L.T. van der Heijden (1), M. van Nuland (1), T.P.C. Dorlo (1), J.H. Beijnen (1,2), A.D.R. Huitema (1,3,4).
(1) The Netherlands Cancer Institute, the Netherlands, (2) Utrecht University, The Netherlands, (3) University Medical Center Utrecht, Utrecht University, The Netherlands, (4) Princess Máxima Center Utrecht, The Netherlands.
Objectives: Microdosing is a method to predict pharmacokinetics of a therapeutic dose from microdose (max. 100 µg) data (1). The conventional evaluation method of microdosing trials is the two-fold criterion. This criterion states that dose-normalised pharmacokinetic metrics of the microdose and therapeutic dose fall within two-fold of one another (2, 3). The major disadvantage of the two-fold criterion is its inability to simulate different therapeutic doses for Phase I studies. An alternative method to evaluate microdose scalability is a naïve pooled data (NPD) modelling approach. With this approach pharmacokinetic models based on microdose data are used to extrapolate and compare different therapeutic dosing regimens. This allows informed decision making about the dose regimen to be used in future Phase I studies. The objective was to demonstrate the feasibility of a NPD modelling approach to evaluate microdose scalability.
Methods: Gemcitabine and anastrozole were used as model drugs for intravenous and oral administration, respectively. Data from two microdose studies and one clinical dose finding study were used (4-6). Plasma concentrations of gemcitabine, its metabolite 2’,2’-difluorodeoxyuridine (dFdU) and anastrozole were analysed using NONMEM (version 7.3).
Model development of gemcitabine and dFdU was performed sequentially. The Laplacian estimation method and subroutine ADVAN13 (TOL = 9) were used. Separate models described the residuals errors for gemcitabine observations above the lower limit of quantification (LLOQ), between the LLOQ and limit of detection (LOD) and dFdU observations. Beal’s M3 method was used for observations
First-order conditional estimation method with interaction (FOCE-I) with subroutine ADVAN4 TRANS4 was used for anastrozole. First and zero order absorption were tested to describe absorption. Residuals errors were described by a proportional error model.
The final microdose models were used to simulate therapeutic dosing regimens of gemcitabine (1,250 mg/m2) and anastrozole (1 mg). A visual predictive check (VPC) was performed by simulating the therapeutic trials with 10 and 6 patients, respectively, 1000 times using the final microdose models. The 2.5th and 97.5th percentiles of the simulated therapeutic plasma concentrations were compared against the observed therapeutic concentrations.
Results A total of 99 gemcitabine (n=9) and 88 dFdU (n=8) plasma concentrations were included in the final dataset of which 7 (7.1%) were below the limit of quantification. The data was best described by a 2-compartment model for gemcitabine and 1-compartment model for dFdC with a first-order conversion constant from gemcitabine to dFdU and a first order elimination for dFdU.
A total of 66 anastrozole plasma concentrations (n=6) were included in the data analysis. A two-compartment model with linear elimination best described the data. Absorption was best described with a zero-order absorption with a duration of 1 h (time to maximum concentration).
Extrapolation of a microdose of gemcitabine showed good scalability of the initial elimination phase with 83.3-100% of the therapeutic observations inside the 2.5-97.5th percentiles between 9.4 and 75.5 minutes after infusion. The terminal elimination phase was overpredicted by the microdose model (50-100% of the therapeutic observations was below the 2.5-97.5th percentiles between 75.5 minutes and 10 h after infusion). Furthermore, microdose dFdU pharmacokinetics was not scalable to therapeutic observations due to a different trend in pharmacokinetics. Anastrozole microdose pharmacokinetics was well extrapolated to therapeutic pharmacokinetics with 55% and 82% of the reported mean concentrations and standard deviations inside the 2.5-97.5th percentiles.
Conclusion A NPD modelling approach is a feasible method for the evaluation of Phase 0 microdose studies. This approach can identify non-linear pharmacokinetics that might cause discrepancies between microdose and therapeutic pharmacokinetics. More importantly this approach allows the prediction of different therapeutic dosing regimens. The expected pharmacokinetics of these different dosing regimens can be compared. This comparison can be used for informed decision making about the continuation of drug development and which dose to use in Phase I clinical trials.
References:
[1] Garner RC, Lappin G. The phase 0 microdosing concept. Br J Clin Pharmacol. 2006;61(4):367-70.
[2] Lappin G, Garner RC. The utility of microdosing over the past 5 years. Expert Opin Drug Metab Toxicol. 2008;4(12):1499-506.
[3] Rowland M. Commentary on ACCP position statement on the use of microdosing in the drug development process. J Clin Pharmacol. 2007;47(12):1595-6; author reply 7-8.
[4] Van Nuland M, Rosing H, Thijssen B, Burgers JA, Huitema ADR, Marchetti S, et al. Pilot Study to Predict Pharmacokinetics of a Therapeutic Gemcitabine Dose From a Microdose. Clin Pharmacol Drug Dev. 2020.
[5] Nomura Y, Koyama H, Ohashi Y, Watanabe H. Clinical Dosage Determination of a New Aromatase Inhibitor, Anastrozole, in Postmenopausal Japanese Women with Advanced Breast Cancer. Clinical Drug Investigation. 2000;20(5):357-69.
[6] Kusuhara H, Takashima T, Fujii H, Takashima T, Tanaka M, Ishii A, et al. Comparison of pharmacokinetics of newly discovered aromatase inhibitors by a cassette microdosing approach in healthy Japanese subjects. Drug Metab Pharmacokinet. 2017;32(6):293-300.
[7] Beal SL. Ways to fit a PK model with some data below the quantification limit. J Pharmacokinet Pharmacodyn. 2001;28(5):481-504.
Reference: PAGE 29 (2021) Abstr 9753 [www.page-meeting.org/?abstract=9753]
Poster: Methodology - Other topics