Zinnia P Parra-Guillen(1), Eduardo Asin-Prieto(1), MarÃa J Garrido(1), Iñaki F Trocóniz(1), Youfang Cao(2), Kapil Mayawala(2), Dinesh de Alwis(2), Tomoko Freshwater(2)
(1)Pharmacometrics & Systems Pharmacology, Department of Pharmaceutical Technology and Chemistry, School of Pharmacy and Nutrition, University of Navarra, Pamplona, Spain 2Merck & Co., Inc., Quantitative Pharmacology & Pharmacometrics, Department of Pharmacokinetics, Pharmacodynamics, and Drug Metabolism, 2000 Galloping Hill Road, Kenilworth, NJ, USA
Introduction/Objectives: Oncolytic viruses (OVs) offer a promising immunotherapeutic modality for cancer treatment. These viruses are characterized by their capability to directly infect and lyse tumor cells, thus activating an antitumoral immune response [1]. Different mathematical models have been developed to model in-vitro kinetics of OV leaving a gap for their utility for clinical data. Furthermore, these models have either ignored or inadequately represented viral kinetics and tumor distribution – limiting their applicability in drug development.
The objective of this work was to develop a quantitative systems pharmacology (QSP) model linking viral kinetics/replication and distribution to tumor, and to explore model behavior under different physiologically relevant parameters. Such a model can be used to inform intratumoral (IT) as well as intravascular (IV) administration of OV in the clinic.
Methods: A QSP model accounting for biological processes, including viral clearance, distribution and uptake in tumor tissues, viral infection and virus replication in the tumor was developed. Specifically, the model consisted of 3 compartments, namely central, peripheral and tumor. Kinetic parameters were fitted to mean viral profiles extracted from literature after IV doses ranging from 1010 to 1013 copies [2, 3]. Tumor compartment was further divided into two sub-compartments representing (1) vasculature compartment of injected and non-injected lesions, and (2) cellular compartment which accounts for the virus within the infected tumor cells from injected and non-injected lesions. A viral replication model was incorporated at tumor level using classical structures [4]: infectivity was proportional to the viral levels at the vascular tumor compartment and the number of non-infected cells. The number of initial tumor cells was derived using clinical data on tumor lesion sizes [5]. Infected cells could in turn produce more virus, but also die at rates that are faster than non-infected (replicating) tumor cells accounting for immune and viral responses. The range of physiologically relevant parameter values for viral tumor distribution [6], as well as viral dymanics (infectivity, replication and induced cell death) [4-5, 7-8] were obtained from the literature. Simulations (using Berkeley Madonna 9.1.18) were performed to assess how the model behaves over various combinations of parameter values.
Results: From the viral kinetics model, a clearance of 41 L/d and volume at steady-state of 84 L were obtained. Simulations including viral distribution and dynamics illustrate 3 behaviors of the model: 1) In the absence of replication, infected tumor cells can be up to 3 orders of magnitude higher after IT than IV. However, with replication rates above 10 copies/cell/d, comparable levels of infectivity can be obtained, even for situations of low viral tumor distribution (< 2L/h) and or affinity. In this context, the predicted virus levels can trigger tumor response under high infectivity conditions (10–8 1/virions/d). 2) Highly oncolytic viruses (i.e. virus triggering a fast death of infected cells, >1d-1) can limit viral replication, thus leading to lower virus levels in both serum and tumor which potentially hampers tumor response. 3) Different magnitudes of tumor response can be observed even with similar serum levels e.g. low replication (<10 copies/cell/d) and high infectivity. These simulation results showed a balance among exposure, infectivity and replication that can be explored through simulations in order to guide clinical study design for optimal dose ranges and sampling time points.
Furthermore, the model also identifies sensitive biological processes, e.g., replication rate and infectivity ratio that cannot be uniquely parameterized using only the serum viral load data from the clinic. For such parameters, in-vitro experiments were suggested.
Conclusions: A general QSP model is proposed to characterize key disposition mechanisms of oncolytic virus. This framework enabled in silico exploration of the impact of different dosing strategies and potential sets of parameters on viral levels in serum and tumor, thus representing a suitable tool to support clinical development of novel OVs. In future, the link between viral exposure and clinical efficacy will be established with emerging data from ongoing clinical trials.
References:
[1] Kaufman et al. Nat Rev Drug Discov. 14:642-662 (2015)
[2] Machiels et al, J Immunother Cancer 7:20 (2019)
[3] García-Carbonero et al. J Immunother Cancer 5:71-85 (2017)
[4] Titze et al. Eur J Pharm Sci. 97:38-46 (2017)
[5] Malinzi et al. Math Biosci Eng. 15: 1435-1463 (2018)
[6] Makkat et al. J Magn Reson Imaging. 25:1159-1167 (2007)
[7] Bajzer et al. J Theor Biol. 252:109-122 (2008)
[8] Cao et al. PLoS Pathog. 14: e1007350 (2018)
Reference: PAGE () Abstr 9282 [www.page-meeting.org/?abstract=9282]
Poster: Oral: Drug/Disease Modelling