Pharmacokinetic variability of valproate during pregnancy – Implications for the use of therapeutic drug monitoring

•The pharmacokinetic variability of valproate as total serum concentration/dose-ratio was extensive, 13-fold, 51 pregnancies in 33 women.•Total serum concentrations were most commonly measured.•Unbound concentrations were only measured in 10% of the cases.•The unbound concentration increased up to 5-fold during two pregnancies in one woman.•Relying on total serum concentrations only may be misleading in evaluation of efficacy and fetal exposure of valproate. Use of valproate (VPA) in women of childbearing age is restricted due to dose-dependent risk of teratogenicity. The purpose of this study was to characterise pharmacokinetic variability of VPA in pregnancy, and discuss use of therapeutic drug monitoring (TDM) as guidance to exposure in women. Measurements of trough total and unbound VPA concentrations before, during and after pregnancy, at assumed steady-state were collected from the TDM-database (2006–2016) at the National Center for Epilepsy in Norway. Additional clinical data were obtained from the Oppland county Perinatal Database (1994–2011). Data from 51 pregnancies in 33 women aged 19–40 years were included. Each woman underwent 1–4 pregnancies, and 1–7 measurements per pregnancy were performed. The variability in total concentration/dose (C/D)-ratios between women was 13-fold, and intra-patient variability extensive. Total C/D-ratios were reduced by 46% from before pregnancy to third trimester (0.48–0.29 μmol/L/mg). Unbound concentrations of VPA were only requested in 10% of the pregnancies. Repeated measurements from two pregnancies in one women revealed increased unbound concentration of VPA during pregnancy. There were 19 with idiopathic generalized epilepsy and two focal based on clinical data from 21 women and 38 pregnancies; 1 major congenital malformation was noted. There is pronounced pharmacokinetic variability of VPA during pregnancy. Unbound concentrations are rarely requested. TDM should be used by measurements of both total and unbound concentrations since total concentrations may be misleading for efficacy and fetal exposure of VPA.