Building a pharmacogenetic model to describe the pharmacokinetics of digoxin

Emmanuelle Comets (1), Céline Verstuyft (2) and France Mentré (1)

(1) INSERM U738, University Hospital Bichat-Claude Bernard (2) Centre d'Investigation Clinique St Antoine, Paris, France

PDF of poster

Objectives: Over the past few years pharmacogenetic data has become increasingly available. In pharmacokinetic (PK) or pharmacodynamic (PD) studies, the focus is on a small number of Single Nucleotide Polymorphisms (SNP) or haplotypes to explain part of the interindividual variability.
The objectives of this work were, first, to review the literature on the statistical methods in this field, second, to apply model building strategies to study the pharmacokinetics of digoxin, a well-known probe for the activity of P-glycoprotein (PgP).

Methods: Papers dealing with pharmacogenetics and pharmacokinetics, published in Clinical Pharmacology and Therapeutics between 2003 and 2004, were retrieved based on the table of contents and the abstracts. In a second search, we scanned PubMed to retrieve all the papers using nonlinear mixed effect models (NLMEM) for pharmacokinetic modelling in the presence of genetic data.
The pharmacokinetics of digoxin have been previously analysed using non-compartmental methods [1], pooling three drug interaction studies in 32 healthy volunteers with extensive pharmacokinetic sampling. All patients were genotyped for the two main mutations in the MDR-1 gene which controls the expression of PgP (C3435T in exon 26 and G2677T/A in exon21). We used several approaches to include the genetic covariates in the pharmacokinetic model: stepwise selection with log-likelihood ratio tests, exhaustive search using the Bayesian Information Criterion (BIC) or the Akaike Information criterion (AIC), backward selection from a full model where covariates were selected using the Wald test. We used FO (NONMEM) and SAEM (MONOLIX) as estimation methods.

Results: In 2003-2004, 28 papers and 22 abstracts including pharmacogenetics and pharmacokinetics were published in Clinical Pharmacology and Therapeutics. A vast majority (93%) used non-compartmental analysis or observed data such as maximum concentrations and performed standard statistical tests. We found 15 papers in PubMed where NLMEM have been used, but they showed a variety of ways to deal with the categorical nature of pharmacogenetic data, resulting in different coding schemes. Model building strategies were also variable.
The PK model for digoxin was a two-compartment model. Using stepwise selection with NONMEM we found that carriers of the TT genotype on exon 26 exhibited an increase of 30% in bioavailability compared to the other genotypes. We did not find any other effect when treating the genetic data as haplotype instead of SNPs, but because of linkage disequilibrium there was limited additional information in haplotypes when compared to exon 26 alone. We will illustrate the differences in covariate selection with the other approaches, using both FO and SAEM.

Conclusion: A wide array of methods have been used to deal with pharmacogenetic data when studying the pharmacokinetics of drugs, with no standard approach. NLMEM can provide a better understanding of the influence of genetic data on pharmacokinetics. For digoxin, NLMEM showed an increase in bioavailability in homozygote carriers of the T allele for the exon 26 polymorphism in MDR-1, explaining the differences in AUC observed previously [1].

[1] C Verstuyft, M Schwab, E Schae eler, R Kerb, U Brinkmann, P Jaillon, C Funck-Brentano, and L Becquemont. Digoxin pharmacokinetics and MDR1 genetic polymorphisms. Eur J Clin Pharmacol, 58(12):809-12, 2003.

Reference: PAGE 14 (2005) Abstr 822 [www.page-meeting.org/?abstract=822]

Poster: poster