Development of a pharmacokinetic population model for atazanavir on a drug user population

Objectives: Atazanavir (ATV) is one of the most frequently used antiretroviral in the treatment of AIDS. This study primarily aimed to develop a structural pharmacokinetic population model for ATV. Additional analysis focused on the development of a covariate model for an HIV-infected drug user population.

Methods: A structural model was developed using NONMEM VII on the full profiles of 20 non-HIV infected subjects who were given a single oral dose of ATV/ritonavir (200 ATV measures). The model discrimination was based on the objective function value (OFV), the goodness of fit plots, and parameter estimates. A covariate model was developed using the forward inclusion backward elimination method (p<0.05) on a second dataset with 66 HIV-infected subjects (315 ATV measures) who were given atazanavir daily in addition to their pre-existing anti-HIV treatment.

Results: A two-compartment model with first order absorption, lag time, inter-subject variability (ISV) with log-normal distribution and heteroscedastic residual error, was found to be a better fit (R² = 97.4 with IPRED) than a one-compartment model with the same properties (-62.4 OFV). Using the HIV-infected population, an effect of ritonavir on the clearance and an inter-occasion variability were added to the model before the covariate selection. Over our primary analysis, numerous covariates were found to have an effect on the clearance and on the first order absorption rate constant (p<0.05). However, no significant effect was found on the volume of distribution.

Conclusion: Most of the papers found in the literature on ATV describe one-compartment pharmacokinetic models. Our study showed that the pharmacokinetic predictions of ATV concentration could be improved by the use of a two-compartment model and significant covariates especially in a drug user population.