Physiologically Based Pharmacokinetic (PBPK) Modelling to Predict Breast Milk Exposure of Valproic Acid.

Breastfeeding has many positive effects on the health and well-being of mother and baby. During lactation, a pharmacological therapy may be required. Currently there is a knowledge gap regarding the safety of medicines during lactation, which may lead to unnecessary cessation of breastfeeding or incur unknown risks to the child when continuing breastfeeding while on medication. The Innovative Medicines Initiative (IMI) ConcePTION aims to address this and other issues related to medication use during pregnancy and lactation. The objective of this study performed in the IMI ConcePTION framework was to explore the utility of Physiologically-Based Pharmacokinetic (PBPK) modelling in the generation of information regarding the safety of medicines (e.g. valproic acid, VPA) during breastfeeding. To achieve this goal, PK-Sim® (v 9.1) (Open Systems Pharmacology) was used to first develop a reference PBPK model for VPA in adults using 7 training datasets. The median fold errors of the Area Under the Curve (AUC) and maximum plasma concentration (Cmax) for the 17 validation datasets were 0.89 (range: 0.74-0.134) and 0.90 (range: 0.60-1.36), indicating a good predictive performance. In a next step, this PBPK model was extended with a breast compartment (including a subcompartment representing the breast milk) according to the work from Dallmann et al. (2018). Transfer of medicines between the milk and the intracellular compartment was parametrized with a clearance for secretion to milk (Clsec), and a clearance for reuptake (Clre). These clearance values were determined based on physicochemical properties using the empirical model from Koshimichi et al. (2011). Clsec and Clre were 235.05 and 1452.78 mL/h. The PBPK model for lactating women was used to predict exposure of VPA in plasma and human breast milk. The simulated milk-to-plasma (M/P) ratio was 0.016, similar to reported M/P ratio's (range: 0.007-0.25). Finally, the daily infant dose was calculated based on the simulated concentration in the milk and used as input for an infant PBPK model. The relative infant dose was 1.4 %. Average steady-state infant plasma concentration was estimated to be 0.83 mg/L (relative infant exposure: 1.2 %), compared to a wide range of observations starting from undetectable concentrations up to 41 mg/L. PBPK modelling was successfully used to predict the transfer of VPA into the human breast milk, and to estimate subsequent infant exposure via breastfeeding. Based on ongoing application to 11 other medicines, a generic workflow for PBPK-based breast milk transfer of medicines has been developed. Current efforts to determine the in vitro permeability of VPA across the blood-milk barrier may obviate the current need for (sparse) milk concentrations. This study illustrated the potential of PBPK modelling as a tool to fill the knowledge gap regarding the exposure to medicines in lactating women and their infants.