Abstracts

Rationale: Therapeutic drug monitoring (TDM) is often employed to determine if changes are needed in antiseizure medication (ASM) dosing to maintain a patient’s established effective concentration. The changing pharmacokinetics of ASMs during pregnancy occurs on a short timescale relative to chronic therapy which can make TDM complex. As ASMs exhibit differing half-lives with no standardized sample collection times, there is potential for bias in measured concentrations due to timing of samples collection relative to dose administration. These biases can compound if concentrations are used for further exposure-response analyses.

Methods: A one-compartment pharmacokinetic model was used to simulate concentrations to demonstrate the theoretical differences of trough (i.e., consistent times) versus convenience (i.e., varied times) sampling. Levetiracetam was used as the model drug with a volume of distribution of 50 L and clearance of 4 L/h (half-life=12.5 h) increasing by 0.25 L and 0.05 L/h per week gestation, respectively. Simulated samples were collected with random time after dose (TAD) at preconception and 4, 8, 12, 16 weeks gestation to model convenience sampling for TDM. The difference between preconception and each sample was calculated with direction of change recorded. Levetiracetam concentrations (normalized to 2000 mg/day dose) collected during the prospective, observational, 20-site Maternal Outcomes and Neurodevelopmental Effects of Antiseizure Drugs study (n=147) were used to show differences in convenience and near trough samples in practice. Samples collected 8+ hours after dose were considered trough samples. Sampling strategies were compared using mean and standard deviation.

Results: When TAD for samples was kept constant, 100% of patients had lower concentrations during pregnancy, as expected from increased clearance. When TAD varied between samples, only 56%, 61%, 66%, 70% of simulated patients had concentrations lower than preconception at 4, 8, 12, 16 weeks, respectively. Increases in volume of distribution led to increased trough concentrations and decreased peak concentrations, while increases in clearance led to decreased average concentrations. The mean dose normalized concentration of convenience samples was 16.0±8.4 ug/mL (mean±standard deviation) versus 11.1±6.4 ug/mL of trough samples.


Conclusions: Theoretically, volume of distribution affects range between peak and trough concentrations, and clearance affects average concentration. This leads to particularities for both convenience and trough sampling during TDM for pregnancy. Increased variability in convenience sampling without considering TAD can conceal concentration changes, while sampling at consistent times, such as trough sampling, cannot differentiate whether a concentration change is due to changes in clearance or volume of distribution. Although sampling at consistent times can be used for simple TDM in clinical practice, convenience sampling with inclusion of TAD analysis would provide a more complete picture of pregnancy pharmacokinetics in clinical studies.


Funding: NIH/NINDS/NICHD U01 NS038455, NIH/NICHD R01HD105305