Publications : 2019

Swartz C, Howard AS, Choksi N, Rauer A, Allen DG, Recio L, Karmaus AL. An integrated approach for animal-free genotoxicity testing: In vitro and in silico evaluation and mode-of-action classification. Poster presented at Society of Toxicology Annual Meeting, Baltimore, MD, March 2019.

Abstract

New alternative approaches (NAMs) for the rapid and cost-effective screening of genotoxicity are promising potential replacements for traditional tests. ILS has developed a workflow to evaluate chemicals for possible genotoxicity potential as determined by in silico predictive modeling and human-relevant in vitro testing. High-throughput mammalian cell-based genotoxicity screening assays in human relevant TK6 cells were performed using a validated MultiFlow™ DNA Damage – p53, γH2AX, Phospho-Histone H3 kit which informs on genotoxicity and potential mode-of-action. Chemicals were also evaluated using 14 CASE Ultra software (MultiCase Inc.) for quantitative structure activity relationship (QSAR) genotoxicity predictions, including assessment of an overall genotoxicity prediction as well as consensus predictions from 4 models for carcinogenicity and 10 models for mutagenicity. For this proof-of-concept study, food-use chemicals were evaluated, including direct food additives and flavoring compounds. Results highlight that the MultiCase predictive modeling dataset rarely contained similar chemistries as seen in many of the food-use chemicals, namely flavorings. Since the predictive modeling approach relies on QSAR, which is based on chemical structure analog similarity, this is a serious limitation. Conversely, the in vitro approaches are more amenable to screening compounds with a variety of structural features. Overall, food-use compounds were correctly identified as negative for genotoxicity based on in silico and in vitro data. While this reflects high accuracy, it also limits the capacity to evaluate performance metrics due to a heavy bias toward the resulting dataset’s abundance of true negative findings with very few positive chemicals. This study demonstrates effective NAM use for evaluating genotoxicity, emphasizes the need for study design considerations when considering predictive modeling software, and exemplifies the impact of domain of applicability limitations when considering alternative approaches. Ultimately, this integrated approach, leveraging both in vitro assays and in silico predictions, can be applied to any class of chemicals to effectively screen for genotoxicity potential.