Mechanistic prediction of apixaban and rivaroxaban secretion into human milk and infant systemic exposure through an integrated physiologically based pharmacokinetic framework.

Accurately characterizing the transfer of apixaban and rivaroxaban into human milk is essential for evaluating their safety during lactation and for guiding anticoagulant therapy in breastfeeding patients. This study aimed to establish a mechanistic physiologically based pharmacokinetic (PBPK) modeling approach linking maternal exposure, mammary drug transport, and infant systemic levels for both anticoagulants. Passive permeability and efflux transport across the mammary epithelium were measured using an in vitro human mammary epithelial cell model, providing information on diffusion and transporter-mediated clearance, whereas milk protein binding was determined using rapid equilibrium dialysis to account for unbound drug available for transfer. These experimentally derived parameters were incorporated into an adult PBPK lactation model, which closely reproduced observed plasma and milk pharmacokinetics across dosing regimens, with simulated milk-to-plasma ratios deviating less than 10% from clinical measurements. The simulated maternal breast milk concentration-time profiles were subsequently used to drive a pediatric PBPK model, enabling an estimation of infant exposure with 2-hour feeding intervals. Simulations showed low infant plasma concentrations for both drugs, with rivaroxaban producing relative infant daily dose values below the commonly accepted 10% safety threshold, whereas apixaban exceeded this threshold with high absolute systemic exposure in infants. Overall, this work demonstrates that experimentally informed PBPK modeling provides a robust and quantitative framework to predict in vivo maternal-infant drug exposure. This approach enhances the ability to assess lactational safety for anticoagulants and establishes a generalizable strategy for evaluating other drugs used during breastfeeding. SIGNIFICANCE STATEMENT: This study establishes an integrated in vitro-in silico PBPK modeling framework to predict drug transfer into human milk and resulting infant exposure. By incorporating permeability, transporter activity, and milk protein binding into maternal and pediatric models, the approach accurately reproduces milk pharmacolinetics and enables mechanistic prediction of infant exposure. The model reveals substantially higher infant exposure to apixaban than rivaroxaban, supporting quantitative assessment of lactational drug safety.