A Method to Calibrate Chemical-Agnostic Quantitative Adverse Outcome Pathways on Multiple Chemical Data.

Quantitative Adverse Outcome Pathways (qAOPs) can support next-generation risk assessment by integrating new approach methods (NAMs) for deriving points of departure. To be useful, a qAOP should be chemical-agnostic. However, existing qAOP studies often pool multichemical data without adequately addressing cross-chemical heterogeneity. Consequently, pathway relationships become obscured by heterogeneity-induced variations, thereby compromising model reliability and generalizability. We developed a calibration approach to address this challenge by leveraging hierarchical structures to systematically separate chemical-specific heterogeneity from the underlying pathway effects. Chemical-specific deviations are explicitly modeled as random effects, enabling the extraction of pathway-level parameters that represent core mechanistic relationships independent of the individual chemical properties. We demonstrated through a simulation study that performance differences between models with and without hierarchical calibration can reveal the magnitude of the heterogeneity in the data. When heterogeneity is substantial, an uncalibrated qAOP should not be considered truly chemical-agnostic in practice, as it confounds pathway-level effects with chemical-specific variations. Finally, we demonstrated the application of this calibration approach through deriving points of departure to a case study of nonmutagenic liver tumor qAOP.