Population pharmacokinetic/pharmacogenetic model for optimization of efavirenz therapy in Caucasian HIV-infected patients

Abstract

Despite extensive clinical experience with efavirenz (EFV), unpredictable interindividual variabilities in efficacy and toxicity remain important limitations associated with the use of this antiretroviral. The purpose of this study was to determine the factors affecting EFV pharmacokinetics and to develop a pharmacokinetic/pharmacogenetic (PK/PG) model in a Caucasian population of HIV-infected patients. In total, 869 EFV plasma concentrations from 128 HIV-infected patients treated with EFV were quantitatively assessed using a validated high-performance liquid chromatography technique. All patients were genotyped for 90 single nucleotide polymorphisms (SNPs) in genes coding for proteins involved in the metabolism and transport of EFV, using a MassArray platform provided by Sequenom. The influence of these polymorphisms on EFV pharmacokinetics and the effects of demographic, clinical, biochemical, lifestyle, and concurrent drug covariates were evaluated. Plasma concentrations were fitted by a one-compartment model, with first-order absorption and elimination using nonlinear mixed-effect modeling (NONMEM program). The CYP2B6*6 allele, multidrug resistance-associated protein 4 (MRP4) 1497C → T, and gamma-glutamyltranspeptidase (GGT) were identified as major factors influencing the apparent EFV oral clearance (CL/F), reducing the initial interindividual variability by 54.8%, according to the model CL/F = (12.2 - 0.00279 · GGT) · 0.602(CYP2B6*6 [G/T]) · 0.354(CYP2B6*6 [T/T]) · 0.793(MRP4 1497C → T), where CYP2B6*6 [G/T], CYP2B6*6 [T/T], and MRP4 1497C → T take values of 0 or 1 to indicate the absence or presence of polymorphisms. The detailed genetic analysis conducted in this study identified two of 90 SNPs that significantly impacted CL/F, which might indicate that the remaining SNPs analyzed do not influence this PK parameter, at least in Caucasian populations with characteristics similar to those of our study population.

Fig. 1.

Scatterplot of measured EFV plasma concentrations (Cobs) versus population (A) and individual (B) EFV concentrations predicted (Cpred) by the basic (left) and the final (right) models.

Fig. 2.

Normalized prediction distribution errors (NPDE) applied to the three relevant dose groups defined by the CYP2B6 genotypes.

Fig. 3.

Mean (solid lines) and 90% prediction intervals (dotted lines) of the simulated plasma concentration-time profiles of efavirenz for patients with CYP2B6*6 G/G (A), CYP2B6*6 G/T (B), and CYP2B6*6 T/T (C), who received doses of 600 mg/day, 400 mg/day, and 200 mg/day, respectively.