Identification, annotation and toxicity estimation of organic pollutants in human serum via non-target analysis

Environmental organic pollution causes a threat to the ecological environment, constrains social development and can also potentially harm human health. We applied non-target analysis to screen organic pollutants from the serum of 89 individuals, identifying 67 pollutants in the categories of industrial intermediates, plasticizers, surfactants, pharmaceuticals, pesticides, and exogenous pollutant metabolites. The detection rate of chemicals for industrial use (50.3%; 95% CI: 39.7, 60.8) was higher, reflecting the environmental exposure characteristics of the surrounding functional areas. In addition, 1168 potential pollutant features were annotated to 10 superclasses. Exposure levels of identified pollutants were semi-quantified by predicting response factors via machine learning model. Highly exposed pollutants involved various categories, especially pharmaceuticals due to their property of being easily absorbed by human body cross biological barriers. Toxicity of developmental toxicity, bioconcentration, mutagenicity and oral rat median lethal dose (LD₅₀) were predicted with the occurrence rates of 62.7%, 10.4%, 11.9% and 11.9% of the identified pollutants respectively. 4-[3-(Trifluoromethyl)benzyl]piperidine (industrial intermediate), risperidone (pharmaceutical), and aminocarb (insecticide) were predicted to have multiple toxic effects, which deserved attention and further hazard assessment. This study provides a comprehensive pattern of human exposure to organic pollutants, contributing to evaluate the health risks caused by pollutants to the population, thus providing data support for the monitoring and management of pollutants.