A hybrid In-Vitro/In-Silico Framework approach for PFAS Toxicity and Synergy Assessment

Per- and Polyfluoroalkyl Substances (PFAS), a vast family of ~15,000 synthetic fluorinated chemicals, pose significant toxicological challenges due to their environmental ubiquity and bioaccumulative nature. Understanding the individual and co-exposure effects of these poorly characterized compounds is critical. This introduces this novel in-vitro-ML-in-silico framework designed to be tailored on PFAS toxicity. Using in vitro methods, morphological effects have been investigated after individual and co-exposures of a selected cell line to a group of PFAS. Cytotoxicity dose-response analysis enabled the detection of toxic effects, and changes have been traced via advanced high-content-image-based morphological profiling. Diverse databases and extrapolated experimental data have been employed to train and evolve the ability of this proposed framework. In vitro results have been integrated with machine learning to generate QSAR Read-Across in-silico models, and a Neural Network to predict complex PFAS-mediated biological disruptions and synergistic interactions, offering a multi-output data-driven approach for potential NAMs or New Approach Methodologies towards tailored-chemical risk assessments.