Zhiyuan Tan (1), Swantje Völler (1,2), Anyue Yin (3), Amy Rieborn(3,4), A. J. Gelderblom(4), Tom van der Hulle (4), Catherijne A. J. Knibbe (1,5), Dirk Jan A. R. 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 currently used for treatment of metastatic renal cell carcinoma (mRCC) and soft tissue sarcoma (STS) at an approved fixed daily dose (QD) of 800 mg[1, 2]. In clinical practice, however, real-world patients are less able to tolerate the 800 mg QD dose with approximately 60% of patients requiring a dose reduction[3]. Severe liver toxicity necessitates treatment interruption in more than 10% of patients, which poses a risk of disease progression[4]. Even though a trough concentration (Cmin) efficacy target of ≥ 20.5 mg/L was reported in several mRCC studies[5], there is no such target specified specifically for STS. Furthermore, there is only limited evidence regarding the optimal toxicity threshold for which 46 to 50 mg/L has been suggested[5]. It is therefore of interest to refine the dose and exposure target in real-world setting.
Objectives: (1) To develop an exposure-liver toxicity time-to-event (TTE) model and identify predictive factors; (2) To develop a tumor size dynamics model and explore and compare exposure-tumor growth relationship of pazopanib in mRCC and STS patients.
Methods: Clinical data from mRCC and STS patients treated with pazopanib at the Leiden University Medical Center (LUMC) between February 2014 and July 2022 was retrieved. Pazopanib exposure metrics for the patients were derived from a previously established population pharmacokinetic (PPK) model[6], based on therapeutic drug monitoring (TDM) data and dose information from the same population.
Toxicity was defined as Common Terminology Criteria for Adverse Events (CTCAE V4.03) Grade ≥ 2 liver toxicity. Different base hazard functions were evaluated and subsequently covariates analysis including dose, exposure metrics, demographics and tumor type was performed.
Tumor size was defined as the sum of longest tumor diameter (SLD) of target lesions, using RECIST 1.1 criteria[7]. The modeling strategy was: (1) Determine a semi-mechanistic model with apparent growth rate (aKG), apparent killing rate (aKD) and acquired resistance (AR); (2) Implement a mixture model to identify patients with pre-existing resistance; and (3) Investigate effect of exposure on tumor growth.
Results: Retrospective data from 135 patients (median age 65 (IQR 58-74.5), median dose of 600 (IQR 200-800) mg, and 63% male, 71% diagnosed with mRCC) was available for the TTE liver toxicity analysis. The data was best described by a Gompertz hazard function with Cmin significantly influencing the hazard (dOFV=8.03). The hazard of developing liver toxicity did not differ between mRCC and STS. A Cmin of ≥ 34 mg/L was found as liver toxicity threshold, resulting in a 4.85-fold hazard increase (P <0.01) compared to Cmin < 34 mg/L. Model-based simulations showed that 600 mg QD significantly decreases the hazard of liver toxicity compared with 800 mg QD (P<0.001) while in 70% of the population Cmin remains ≥ 20.5 mg/L.
For the tumor growth model, 111 (63% male and 72% mRCC) out of 135 patients, with 111 baseline and 403 post-treatment SLD, were included. Median Cmin was 26.6 [IQR 20.6-31.1] mg/L. Median baseline SLD was 79 [IQR 50-129] mm for mRCC and 88 [IQR 64-148] mm for STS. Median SLD after treatment start was 53 [IQR 33-103] mm for mRCC and 112 [IQR 61, 160] mm for STS. For tumor growth rate, first-order aKG (day-1) was estimated to be 0.0005 for mRCC and 0.0086 for STS. For killing effect, linear aKD (day-1) was estimated to be 0.004 for mRCC and 0.008 for STS, while neither exposure nor dose dependency was found for both types of tumors. The introduction of pre-existing resistance by using a mixture model, improved the model fit significantly (27% and 13% for mRCC and STS, dOFV > 3.84). Resistance rate, AR (day-1) was described by an exponential function with time dependency (0.008 for mRCC and 0.0003 for STS). Relative standard error (RSE) for all estimates were ≤30% except AR of STS (59%).
Conclusions: The TTE model indicates that a Cmin ≥ 34 mg/L significantly increases the risk of liver toxicity. In contrast, no clear effect of exposure was observed in the tumor growth model for mRCC and STS within our cohort with a median Cmin of 26.6 mg/L and IQR 20.6-31.1 mg/L. A decreased pazopanib initial dose of 600 mg QD, followed by routine TDM practice once 2-3 months aiming at a Cmin target of 20.5-34 mg/L with the developed PPK model[6] could potentially improve the balance between efficacy and treatment consistency.
References:
[1] MOTZER R J et al. The New England journal of medicine (2013) 369(8), 722-731
[2] VAN DER GRAAF W T et al. Lancet (2012) 379(9829), 1879-1886
[3] JOHNSTON H et al. Clinical genitourinary cancer (2023) 21(3), 357-365
[4] KAPADIA S et al. Acta oncologica (2013) 52(6), 1202-1012
[5] YU H et al. Clinical pharmacokinetics (2014) 53(4), 305-325
[6] Tan Z et al. PAGE 31 (2023) Abstr 10553 [www.page-meeting.org/?abstract=10553]
[7] SCHWARTZ L H et al. European journal of cancer (2016) 62, 132-137
Reference: PAGE 32 (2024) Abstr 11107 [www.page-meeting.org/?abstract=11107]
Poster: Drug/Disease Modelling - Oncology