Anna Mueller-Schoell (1,2), Lena Klopp-Schulze (1), Robin Michelet (1), Wilhelm Huisinga (3), Markus Joerger (4), Charlotte Kloft (1)
(1) Dept. of Clinical Pharmacy and Biochemistry, Institute of Pharmacy, Freie Universitaet Berlin, Germany and (2) Graduate Research Training Program PharMetrX, Germany, (3) Institute of Mathematics, University of Potsdam, Germany (4) Medical Oncology and Clinical Pharmacology, Dept. of Internal Medicine, Cantonal Hospital St. Gallen, Switzerland
Objectives: In 2011, Madlensky et al. [1] established an efficacy threshold for endoxifen, (ENDX), the active metabolite of tamoxifen (TAM). Minimum concentrations at steady state (CSS,min ENDX) >5.97 ng/mL were associated with a 26% lower breast cancer (BC) recurrence rate. Six years later, de Vries Schultink et al. [2] developed the Antiestrogenic Activity Score (AAS), considering TAM, its three major metabolites (including ENDX) and their respective antiestrogenic potencies. Based on the same dataset, a relation between AAS ≥1798 and a 31% lower BC recurrence rate compared to AAS <1798 was identified. Still, current therapeutic drug monitoring approaches use the ENDX threshold as target attainment (TA) metric [3]. The presented simulation study aimed to investigate the impact of using the AAS vs. the ENDX threshold as TA metric on dose selection in model-informed precision dosing. Individual TAM doses could range between 5 and 120 mg (all once-daily (q.d.)), considering available tablet formulations and maximum reported doses in TAM dose escalation trials [4]. Differences in dose selection were investigated for the population as whole and for genotype-predicted CYP2D6 normal, intermediate and poor metabolisers (gNM, gIM, gPM), respectively. For the genotype-to-phenotype classification based on patients’ CYP2D6 activity scores (AS), we used the most recent Clinical Pharmacogenetics Implementation Consortium guideline for CYP2D6 and tamoxifen therapy (gNM: AS ≥1.5, gIM: AS=0.5-1.0, gPM: AS=0) [5].
Methods: A previously developed NLME-PBPK model of TAM and its three major metabolites [6] was used to simulate TAM treatment in 10.000 virtual BC patients. Covariates (age and CYP2D6 AS) in the virtual population were generated to represent the respective frequencies observed in a pooled database of 6 clinical studies [7]. In the first simulation step, virtual patients received CYP2D6 phenotype-adjusted initial doses (gNM: 20 mg, gIM: 40 mg, gPM: 60 mg; all q.d.). In step 2, virtual minimum concentrations of TAM and its three major metabolites were measured after 2, 4 and 6 weeks of treatment. CSS,min ENDX and AAS at 6 months were predicted for each virtual patient for the full dose range by Bayesian Forecasting (BF), considering patient covariates, prior knowledge (model parameters) and virtually measured concentrations of (1) TAM and ENDX in the ENDX-guided dosing group and (2) TAM and its three major metabolites in the AAS-guided dosing group. The lowest respective doses required for TA according to (1) the ENDX (CSS,min ENDX >5.97 ng/mL) or (2) the AAS (AAS>1798) threshold were selected as adjusted individual doses for each patient. Finally, the predictive performance of BF using either TA metric was assessed by calculating “true” TA at 6 months according to both (1) and (2), using the respective doses selected in step 2 and PK parameters chosen in step 1. Covariates and PK parameters of patients with different doses selected in (1) and (2) were further investigated. Modelling and simulation were performed using NONMEM (v.7.3), pre- and post-processing was conducted in R (v.3.4.4).
Results: In 76% of patients, the same doses were selected with either TA metric used, while in 24% of patients, different doses were selected (23.2% of gNM, 19.1% of gIM and 61.7% of gPM). For 21.9% of all gNM, a higher dose in the AAS vs. the ENDX group was selected, for 9.94% of all gIM, a higher dose, and for 9.21% of all gIM, a lower dose. For 61.5% of all PM, a lower dose in the AAS vs. the ENDX group was selected. Among patients with different dose selections, 98.9% of gNM, 92.2% of gIM and 88.3% of gPM reached at least one target in the AAS group, while 77.2% of gNM, 85.3% of gIM and 100% of gPM reached at least one target in the ENDX group. gNM with high apparent formation to and clearance of the ENDX-precursor metabolite N-desmethyltamoxifen (NDMT) were at highest risk for selection of a too low dose in ENDX-guided dosing.
Conclusions: In this simulation study, more than three out of four patients received the same dose regardless of the TA metric used. In patients with different dose selections, dose selection according to the AAS target seemed preferable for gNM and gIM, while dose selection according to the ENDX target was better for gPM. A vulnerable subpopulation with high apparent NDMT formation capacity and eliminating clearance was discovered and the clinical relevance of this finding should be further investigated.
References:
[1] Madlensky, L. Natarajan, S. Tchu, M. Pu, J. Mortimer, S.W. Flatt, D.M. Nikoloff, G. Hillman, M.R. Fontecha, H.J. Lawrence, B.A. Parker, A.H.B. Wu, J.P. Pierce. Tamoxifen metabolite concentrations, CYP2D6 genotype, and breast cancer outcomes. Clin. Pharmacol. Ther. 89: 718–725 (2011).
[2] A.H.M. de Vries Schultink, X. Alexi, E. van Werkhoven, L. Madlensky, L. Natarajan, S.W. Flatt, W. Zwart, S.C. Linn, B.A. Parker, A.H.B. Wu, J.P. Pierce, A.D.R. Huitema, J.H. Beijnen. An Antiestrogenic Activity Score for tamoxifen and its metabolites is associated with breast cancer outcome. Breast Cancer Res. Treat. 161: 567–574 (2017).
[3] A.H.M. de Vries Schultink, A.D.R. Huitema, J.H. Beijnen. Therapeutic Drug Monitoring of endoxifen as an alternative for CYP2D6 genotyping in individualizing tamoxifen therapy. Breast 42: 38–40 (2018).
[4] V.O. Dezentjé, F.L. Opdam, H. Gelderblom, J. Hartigh den, T. Van der Straaten, R. Vree, E. Maartense, C.H. Smorenburg, H. Putter, A.S. Dieudonné, P. Neven, C.J.H. Van de Velde, J.W.R. Nortier, H.J. Guchelaar. CYP2D6 genotype- and endoxifen-guided tamoxifen dose escalation increases endoxifen serum concentrations without increasing side effects. Breast Cancer Res. Treat. 153: 583–590 (2015).
[5] M.P. Goetz, K. Sangkuhl, H.J. Guchelaar, M. Schwab, M. Province, M. Whirl-Carrillo, W.F. Symmans, H.L. McLeod, M.J. Ratain, H. Zembutsu, A. Gaedigk, R.H. van Schaik, J.N. Ingle, K.E. Caudle, T.E. Klein. Clinical Pharmacogenetics Implementation Consortium (CPIC) Guideline for CYP2D6 and Tamoxifen Therapy. Clin. Pharmacol. Ther. 103: 770–777 (2018).
[6] L. Klopp-Schulze. Towards the individualisation of tamoxifen breast cancer treatment: Leveraging pharmacometric approaches. Freie Universitaet Berlin (2018).
[7] L. Klopp-Schulze, M. Joerger, S.L.W. Koolen, P. Neven, R.H.J. Mathijssen, C. Kloft.Exploring and explaining variability in tamoxifen and endoxifen pharmacokinetics in breast cancer patients: A pooled analysis. 26th Population Approach Group Europe (PAGE), Budapest, Hungary, 06-09 June 2017. PAGE 26: 7314 [www.page-meeting.org/?abstract=7314], (2017).
Reference: PAGE 28 (2019) Abstr 8900 [www.page-meeting.org/?abstract=8900]
Poster: Drug/Disease Modelling - Oncology