High-throughput genetic screening and pharmacokinetic population modeling in drug development
Tessier A (1, 2), Bertrand J (3), Fouliard S (2), Comets E (1), Chenel M (2)
(1) Univ Paris Diderot, Sorbonne Paris Cité, UMR 738, F-75018 Paris, France and INSERM, UMR 738, F-75018 Paris, France; (2) Clinical Pharmacokinetic Department, Institut de Recherches Internationales Servier, Suresnes, France; (3) University College London, Genetics Institute, London, UK
Objectives: To develop a population PK model and integrate a large number of SNPs, genotyped in clinical studies, as covariates in this model.
Methods: The dataset included 78 subjects receiving molecule S (Servier laboratories) in Phase I studies (60 as single dose, 18 as repeated doses daily over 21 days). Subjects were sampled extensively on Day 1 and 21 (for repeated doses) along with additional trough samples. They were genotyped for 176 Single Nucleotide Polymorphisms (SNPs) using a DNA microarray. Forty genes corresponding to markers of metabolic enzymes, carriers and nuclear receptors were screened for their influence on PK parameters. A population PK model was developed to describe the evolution of concentrations data, and the parameters were estimated using NONMEM 7.2 [1], FOCE-I. Here the number of observations (PK profiles) is less than the number of covariates to explore (SNPs). Thus the analysis requires reducing the number of covariates to include. We used the algorithm proposed by Lehr et al. [2] based on a preliminary screening using univariate ANOVA on the individual Bayesian estimates (EBE).
Results: The pharmacokinetics of drug S followed a two-compartment model with zero order absorption (D1) and a lag time (Tlag). Non-linearity in the PK with dose was modeled through a variable bioavailability (F). The PK profile showed a rebound at approximately 24h, which was described assuming a gallbladder compartment with an emptying time-window, corresponding to enterohepatic circulation (EHC). Inter-individual variability was estimated for F, D1, Tlag, the intercompartmental clearances between central and peripheral compartment (Q) and between central and gallbladder compartment (QGB), and the oral clearance (CL). Three genotypic models (additive, dominant and recessive) were tested for SNPs on EBE. One SNP was associated additively with CL on the gene coding for metabolism enzyme NAT1 and another recessively with QGB on the gene coding for nuclear receptor VDR.
Conclusions: Using NLME modeling rather than observed AUC to study the impact of many genetic variants during the development of drug S (nonlinearity, EHC) enabled identifying how the genetic markers impact the different phase of ADME (Absorption, Distribution, Metabolism, Excretion) process. Effects of SNPs revealed by this work will be explored through focused in vitro studies.
References:
[1] Sheiner LB, Rosenberg B, Melmon KL. Modelling of individual pharmacokinetics for computer-aided drug dosage. Computers and Biomedical Research 1972;5(5):411-459
[2] Lehr T, Schaefer HG, Staab A. Integration of high-throughput genotyping data into pharmacometric analyses using nonlinear mixed effects modeling. Pharmacogenetics and Genomics 2010;20(7):442-450
