Alicja Puszkiel (1), Cécile Arellano (1), Henri Roché (2), Christelle Vachoux (1), Alexandre Evrard (3), Valérie Le Morvan (4), Etienne Chatelut (1,5), Fabienne Thomas (1,5), Mélanie White-Koning (1)
(1) Cancer Research Center of Toulouse (CRCT), Inserm U1037, Université Paul Sabatier, Toulouse, France; (2) Department of Medical Oncology, Institut Universitaire du Cancer de Toulouse – Oncopole, France; (3) Laboratoire de Biochimie et Biologie Moléculaire, CHU Carrémeau, Nîmes, France; (4) Institut Bergonié, Bordeaux, France; (5) Institut Claudius Regaud, Institut Universitaire du Cancer de Toulouse – Oncopole, France.
Objectives: To develop a population pharmacokinetic (PopPK) model describing the disposition of tamoxifen (TAM) and three of its metabolites in breast cancer patients and to investigate the impact of genetic polymorphisms of CYP2D6 on plasma levels of active metabolites: 4-hydroxy-tamoxifen (4-OH-TAM) and endoxifen (ENDO).
Methods: PK data for TAM and three metabolites (N-desmethyl tamoxifen [NDT], 4-OH-TAM and ENDO) come from a prospective, multicenter, 3-year follow up study including 888 patients starting treatment with TAM at 20 mg once daily. This report focuses on preliminary results from 209 patients. Plasma samples were available for each patient every 6 months during a 3-year period, i.e. at inclusion (before treatment start) and 24-hours post-dose at 6, 12, 18, 24, 30 and 36 months. Plasma concentrations of TAM and metabolites were measured by a validated UPLC-MS/MS method [1]. Patients were genotyped at inclusion for single nucleotide polymorphisms (SNP) in the gene encoding for CYP2D6 and classified into poor (PM), intermediate (IM), extensive (EM) or ultrarapid metaboliser (UM) based on the presence of functional (*1), reduced function (*9, *10, *17, *41) or non-functional alleles (*4, *6, *7) and the number of CYP2D6 copies (*5 or gene duplication). Concentration-time data were analysed using non-linear mixed-effects modelling in NONMEM 7.4 using FOCE with interaction option. The impact of CYP2D6 activity on certain metabolic rate constants was tested according to the power equation and was considered significant if the objective function value (OFV) and the interindividual variability (IIV) of the estimate of corresponding parameter decreased significantly.
Results: PK data for TAM and the three metabolites (n = 934 samples) were analysed simultaneously with a four-compartment model with ENDO formed from either NDT or 4-OH-TAM, themselves formed from TAM. The absorption of TAM from gut compartment was described by a first-order rate constant ka. Volumes of distribution (Vd) of the metabolites were not identifiable and were fixed to a value of Vd of TAM previously reported in the literature [2]. The formation of metabolites was described by linear conversion as follows: k23 is the conversion rate constant of TAM to NDT, k24 TAM to 4-OH-TAM, k35 NDT to ENDO and k45 4-OH-TAM to ENDO. Since TAM, NDT and 4-OH-TAM can follow other metabolising pathways which do not lead to the formation of the metabolites accounted for in the model, the inclusion of elimination rate constants for TAM (k20), NDT (k30) and 4-OH-TAM (k40) were tested by comparing the OFV of models with and without each of these constants. Inclusion of k30 resulted in improved goodness-of-fit plots and a significant decrease in the OFV, therefore it was retained in the model. The elimination rate constant of ENDO (k50) was fixed to a value previously reported in the literature [2]. The residual error was coded separately for each compound using a proportional model. Inclusion of CYP2D6 activity as covariate on k35 significantly decreased OFV (∆OFV = -145). IIV in k35 decreased from 71.3% in the base model to 47.5% in the final model. The mean (relative standard error, RSE%) [CV% for IIV] estimates of the final model were: k23 = 6.94 x 10-3 h-1 (2%) [29.5%], k24 = 3.64 x 10-5 h-1 (34%) [73.6%], k35 = 4.48 x 10-4 h-1 (5%) [47.5%], k30 = 3.94 x 10-3 h-1 (3%) [44.6%] and k45 = 1.26 x 10-3 h-1 (33%). k35 decreased by 84.9% and 69.9% in PM and IM patients, respectively, and increased by 58.0% in UM patients, compared to EM patients. Residual error estimates were 25.3% (7%), 26.3% (6%), 29.6% (8%) and 31.5% (7%) for TAM, NDT, 4-OH-TAM and ENDO, respectively. The correlation between TAM and NDT residual error was 79.1% (8%) and between 4-OH-TAM and ENDO 81.7% (10%). Eta-shrinkage was less than 7% and epsilon-shrinkage was less than 9%.
Conclusions: The developed model gave a good description of steady-state concentrations of TAM and three of its major metabolites. CYP2D6 activity significantly influenced the formation of ENDO from NDT. Future analyses will focus on inclusion of the three remaining metabolites in order to develop a joint PK model describing major metabolism pathways of TAM. The developed model could be useful to establish recommendations for monitoring of plasma concentrations of TAM active metabolites and to individually adjust the dose based on previously proposed target plasma ENDO concentration [3].
References:
[1] Arellano C, Allal B, Goubaa A, Roché H, Chatelut E. An UPLC–MS/MS method for separation and accurate quantification of tamoxifen and its metabolites isomers. J. Pharm. Biomed. Anal. Elsevier; 2014;100:254–61.
[2] Dahmane E, Zaman K, Perey L, Bodmer A, Leyvraz S, Eap CB, et al. Population Pharmacokinetics of Tamoxifen and three of its metabolites in Breast Cancer patients. 2014;89:3402. Available from: http://www.page-meeting.org/pdf_assets/6546-Poster_Elyes Dahmane.pdf
[3] Madlensky L, Natarajan L, Tchu S, Pu M, Mortimer J, Flatt SW, et al. Tamoxifen metabolite concentrations, CYP2D6 genotype, and breast cancer outcomes. Clin. Pharmacol. Ther. 2011;89:718–25.
Reference: PAGE 27 (2018) Abstr 8616 [www.page-meeting.org/?abstract=8616]
Poster: Drug/Disease Modelling - Oncology