Exposure-Response Analysis Using Time to Event Data for Bevacizumab Biosimilar, SB8 and the Reference Bevacizumab
Suemin Park (1,2), Seung Chan Choi (1,2), Jin Ah Jung (3), Minjeong Park (3) Hyeong-Seok Lim (1,2)
(1) Department of Clinical Pharmacology and Therapeutics, Asan Medical Center, University of Ulsan College of Medicine, Seoul, Republic of Korea, (2) Department of Medical Science and Asan Medical Institute of Convergence Science and Technology, Asan Medical Center, University of Ulsan College of Medicine, Seoul, Republic of Korea, (3) Samsung Bioepis Co., Ltd., 76, Songdogyoyuk-Ro, Yeonsu-Gu, Incheon, 21987, Republic of Korea
Introduction:
SB8 is a biosimilar of the reference Avastin® (bevacizumab), a humanized anti-VEGF monoclonal antibody for treatment of solid tumors. SB8 is approved in the European Union based on demonstration of comparable physicochemical, biological, clinical pharmacokinetic (PK), efficacy and safety profiles as the reference bevacizumab [1]. As with most biologics in cancer treatment, interpatient variability may be observed in treatment with bevacizumab, which can be attributed to various factors including its PK. Several studies have been performed assessing correlation between bevacizumab exposure and survival in colorectal cancer [2, 3], but as of current, there is no known threshold for the minimum effective concentration of bevacizumab for treatment of non-small cell lung cancer (NSCLC). Thus, modeling analysis was conducted to characterize exposure-response relationship of SB8 and Avastin® in NSCLC patients and further evaluate impact of the PK profiles of SB8 and Avastin® on therapeutic efficacy.
Objective:
- To characterize exposure-response relationship of SB8 and Avastin®
- To evaluate therapeutic efficacy of SB8 and Avastin® based on exposure of the two treatments observed in the clinical Phase III study (SB8-G31-NSCLC)
Methods:
Progression-free survival (PFS) (central review [Avastin®, n=123; SB8, n=100] or investigator review [Avastin®, n=124; SB8, n=100]), and overall survival (OS) (Avastin®, n=124; SB8, n=100) data from advanced non-squamous NSCLC patients with available trough concentration (Ctrough) in the Phase III study (SB8-G31-NSCLC) were used to construct the time-to-event (TTE) model [4]. Various types of models were assessed using graphical and statistical methods. Base models without any covariates were selected first and potential covariates including treatments (SB8 vs. Avastin®), and trough concentration at steady state (Css, trough), were tested thereafter. For covariate assessment, proportional hazard covariate model was used for treatments and Css, trough, and inhibitory sigmoid maximum effect model was further implemented for Css, trough as more complex nonlinear relationship between hazard the Css,trough may be present [5,6]. Wald statistics test was used for model evaluations, and a likelihood ratio test was used to compare hierarchical models. The predictability of the TTE model was evaluated using visual predictive check plots and compared with the Kaplan-Meier curves. Median OS and PFS by Css, trough were simulated by implementing inverse functions for the TTE functions of the final model. Modeling and simulation were performed using NONMEM® v7.3 [7] and data were processed and plotted using R software v3.5.1 [8].
Results:
Log-logistic model with Css, trough as a covariate was selected as the final model for both OS and PFS. Treatments did not affect the model, indicating that the treatment effects of SB8 and Avastin® are equivalent. Css, trough significantly improved the model when a sigmoid inhibitory maximum effect model was implemented. Hill coefficients from 1 to100 were tested and 20 was selected, suggesting all-or-none effect of SB8 and Avastin® on OS and PFS. EC50 for OS, PFS (central), PFS (investigator) were estimated as 77.5, 77.7, and 72.9 mcg/mL, respectively. In simulations for median OS and PFS by Css, trough, predicted median survivals initially increased rapidly with increasing concentration, but after a certain value of Css, trough, median survivals remained at plateau. Css, trough of SB8 and Avastin® observed in the Phase III study were on plateaus of the exposure-response curve for both OS and PFS.
Conclusion:
Simulations based on the constructed TTE models for OS and PFS showed stiff exposure (Css, trough)-response relationship where the median OS and PFS increase sharply as Css, trough increases up to a certain value and remain on plateaus thereafter. Css, trough values of SB8 and Avastin® observed after administration of therapeutic doses of bevacizumab in the Phase III study were on plateaus of the exposure-response curves for both OS and PFS. These findings show that therapeutic effects of SB8 and Avastin® are similar even after accounting for the interpatient variability that might arise in treatment NSCLC. Thus, these modeling and simulation demonstrated comparable efficacy of SB8 and Avastin® based on the exposure of the two treatments.
References:
[1] European Medicines Agency. Aybintio: Public assessment report https://www.ema.europa.eu/en/documents/assessment-report/aybintio-epar-public-assessment-report_en.pdf
[2] Papachristos A, Kemos P, Kalofonos H, et al., Correlation Between Bevacizumab Exposure and Survival in Patients with Metastatic Colorectal Cancer. Oncologist. 2020 Oct; 25(10): 853–858. Published online 2020 Apr 28.
[3] Caulet M, Lecomte T, Bouche O et al., Bevacizumab Pharmacokinetics Influence Overall and Progression-Free Survival in Metastatic Colorectal Cancer Patients. Clin Pharmacokinet. 2016 Nov;55(11):1381-1394.
[4] ClincalTrial.gov. A Study Comparing SB8 and Avastin® in Patients with Advanced Non-squamous Non-small Cell Lung Cancer NCT02754882. https://clinicaltrials.gov/ct2/show/NCT02754882
[5] HS Lim. Brief introduction to parametric time to event model. Trans Clin Pharmacol. 2021 Mar;29(1):1-5.
[6] Chigutsa E, Long AJ1, Wallin JE. Exposure-Response Analysis of Necitumumab Efficacy in Squamous Non-Small Cell Lung Cancer Patients. CPT Pharmacometrics Syst Pharmacol. 2017 Aug;6(8):560-568.
[7] Robert J, Bauer. NONMEM Tutorial Part I: Description of Commands and Options, With Simple Examples of Population Analysis. CPT Pharmacometrics Syst. Pharmacol. (2019) 8, 525–537.
[8] R Development Core Team. R: A language and environment for statistical computing. Vienna, Austria, 2018.