In silico studies in toxicology: Experiences, challenges, and perspectives

Abstract

Animal testing remains the standard for assessing the safety of drugs and chemicals; however, computer-aided (in silico) methods are increasingly being developed as efficient alternatives or complements to in vivo testing, supported by regulatory frameworks such as REACH. Their advantages include low cost, the ability to analyze hypothetical compounds, and reliance on human-relevant data, thereby reducing extrapolation of results between different biological species. Since the mid-1990s, these technologies have become essential in drug development, particularly for assessing ADMET properties and predicting toxic effects, which account for a significant proportion of late-stage failures in pharmaceutical development. Bioinformatics and in silico approaches in toxicology enable faster, more ethical, and more cost-effective chemical safety assessment, in line with the 3R principles. Previous experience includes QSAR modeling, identification of toxicophores, toxicogenomic analyses, and data mining of public databases. Notably, Serbian–Chinese collaborative research on bioactive compounds demonstrated that in silico toxicogenomic analysis of sulforaphane in colorectal cancer identified gene signatures and molecular patterns associated with therapeutic effects, as well as potential adverse outcomes depending on the patient’s genetic profile. The application of the Adverse Outcome Pathways (AOP) framework and predictive tools has further improved understanding of toxicity mechanisms. Despite these advantages, challenges remain, including limited interpretability of complex models, dependence on data quality, insufficiently precise dose–response modeling, and the need for experimental validation. Future directions in toxicology rely on integrated approaches to testing and assessment (IATA), combining in silico, in vitro, and in vivo methods to develop more reliable and personalized risk assessment strategies.