Development and prospective validation of a read-across approach to assess the in vivo toxicokinetic profiles of chemicals in humans.

The use of read-across is established for assessing toxicological hazards based on structural and biological similarities. Toxicokinetic (TK) read-across extends this concept to infer in vivo TK of target compounds from source analogs with available data, but its application to non-pharmaceutical chemicals remains limited. Here, we evaluated TK read-across for non-pharmaceutical compounds, focusing on cosmetic ingredients across multiple administration routes and diverse TK properties, including poor oral bioavailability. A TK analog workflow was applied to identify analogs, with suitability ratings assigned based on structure, reactivity, metabolism, and physicochemical properties. Case studies showed that "suitable" pairs exhibited comparable in vivo TK and in vitro absorption, distribution, metabolism and excretion (ADME) characteristics, typically within 0.5-2-fold differences. Pairs rated "suitable with interpretation" displayed >2-fold TK differences, consistent with structural or ADME dissimilarities. "Suitable with pre-condition" pairs were poor TK analogs, as analogs were often metabolites or metabolic precursors of targets, with >5-fold differences in clearance. "Not suitable" pairs showed divergent TK profiles and structures. A human study using a "suitable" pair demonstrated that terpinolene TK aligned with limonene data within two-fold. Overall, this workflow supports evidence-based analog selection and enables TK read-across to inform safety and TK assessments when human data are limited.