Computational modelling of detected persistent organic pollutants in adriatic pelagic fish: molecular and toxicokinetic perspectives for human health risk assessment.

Abstract

In this study, an integrative computational approach combining molecular descriptors, genetic algorithm - multiple linear regression (GA-MLR), and toxicokinetic simulations was employed to characterize the preliminary profile of Persistent Organic Pollutants (POPs) in various pelagic fish species from the Adriatic Sea. The molecular basis and toxicokinetic profile based on previously obtained experimental data on POP concentrations were followed by preliminary environmental characterization to assess the relationship between fish-specific parameters (e.g. species, lipid content) and determined pollutant concentrations. A GA coupled with MLR was applied to develop predictive models identifying the most influential physico-chemical properties of POPs that contribute to non-carcinogenic and carcinogenic health risks associated with chronic fish consumption. The toxicokinetic modelling and Hansen solubility parameters (HSPs) were used to assess the consumers' risk, gastrointestinal absorption and bioaccumulation. The use of GA-MLR and 3D-MoRSE descriptors directly linked chemical structure to environmental fate, bioaccumulation, and toxicological outcomes. The modelled concentrations of POPs in key organs and tissues were compared to in vivo data reported in the literature. This integrated analysis establishes a scientific basis for future toxicological and risk assessments of POPs in Adriatic pelagic fish, with emphasis on experimental validation and toxicokinetic profiling relevant to human exposure.