Comparative in Silico Toxicological Assessment of 16 Selective COX-2 Inhibitors (Coxibs) Using Multi-Platform Predictive Models

Selective cyclooxygenase-2 (COX-2) inhibitors (coxibs) offer reduced gastrointestinal toxicity compared to traditional non-steroidal anti-inflammatory drugs (NSAIDs), yet their safety remains controversial as evidenced by several market withdrawals due to severe cardiovascular and hepatic risks. This study systematically evaluated 16 human and veterinary coxib compounds for mutagenicity, carcinogenicity, micronucleus activity, and hepatotoxicity through a multi-platform in silico approach using VEGA-QSAR, ToxTree, and T.E.S.T. software. The results indicated that while widely used agents like Celecoxib and Etoricoxib exhibit safer profiles, compounds such as Lumiracoxib and NS-398 possess significant carcinogenic and mutagenic potential. Notably, high-reliability hepatotoxicity alerts were identified for several veterinary-grade coxibs, raising significant public health concerns regarding drug residues in the equine-derived food chain and potential human exposure. Discrepancies observed between different predictive models highlight the limitations of individual algorithms and underscore the necessity of a consensus approach to improve the robustness of risk evaluations. This study provides an essential foundational framework for the safety assessment of coxib derivatives and emphasizes the critical importance of validating in silico alerts with targeted in vivo research, particularly in the context of food safety and drug residue management.