Mega-model of nevirapine population pharmacokinetics
Elin Svensson(1), Jan-Stefan van der Walt(1), Karen Barnes(2), Karen Cohen(2), Tamara Kredo(2,3), Phumla Sinxadi(2), Mats Karlsson(1), Paolo Denti(2)
(1)Department of Pharmaceutical Bioscience, Uppsala University, Sweden; (2)Division of Clinical Pharmacology, Department of Medicine, University of Cape Town, South Africa; (3) South African Cochrane Centre, South African Medical Research Council, Cape Town, South Africa
Objectives: To develop a mega-model, i.e. a model driven by raw data from multiple sources, of nevirapine (NVP) pharmacokinetics (PK) in a diverse population of HIV-infected South African adults on anti-retroviral therapy (ART).
Methods: Data of NVP plasma concentrations from four different South African clinical studies including 116 patients, some intensively and some sparsely sampled were available. All patients were HIV-infected and on ART regimens including 200 mg NVP BID. Sampling was conducted at steady state in all the studies, but sampling schedule and conditions varied. A population model was developed in NONMEM7, using a stepwise approach, starting with the rich data and subsequently adding the sparse data. Inclusion of significant covariates was explored. Model development was guided by goodness of fit and VPCs.
Results: NVP PK was described by a one compartment disposition model with absorption through two transit compartments and first-order elimination. The model included a mixture of two sub-populations with different typical values of clearance (CL): 3.14 L/h/60 kg and 1.39 L/h/60 kg, respectively. The proportion of the lower CL sub-population was estimated, although with poor precision, to 14% (RSE 47%). Volume of distribution (V) was 91.4 L/60kg, while the mean transit time (MTT) was 2.41 h if the drug was ingested with food, and 0.59 h when taken in fasting conditions. Concomitant antitubercular-treatment including rifampicin decreased bioavailability (F) by almost 40%, and between-subject variability (BSV) was supported in F for patients on such antitubercular-treatment. BSV in CL (19%) and between-occasion variability (BOV) in MTT (61%) and F (24%) were also significant. The residual error (proportional) was estimated to 7.8% with BOV of 20%. The model included allometric scaling with body weight of CL and V.
Conclusions: This is the first population model including a mixture on clearance to describe individuals with extremely high concentrations, potentially due to genetic polymorphisms. The model identified effects of antitubercular-treatment and food with precise and stable estimates of the typical values and variability of the model parameters. This model may serve as a starting point for development of a larger mega-model, possibly including data from other countries.