Prediction and modeling on bioaccumulation of chemicals in aquatic organisms: A critical review

Chemical bioaccumulation, which links internal exposure to hazards, plays a crucial role in ecological and human health risk assessment. Traditional bioaccumulation evaluations rely on laboratory-simulated exposure experiments and field monitoring studies, facing limitations in high cost, time constraints, and ethical concerns. Mechanistic toxicokinetic (TK) models and data-driven quantitative structure-activity relationship (QSAR) approaches have emerged as efficient alternatives. The review synthesizes decades of progress in TK and QSAR models for bioaccumulation prediction of chemicals in aquatic organisms, encompassing theoretical foundations, mathematical equations, practical applications, and current research gaps. While traditional compartmental TK models provide simplicity, physiologically-based toxicokinetic (PBTK) models enhance accuracy by incorporating species-specific physiology and chemical-dependent biochemical parameters. Existing PBTK models are currently limited to 10 fish species and 133 chemicals across 29 categories, with applications focused on internal concentration dynamics predictions, cross-species extrapolation, in vitro to in vivo extrapolation, and toxicity integration. Major limitations include a taxonomic bias toward fish, inadequate contaminant coverage, oversimplified environmental parameterization, incomplete mechanistic representations, and parameter scarcity. QSAR models have evolved from linear regression to machine learning approaches, achieving superior prediction performance through enhancing molecular representations and incorporating advanced algorithms. Nonetheless, challenges remain in TK parameter prediction and high-quality data acquisition. The review offers valuable insights and methodological advancements to support chemical risk assessment.