Resolving data gaps in the acute toxicity profile of phosgene oxime (CX): a comprehensive in silico evaluation

Abstract

Phosgene oxime (CX) is a highly reactive chemical warfare agent classified as a nettle agent due to its rapid induction of corrosive skin lesions, severe pain, and tissue necrosis upon exposure. Despite its recognised extreme toxicity, substantial gaps remain in our understanding of CX's acute toxicological profile, primarily owing to limited experimental data stemming from significant ethical, safety, and regulatory constraints. Consequently, accurate risk assessment for CX exposure has been challenging, necessitating the application of alternative predictive methodologies. In response, this study aimed to comprehensively evaluate the acute toxicity of CX via state-of-the-art in silico methods that integrate multiple computational toxicology tools, including STopTox, ADMETlab, admetSAR, TEST, ProTox-III, VEGA, OPERA, the QSAR Toolbox, and Percepta ACD/Labs. Predictive modelling encompasses acute oral, dermal, and inhalation toxicity endpoints, providing theoretical LD₅₀/LC₅₀ values for rats, which are extrapolated to human-equivalent doses via established allometric scaling techniques. The results consistently demonstrated a high acute toxicity profile of CX across all exposure routes, particularly via inhalation and dermal contact, emphasizing the substantial health risks associated with potential CX incidents. However, notable variability among computational predictions has led to limitations related to the applicability domain and dataset constraints, highlighting areas requiring further methodological refinement. This research represents the extensive application of validated in silico approaches to elucidate the acute toxicity parameters of CX. These findings underscore the utility of computational toxicology methodologies as ethically sound, hypothesis-generating alternatives to experimental testing, enhancing chemical threat preparedness and regulatory toxicological assessment.