Suemin Park1, Seung Chan Choi1,2, Nari Yun3, Myoung Ho Jang3, Hyeong-Seok Lim1,2
1Department of Clinical Pharmacology and Therapeutics, Asan Medical Center, University of Ulsan College of Medicine, 2Department of Medical Science and Asan Medical Institute of Convergence Science and Technology, Asan Medical Center, University of Ulsan College of Medicine, 3Clinical Development, GI Innovation, Inc.
Introduction/Objectives: GI-102 (CD80/IL2v3) is a novel immunocytokine designed to direct interleukin-2 variant (IL-2v3) to immune cells and tumor cells. The proliferation and activation of natural killer (NK) and cytotoxic T (Tc) cells are enhanced through selective binding affinity to IL-2Rß? and cis-binding of CD80 to PD-L1, and the effects of IL-2 on regulatory T cells are minimized. A phase 1/2a, first-in-human (FIT) study is underway to determine the optimal dose(s) in patients with advanced or metastatic solid tumors, and to assess anti-tumor activity of the drug along with its safety and tolerability profile [1]. Based on the interim data, a cell-level pharmacodynamics-mediated drug disposition (PDMDD) model of GI-102 was constructed, and simulations were conducted (1) to assess the relationship between drug exposure, target receptor occupancy, and pharmacodynamic (PD) effect and (2) to provide supportive information for determining the optimal dose range. Methods: The model was constructed by simultaneously fitting the available drug concentrations data (number of observations, n=966) along with total lymphocyte (n=620), Tc cell (n=368) and NK cell (n=368) counts data from a dose escalation phase of the FIH study that included 41 patients administered with GI-102 via intravenous infusion every three weeks, at doses ranging from 0.06 to 0.45 mg/kg. The dose escalation phase of the study was ongoing at the time of analysis (as of Sep 30, 2024). The PK model was based on the target-mediated drug disposition principle where GI-102 binds to IL-2Rß? and undergoes elimination. The model incorporated time-varying PD-mediated disposition, as the increased proliferation of lymphocytes stimulated by the drug-receptor complexes affects the target receptors pool, subsequently influencing drug distribution and elimination. Additionally, a cell-level component was incorporated into the model by describing turnover and binding of the IL-2Rß? to the drug on a single cell [2]. In line with the proposed mechanism of action, it was assumed in the model that the increase in total lymphocytes with GI-102 treatment is attributed to the sum of increases in three simplified subsets: Tc cells, NK cells, and other cells. The margination and diapedesis of lymphocytes into tissues immediately following treatment with GI-102 were described empirically [3]. The time delay between the drug concentrations and the increase in cell counts, driven by the stimulation of cell production by GI-102, was described by indirect response model combined with an operational model of agonism [4,5]. A mixture model was applied to account for the bimodal distribution of the observed cell counts among patients [6]. An interim PK-PD model was constructed mainly based on graphical assessments due to the model’s complexity and instability. Simulations were then performed to assess the relationship between drug exposure, target receptor occupancy, and increase in lymphoid cell counts from baseline. Modeling was conducted using NONMEM® version 7.5.0, and data processing and plotting were performed with R version 4.3.3 [7,8]. Results: The two-compartment, cell-level PDMDD PK-PD model was constructed. The model was deemed adequate for predicting central tendency of the PK/PD data, but may slightly underpredict the lymphocyte and Tc cell. Simulations showed a clear exposure-response relationship across the evaluated dose range of 0 through 0.45 mg/kg. Increase in lymphoid cell counts exhibited a steep rise at lower doses followed by a gradual, dose-dependent increase at higher doses. At 0.45 mg/kg, total lymphocyte counts and Tc cells accounting for diapedesis were predicted to reach a mean level of about 3000 cells/µL and 424 cells/µL, respectively, indicating robust expansion of these cells compared to baseline. NK cells accounting for diapedesis were predicted to reach a mean level of 926 cell/µL. Proliferation of Tc and NK cells were observed at all tested doses with less than 20% IL-2Rß? occupancy. The higher dose range (e.g., 0.24-0.45 mg/kg) tested in the clinical study showed a significant increase in lymphocytes. Conclusion: Our modeling and simulation based on the interim clinical data provides a preliminary prediction of the drug’s PD effect at the tested doses, serving as a reference for determining the optimal dose range.
[1] Lee J, Cho BC, Lee JL et al. A first-in-human, phase 1/2a study of GI-102 (CD80-IL2v3) in patients with advanced or metastatic solid tumors: initial results from dose escalation. Journal of Clinical Oncology 2024 May, 42:16_suppl, 2598-2598 [2] Krzyzanski Wojciech, Harrold John M, Wu Liviawati S. et al. A cell-level model of pharmacodynamics-mediated drug disposition. J Pharmacokinet Pharmacodyn. 2016 Sep, 43:513-527. [3] Dutcher Janice P, Schwartzentruber Douglas J, Kaufman Howard L et al. High dose interleukin-2 (Aldesleukin) – expert consensus on best management practices. Journal for Immuno Therapy of Cancer 2014 Sep, 2:26. [4] Zou Huixi, Banerjee Parikshit, Leung Sharon Shui Yee et al. Application of pharmacokinetic-pharmacodynamic modeling in drug delivery: development and challenges. Front. Pharmacol. 2020 Jul, 11:997 [5] Black JW and Leff P. Operational models of pharmacological agonism. Proceedings of the Royal Society of London. Series B, Biological Sciences. 1983 Dec, 220(1219): 141-162. [6] Yoon S and Lim H. Performance of a mixture model by the degree of a missing categorical covariate when estimating clearance in NONMEM. Trans Clin Pharmacol. 2019 Dec, 27(4):141-148. [7] Beal SL, Sheiner LB, Boeckmann AJ, and Bauer RJ (eds) NONMEM 7.5 Users Guides. (1989–2020). ICON plc, Gaithersburg, MD. https://nonmem.iconplc.com/nonmem750 [8] R Development Core Team. R: A language and environment for statistical computing. In, Vienna, Austria, 2018.
Reference: PAGE 33 (2025) Abstr 11616 [www.page-meeting.org/?abstract=11616]
Poster: Drug/Disease Modelling - Oncology