A novel framework for polycyclic aromatic hydrocarbon dose–effect assessment: A case study of benzo[a]pyrene

• Data standardization-based novel framework for dose–effect modeling was proposed. • The framework predicted human health effects of BaP exposure with 92.31% accuracy. • Females had significantly higher carcinogenic sensitivity than males to oral BaP. • Digestive and metabolic organs were most sensitive to oral BaP’s carcinogenicity. • Refined chronic reference dose and carcinogenic slope factor for BaP were derived. Benzo[ a ]pyrene (BaP), a representative polycyclic aromatic hydrocarbon (PAH), is a priority environmental pollutant with well-documented carcinogenic and non-carcinogenic effects. Despite extensive research, inconsistencies in exposure metrics, effect indicators, and data heterogeneity have hindered the establishment of standardized reference values for reliable health risk assessment. In this study, we developed a systematic framework integrating effect size standardization, effect classification, and benchmark dose (BMD) modeling to characterize the dose–effect relationship of BaP. By converting multi-source toxicological data to standardized effect sizes, generalized additive models (GAMs) were used to explore nonlinear dose–effect relationship. The results revealed tissue-specific sensitivities, with digestive and metabolic organs exhibiting higher carcinogenic risk, while neurological and metabolic endpoints demonstrated particular vulnerability to non-carcinogenic effects. The mean female oral carcinogenic slope factor (CSF o ) was 2.61 times higher than the mean male CSF o ( P < 0.05), indicating greater susceptibility of females to BaP via oral exposure. The derived chronic reference dose (CRfD) value (6.89 × 10 −4 mg/kg·day) and CSF values (oral intake: 0.252–0.290, inhalation: 1.208, and dermal contact: 0.252–0.290 (mg/kg·day)⁻ 1 ) represent refined reference values applicable to human health risk assessment. Furthermore, when validated against 13 occupational epidemiological endpoints, our predictions demonstrated an overall predictive accuracy of 92.31% and a sensitivity of 100% in identifying true human health hazards. This study provides a reproducible and transferable framework for pollutant toxicity assessment and underscores the need for standardized, sex-specific approaches to ensure scientific validity in environmental health risk evaluation.