Population pharmacokinetic analysis of pazopanib in adult patients with metastatic renal cell carcinoma and soft tissue sarcoma
Zhiyuan Tan (1), Swantje Völler (1,2), Anyue Yin (3), Amy Rieborn(3,4), Max Kramer (3), Mike Volwater (3), Hans Gelderblom(4), Tom van der Hulle (4), Catherijne Knibbe (1,5), Dirk Jan Moes (3)
(1) Division of Systems Pharmacology and Pharmacy, Leiden Academic Centre for Drug Research, Leiden University, Leiden, The Netherlands. (2) Pharmacy, Leiden Academic Centre for Drug Research, Leiden University, Leiden, The Netherlands. (3) Department of Clinical Pharmacy & Toxicology, Leiden University Medical Center, Leiden, The Netherlands. (4) Department of Medical Oncology, Leiden University Medical Center, Leiden, The Netherlands. (5) Department of Clinical Pharmacy, St Antonius Hospital, Nieuwegein, The Netherlands.
Introduction: Pazopanib is a multi-target receptor tyrosine kinase inhibitor (TKI), and part of the current standard of care of metastatic renal cell carcinoma (mRCC) and soft tissue sarcoma (STS) at an approved 800 mg oral daily dose[1, 2]. Previous studies have shown a trough concentration (Cmin) of > 20.5 mg/L is correlated with improved progression-free survival (PFS)[3] in mRCC patients, which is the basis for implementing pazopanib therapeutic drug monitoring (TDM). However, the exposure-safety relationship in mRCC patients was reported to vary in different studies[4]. Also, the correlation of exposure with survival/safety for STS patients has not been established separately. A recent study[5] postulated Cmin < 27 mg/L was independently associated with a risk of progression at 3 months in STS patients, which might indicated a different TDM target. Large inter-individual variability (IIV) has been observed for pazopanib pharmacokinetics (PK)[6].
To date, several pazopanib population PK (PopPK) models have been published aiming to describe exposure in real-world patients, mainly mRCC and few on STS[6, 7]. In this study, an integrated PK analysis based on TDM data of both cancer types from routine clinical care has been performed to get better understanding of the real-world patients with mRCC and STS.
Objectives: Develop a pazopanib PK model that can accurately predict individual PK in real-world patients for future exploration of differences in exposure-response between mRCC and STS patients.
Methods: Pazopanib TDM data from routine clinical practice was identified from mRCC and STS patients at the Leiden University Medical Center (LUMC) between February 2014 and July 2022.
One and two compartmental models with linear oral absorption were explored. Due to known dose-dependent non-linear absorption properties of pazopanib tablets, both linear absorption and a published dual absorption rate (Ka) model with dose and time dependent bioavailability (F) were tested (Yu et al [6]). In addition, incorporating non-linearity oral properties (Ka, F) and non-linear apparent clearance (CL/F) was also evaluated. Furthermore, the influence of disease types on the CL/F of pazopanib was investigated.
Data processing and PK analysis were performed using R (V 4.2.1) and NONMEM (V 7.4.4). Object function values (OFV) and goodness-of-fit (GOF) plots were used to evaluate the fit of the studied models. GOF plots were stratified on tumor types to investigate potential difference between mRCC and STS patients. Bootstrap and prediction corrected visual predicted check (pcVPC) were used for model validation.
Results: In the study, retrospective data from 135 patients with 460 TDM observations (median age 65 (IQR 58-74.5) and 63% male) were included. 96 patients (71%) were diagnosed with mRCC. The patients were treated for a median duration of 122 days with a median dose of 600 mg. TDM observations mainly comprised trough samples (87.5%). Mean time after last dose was 22.0 ± 13.2 h and mean concentration was 34.70 ± 15.70 mg/L.
One-compartment model with linear absorption was chosen as the base model. Applying the model of Yu et al [6] showed underprediction at higher dose and overprediction at lower dose. The estimation of absorption-related parameters of the PK model of Yu et al was also not achievable. Additionally, including non-linearity on F, Ka and exponential dose-effect on CL/F did not improve the fit to the data compared to the linear absorption model (dOFV all > 0). Therefore, the linear absorption model was adopted and Ka was fixed to a reported value[7]. Moreover, inter-occasion variability (IOV) on CL/F defined by each sample per individual was introduced which improved individual prediction significantly. Finally, a one-compartment disposition model with linear absorption and IOV on CL/F was chosen as final model. Covariates were not added at current stage and will be considered later.
CL/F was estimated to be 0.793 L/h (RSE 4%), and the CV% of IIV on CL was 23.9% (RSE 54%). IOV on CL/F was estimated to be 21.6%. Comparable CL/F estimates in mRCC (0.79 L/h) and STS (0.83 L/h) patients were identified. No obvious difference was observed on GOF plots split by tumor types as well. Model validation showed stable estimation.
Conclusions: A PopPK model of pazopanib integrating both mRCC and STS patients, with retrospective TDM data was developed for future PK/PD analysis to evaluate differences between mRCC and STS patients.
References:
[1] STERNBERG C N et al. Journal of clinical oncology (2010) 6, 1061-8.
[2] VAN DER GRAAF W T et al. Lancet (2012) 9829, 1879-86.
[3] VERHEIJEN R B et al. Clinical pharmacokinetics (2017) 9, 987-97.
[4] WESTERDIJK K et al. British journal of clinical pharmacology (2020) 2, 258-73.
[5] MINOT-THIS M S et al. Pharmaceutics (2022) 6, 1224.
[6] YU H et al. Clinical pharmacokinetics (2017) 3, 293-303.
[7] OZBEY A C et al. Pharmaceuticals (2021) 9, 927.