Integrating Adverse Outcome Pathway–Bayesian Network and Physiologically Based Pharmacokinetic Modeling for Quantitative Prediction of Developmental Neurotoxicity of F-53B under Human-Relevant Exposure Scenarios

F-53B (6:2 Cl-PFESA), a major replacement for perfluorooctanesulfonate (PFOS), is frequently detected in human cord blood, yet its developmental neurotoxicity risks remain poorly characterized. This study establishes a quantitative testing strategy coupling in vitro phenotypic profiling, transcriptomics, and a probabilistic adverse outcome pathway-Bayesian network (AOP-BN) with pregnancy physiologically based pharmacokinetic (PBPK) modeling. Using human embryonic neural stem cells, we found that F-53B induced dose-dependent oxidative stress and mitochondrial dysfunction, resulting in compromised neurogenesis. Transcriptomics supported these phenotypic results. We derived a benchmark dose of 3.26 μmol/g of protein for learning and memory impairment and utilized AOP-BN to quantify the probability of adverse outcomes across exposure gradients. By coupling this framework with a pregnancy PBPK model, we estimated fetal brain concentrations of 0.09-14.66 ng/mL (Q5-Q95) based on human biomonitoring data. While these levels remain below the point of departure for downstream neurogenic defects, the narrow margins of exposure for early molecular events, specifically ROS elevation, indicate potential safety concerns. Consequently, this study identifies oxidative stress as a sensitive trigger for F-53B toxicity and demonstrates a robust, mechanistically anchored framework for human-relevant risk assessment of emerging PFAS.