Rubin Lubomirov, Laura Pérez-Ramos, Eduardo AsÃn, Salvador Fudio
Department of Clinical Pharmacology, Clinical Development / Oncology Business Unit, PharmaMar S.A., Colmenar Viejo (Madrid), Spain
Introduction: Lurbinectedin (Zepzelca®) is a selective inhibitor of oncogenic transcription, and has a direct effect on the tumor microenvironment by modifying immune-regulatory properties of tumor associated macrophages [1-4]. Lurbinectedin is highly protein-bound, with higher affinity for α-1-acid glycoprotein (AAG) than for albumin, is mainly metabolized by CYP3A4/5 [5], and is a substrate of ABCB1 transporter. On June 2020, lurbinectedin was granted FDA’s accelerated approval as second-line treatment for patients with small cell lung cancer (SCLC) [6]. Up to now, the impact of genetic factors on lurbinectedin pharmacokinetic variability have not been systematically investigated.
Objectives: To develop a population pharmacokinetic (PopPK)-based, two-stage (discovery & validation) approach to assess the impact of genetic variants on lurbinectedin exposure in patients with advanced cancer.
Methods: Total plasma concentrations of lurbinectedin from patients with advanced cancer participating in Phase I to III clinical trials were pooled and fitted using non-linear mixed-effects modelling implemented in NONMEM v7.3 [7]. In the discovery stage, as previously described [8], lurbinectedin interpatient variability on total clearance (etaCL) was used as a phenotype, considering that this parameter would best reflect the remaining unexplained variance in lurbinectedin elimination accounted for by genetic variations. Subpopulations of low (n=10) and high (n=10) etaCL outliers (i.e. patients with high and low lurbinectedin plasma exposure, respectively) from a subset of 180 patients, were selected as cases and controls, respectively. Germline DNA was extracted from peripheral blood mononuclear cells. Targeted exome sequencing of 42 genes involved in lurbinectedin metabolism and transport was performed in an Ion PGM™ System for Next-Generation Sequencing (NGS) using an Ion 318™ chip v2 for every ten samples. Associations between etaCL and genetic variants were assessed by comparing allelic frequencies between cases and controls using PLINK 2.0 [9]. In the validation stage, top rated genetic variants (with lowest p-value) identified in the discovery stage were genotyped using TaqMan® assays in an independent population of patients (n=719). In this stage, the PK parameter used as phenotype was unbound lurbinectedin clearance (CLu), estimated based on empirical Bayesian estimates of individual total CL from the PopPK model and plasma proteins levels (albumin and AAG as well as respective lurbinectedin dissociation constants Kd Albumin and Kd AAG), as described previously [10]. Impact of each genetic variant on individual CLu was assessed using PLINK 2.0 [9], assuming different genetic models (log-additive, recessive or dominant), with adjustments for multiple comparisons. Quantification of the magnitude of differences in CLu among genotypes was assessed using the ratios of log-transformed geometric means with 90% confidence intervals (90% CI).
Results: A three-compartment mammillary model with linear distribution and elimination from central compartment was suitable to describe the time course of total plasma concentrations of lurbinectedin after i.v. administration. The model estimated a volume of distribution at steady state of 460 L that caused the distribution to deep tissues and a low central volume of 11.5 L while total plasma CL was 9.65 L/h. The model detected several covariates that affected CL: serum AAG, the presence of strong or moderate CYP3A4 inhibitors, and BSA. In the discovery stage, 34 genetic variants located in 20 genes were identified (p-value <0.05) [8]. During the validation stage, nine genetic polymorphisms were genotyped in 719 patients. The association with lurbinectedin exposure was confirmed for rs15524 (A>G) in CYP3A5 gene. Carriers of G allele (18% of patients) had 25% (90%CI: 16 – 34%) higher lurbinectedin CLu compared to patients homozygous for allele A.
Conclusion: The applied PopPK-based two-stage pharmacogenetic assessment approach was a valid strategy to identify and evaluate the impact of genetic variants on lurbinectedin exposure along the development program. The clinical impact and applicability of these findings need further prospective evaluation and quantification in an independent population of patients treated with lurbinectedin.
References:
[1] Leal et al. Br J Pharmacol. 2010; 161:1099. doi: 10.1111/j.1476-5381.2010.00945.x
[2] Santamaría Nuñez et al. Mol Cancer Ther. 2016; 15:2399. doi: 10.1158/1535-7163.MCT-16-0172
[3] Costanzo et al. EMBO Mol Med. 2022; e14841. doi: 10.15252/emmm.202114841
[4] Allavena et al. Front Oncol. 2022; 12:851790. doi: 10.3389/fonc.2022.851790
[5] Aviles et al. Drug Metab Dispos. 2022; 50:327. doi: 10.1124/dmd.121.000668
[6] Singh et al. Clin Cancer Res. 2021; 27:2378. doi: 10.1158/1078-0432.CCR-20-3901
[7] Beal SL, Sheiner LB, Boeckmann AJ & Bauer RJ (Eds.) NONMEM Users Guides. 1989-2017. Icon Development Solutions, Ellicott City, Maryland, USA.
[8] Lubomirov et al. PAGE 27 (2018) Abstract 8489 [https://www.page-meeting.org/ default.asp?abstract=8489]
[9] Purcell et al. Am J Hum Genet 2007; 81:559. doi: 10.1086/519795
[10] Fernández-Teruel et al. Cancer Chemother Pharmacol. 2021 Nov 5. doi: 10.1007/s00280-021-04366-3
Reference: PAGE 30 (2022) Abstr 10039 [www.page-meeting.org/?abstract=10039]
Poster: Drug/Disease Modelling - Oncology