Abstract
The objectives of this analysis were to develop a population pharmacokinetic model (PPK) for tenofovir without using potentially unreliable patient reported dosing records and to retrace patient dosing history using pharmacokinetic simulations conditioned on protocol design constraints to assess patient adherence. MTN-001 is a multi-center, open label, 3-way cross over study comparing oral, vaginal and combination (oral and vaginal) administration of tenofovir in healthy women. It was reported that self-reported adherence in this study was high (94%), but serum concentrations indicated only 64% of participants used tablets consistently. A method based on superposition was applied to develop the population PPK considering only observable clinic visit dosing information. A two compartment model described the data well and the parameters agree with the literature reported values. Race was not a significant covariate on clearance. Retracing of the dosing history with 3-past dose resolution was not successful. Trial simulations with full adherence assumption predict a median Cmin of 68 ng/mL, which is in agreement with literature reports. Non-adherence at 25% resulted in 37-51% reduction in Cmin using one coin and two coin models, respectively. Population analyses should consider some method of correction for non-adherence to avoid biased estimates.