Soil degradation toxicity potential (DT50) of VPs followed by biodegradability and leaching: Exploration of possible aquatic and terrestrial component toxicity through QSAR, q-RASAR, and comprehensive screening
摘要
Environmental monitoring has revealed the presence of Veterinary Pharmaceuticals (VPs) in surface waters, groundwater, manure, soils and treated drinking water. The soil degradation half-life (DT50) is a key indication for assessing chemical risks. To address this issue, we developed robust in silico soil degradation prediction models that included QSAR and q-RASAR. A curated dataset of 94 VPs with experimentally determined “typical” DT50 values was used to calculate 0D-2D physicochemical descriptors (PaDEL), as well as similarity-based descriptors. Model development employed GA-MLR in QSARINS, with rigorous descriptor pretreatment, different dataset splitting and OECD-compliant validation. All the developed models were statistically robust and predictive. Mechanistic interpretation identified persistence-enhancing features such as high atomic mass autocorrelation (AATSC7m), bulky aromatic/heteroaromatic moieties (GATS7v), and halogenation, while polarizability (GATS3p), higher connectivity (VC-6) and conformational flexibility (RotBFrac) correlated with faster degradation. The models were applied to 283 VPs without experimental DT50, followed by consensus AD screening and multi-criteria decision-making (MCDM) analysis. Further, 223 VPs were retained within the AD, ensuring only reliable predictions. Predicted DT50 values were combined with soil adsorption coefficients (Koc, via KOCWIN) and biodegradability (TOPKAT, BIOWIN) to assess leaching potential using the Groundwater Ubiquity Score (GUS). Several fluoroquinolones, macrolides, and halogenated VPs were classified as persistent, non-biodegradable and high-leaching, indicating elevated aquatic risk; others showed low leachability but high persistence, suggesting terrestrial accumulation hazards. Overall, this integrated QSAR and q-RASAR framework enables reliable prioritization of VPs based on combined persistence and leaching risks.