Population Pharmacokinetic Analysis Data analysis was performed using NONMEM version 7.5.0 (ICON Development Solutions, Hanover, MD, USA). To account for the differences in molecular weight between S-ketamine and the metabolites, concentration data were converted from ng/ml to nmol/ml. Data were analyzed in a stepwise fashion. First, S-ketamine data were analyzed, followed by the addition of S-norketamine and subsequently S-hydroxynorketamine. The routing of S-ketamine consists of two parts: one direct pathway from the OTF into plasma, and one indirect pathway in which some S-ketamine is stored in saliva which is ingested and absorbed via the gastrointestinal tract. Since S-norketamine was not administered, theoretically, the volume of the central S-norketamine compartment (VN1) was not identifiable. However, since we assumed that 80% of S-ketamine was metabolized VN1 is identifiable.2 The same applies for S-hydroxynorketamine: since we assumed that 70% of Snorketamine is transformed into S-hydroxynorketamine,10 the volume of the central S-hydroxynorketamine compartment is identifiable. The number of Sketamine, S-norketamine and S-hydroxy-norketamine compartments as well as the intermediary metabolism compartments was determined by goodness-of-fit criteria, i.e., a significant decrease in objective function value (OFV) calculated as -2 log likelihood (χ2 test), visual inspection of the data fits and goodness-offit plots (normalized prediction distribution error vs time plots, normalized prediction distribution error vs predicted plots, and predicted vs measured plots). Moreover, prediction-variance-corrected visual predictive checks (VPCs) were performed by simulating 1,000 data sets based on the model parameters and comparing the simulated quantiles with those of the true data. P <0.01 were considered significant. FOCE-I (first-order conditional estimation with interaction) was used to estimate model parameters. To account for inter-individual and inter-occasion variability (IOV), random effects were included in the model with an exponential relation: θi=θ×exp(ηi+ηiov), where θi is the parameter for individual i, θ is the population parameter, ηi is the random difference between the population and individual parameter, and ηiov is the difference between θi and θ as a result of IOV. In addition, proportional and additive errors were evaluated for each separate analyte to account for residual variability. The proportional and combined proportional and additive error models were described by Yij =Pij × (1+ϵij) and Yij =Pij×(1+ϵ1ij )+ϵ2ij , respectively, where Yij is the jth observed plasma concentration for individual i, Pij is the corresponding model prediction, and ϵij is the residual error. Inter-occasion variability was determined for the S-ketamine and S-norketamine absorption parameters, while it was determined for all S-hydroxynorketamine model parameters. 18
RkJQdWJsaXNoZXIy MTk4NDMw