Qualitative and Quantitative In Silico Assessment of Acute Toxicity of PFOA and PFOS PFOA ve PFOS'un Akut Toksisitesinin İn Siliko Nitel ve Nicel Değerlendirilmesi

Creative Commons License

Karakuş F.

Fabad Journal of Pharmaceutical Sciences, cilt.51, sa.2, ss.435-452, 2026 (Scopus)

Özet

Perfluorooctanoic acid (PFOA) and perfluorooctane sulfonic acid (PFOS) have been categorized as persistent organic pollutants under the Stockholm Convention. Moreover, the IARC classifies them as classified carcinogens (Group 1) and possibly carcinogenic (Group 2B) to humans, respectively. Toxicological and epidemiological evidence links PFOA and PFOS to multiple adverse health outcomes, including renal and testicular cancer. However, data on their acute toxicity remain limited. To address this gap, this study performed a comprehensive in silico assessment of PFOA and PFOS using both qualitative and quantitative computational models. Acute toxicity was evaluated for oral, dermal, inhalation routes, eye, and skin exposure. Qualitative predictions from Toxtree, STopTox, admetSAR, and ADMETlab consistently indicated toxicity, particularly via oral and inhalation exposure, as well as through eye and skin contact. The carboxylic acid moiety was highlighted as a key toxicophore in PFOA, while the whole structure was identified as a toxicophore in PFOS, possibly contributing significantly to their toxicity. Quantitative analyses, employing ProTox-3.0, Deep-PK, ToxACoL, TEST, VEGA, and QSAR Toolbox, revealed notable variation in LD50 estimates, with LC50 data being unavailable. Overall, LD50 predictions supported high acute toxicity via oral exposure. These findings underscore the acute toxicity potential of PFOA and PFOS and additionally demonstrate the value of in silico toxicology for screening data-poor chemicals and supporting regulatory decision-making. Although variability among prediction algorithms and training datasets presents limitations, in silico approaches offer a rapid and cost-effective hazard assessment. Future studies should focus on improving predictive performance through expanded datasets and integration with experimental validation.