Initial situation/scientific issue

Since the development of the first quantitative structure-activity relationship (QSAR) model in the early 1960s, QSAR and later machine learning based approaches have been widely used in both drug discovery and toxicity prediction. Specifically relevant for InSilify DrugTox is the fact that computational toxicity modelling bears a high potential to reduce the animal demand in drug development. However, comparative retrospective analyses of how available QSAR models would have performed in predicting toxicity are rare. Consequently, new comparative and retrospective studies are needed with a relevant and preferably pristine data pool from drug development and authorisation sources to test if the available models are accurate and fit-for-use.

Project description/methodology

InSilify DrugTox will perform a large-scale retrospective computational toxicology analysis using a large pool of toxicity data from regulatory sources (EMA-SA final letters and EPARs) that have never been used for this purpose. The overarching research question behind these investigations is to determine whether available QSAR models can reliably predict toxicity in drug development. These investigations will guide and support the in silico transition in drug development, which will translate into a reduced demand of pre- and non-clinical animal studies and a faster and cheaper availability of drugs for which there is still an unmet medical need.

The goals of InSilify DrugTox to curate an extensive data pool out of pristine drug development and authorisation sources and to perform thorough retrospective toxicity modelling with these data is a novel and highly relevant contribution to the field of computational toxicology. In addition, while many comparative QSAR studies have focused exclusively on Ames mutagenicity, the proposed project will analyse a much broader range of safety relevant endpoints. Finally, an important advantage of the proposed project is the collaboration between renowned scientists and drug regulatory experts, which will allow us to combine state-of-the-art science with an enhanced potential to translate relevant results into drug regulatory frameworks.

The proposed project will be a scientific collaboration of three Austrian institutions (AGES Medical Market Surveillance, the Pharmacoinformatics Research Group of the University of Vienna and the Regulatory Toxicology Group of the Medical University of Innsbruck) with broad scientific and regulatory expertise in drug regulatory toxicology, toxicological in silico modelling, NAMs and 3Rs.

Outcomes