ABSTRACT
The assessment of the carcinogenic potential of chemicals with alternative, human-based in vitro systems has become a major goal of toxicogenomics. The central read-out of these assays is the transcriptome, and while many studies exist that explored the gene expression responses of such systems, reports on robustness and reproducibility, when testing them independently in different laboratories, are still uncommon. Furthermore, there is limited knowledge about variability induced by the data analysis protocols. We have conducted an inter-laboratory study for testing chemical carcinogenicity evaluating two human in vitro assays: hepatoma-derived cells and hTERT-immortalized renal proximal tubule epithelial cells, representing liver and kidney as major target organs. Cellular systems were initially challenged with thirty compounds, genome-wide gene expression was measured with microarrays, and hazard classifiers were built from this training set. Subsequently, each system was independently established in three different laboratories, and gene expression measurements were conducted using anonymized compounds. Data analysis was performed independently by two separate groups applying different protocols for the assessment of inter-laboratory reproducibility and for the prediction of carcinogenic hazard. As a result, both workflows came to very similar conclusions with respect to (1) identification of experimental outliers, (2) overall assessment of robustness and inter-laboratory reproducibility and (3) re-classification of the unknown compounds to the respective toxicity classes. In summary, the developed bioinformatics workflows deliver accurate measures for inter-laboratory comparison studies, and the study can be used as guidance for validation of future carcinogenicity assays in order to implement testing of human in vitro alternatives to animal testing.
Subject(s)
Carcinogens/toxicity , Computational Biology , Gene Expression Profiling , Kidney Tubules, Proximal/drug effects , Laboratory Proficiency Testing , Liver/drug effects , Toxicogenetics/methods , Transcriptome/drug effects , Carcinogens/classification , Cell Line, Tumor , Dose-Response Relationship, Drug , Gene Expression Regulation/drug effects , Genome-Wide Association Study , Humans , Kidney Tubules, Proximal/metabolism , Liver/metabolism , Observer Variation , Oligonucleotide Array Sequence Analysis , Reproducibility of Results , Risk Assessment , Time Factors , WorkflowABSTRACT
Statins are widely used lipid-lowering drugs that are effective in reducing cardiovascular disease risk. Although they are generally well tolerated, they can cause muscle toxicity, which can lead to severe rhabdomyolysis. Research in this area has been hampered to some extent by the lack of standardized nomenclature and phenotypic definitions. We have used numerical and descriptive classifications and developed an algorithm to define statin-related myotoxicity phenotypes, including myalgia, myopathy, rhabdomyolysis, and necrotizing autoimmune myopathy.