomic data sets. High levels of technical reproducibility were demonstrated using a widely employed gene expression microarray platform. While differences at the global transcriptome level were observed between the TCs, a common subset of B[a]P responsive genes (n=400 gene probes) was identified at all TCs which included many genes previously reported in the literature as B[a]P responsive. These data show promise that the current generation of microarray technology, in combination with a standard in vitro experimental design, can produce robust data that can be generated reproducibly in independent laboratories. Future work will need to determine whether such reproducible in vitro model(s) can be predictive for a range of toxic chemicals with different mechanisms of action and thus be considered as part of future testing regimes for regulatory risk assessment.