Chenyu Wang1, Dr. Qinhao Wu1, Dr. Xuanlin Liu1, Dr. Lalitya Sudarsono1, Anne-Grete Märtson1, J. G. Coen van Hasselt1
1Leiden University
Introduction There exists an urgent need for new antiviral therapies against emerging viruses[1]. The empirically derived half-maximal effective concentration (EC50) which suppresses viral activity in in vitro studies represents a key parameter in antiviral drug development. However, differences in experimental protocols have been suggested to substantially impact estimates of this antiviral EC50 parameters [2]. These inconsistencies complicate the standardization of antiviral efficacy assessments [3]. Secondly, beyond experimental conditions, virus-specific properties can also impact EC50 estimation. Viruses with high replication rates or faster clearance may exhibit different dose-response relationships, influencing the apparent EC50 measured in vitro. Understanding these biological differences is critical for interpreting EC50 values across different viruses and ensuring accurate comparisons of antiviral efficacy. To address these challenges in translational antiviral drug development, mechanism-based viral dynamic modeling provides a quantitative framework to disentangle the effects of experimental conditions and virus-specific properties on EC50. In this study, we investigate how empirical EC50 values as determined from experimental protocols are influenced by experimental conditions and virus-specific properties. We use a model-based approach to systematically analyze these factors and assess their impact on EC50 estimation. Specifically, we aimed to (1) determine relationships between empirical EC50 (experimental) and mechanism-based model parameters including the intrinsic EC50 (in model); (2) explore how experimental conditions affect the EC50 calculation in vitro. Methods Model definition: The target-cell limited (TCL) viral dynamic model was used to simulate viral burden dynamics over time across multiple experimental scenarios. Our standard model is based on a SARS-CoV-2 viral dynamics model derived from laboratory data, with initial simulation conditions carefully set to match those of the experiment. The equations of TCL model are: dT/dt=-ßVT dI/dt=ßVT-dI dV/dt=(1-E)pI-cV Here, T represents target cell count; ß represents target cell infection rate; V represents viral load; I is the infected cell count; d is the infected cell clearance rate; E is the drug effect defined according to a sigmoidal Emax model; p is viral production rate; and c is viral clearance rate. Simulation settings: Our simulations incorporated virus-specific model parameters, including ß, p, c and d, to capture key aspects of viral dynamics. The parameter range was determined based on differences among viruses, with a maximum variation of 100-fold. Experimental conditions such as initial cell numbers (V0), multiplicity of infection (MOI), time of adding drugs, sampling time and drug concentration were systematically varied to assess their impact on EC50 estimation. The range of simulated experimental conditions was based on real experiments, also with a maximum variation of 100-fold. Empirical EC50 estimation: The empirical EC50 was estimated at a selected time point by fitting a four-parameter logistic model to the terminal simulated viral load. To assess the relationship between EC50 and key variants, including experimental conditions and viral parameters, Kendall’s tau correlation test was performed to evaluate the statistical significance of these associations[4]. Results Our simulation effectively captured viral burden dynamics across various experimental conditions and virus-specific parameters. Among the viral parameters, all four key factors (ß, p, c and d) significantly influenced EC50 estimation, with Kendall’s t values of 0.48, -0.18, 0.21, and 0.45, respectively. Notably, when ß increased from 6 × 10?8 to 6 × 10?6, while keeping other parameters constant, the final EC50 rose from 1.65 to 20.46. Regarding experimental conditions, MOI, sampling time, and time of adding drugs had minimal impact on EC50, with Kendall’s t values of 0.01, 0.18, and -0.11, respectively. In contrast, initial viral load (V0) had a significant effect on EC50 estimation (Kendall’s t = 0.9). Specifically, when T0 increased from 107 to 108, EC50 rose from 1.65 to 4.48. Conclusions Our findings highlight key factors influencing EC50 estimation. Understanding of these factors will help optimize experimental conditions and will improve the accuracy of pharmacodynamic assessments. Moreover, this study facilitates the translation of in vitro antiviral drug research into clinical studies, supporting both clinical practice and drug development.
[1] Andrei G. Frontiers in Virology, 2021, 1: 666548. [2] Damiani E, et al. Toxicol Lett. 2019 Mar 1;302:28-34. [3] Chemaly RF, et al. Antiviral Res. 2019 Mar;163:50-58. [4] Liebscher E. Computational Statistics & Data Analysis. 2021 May; 157: 107140.
Reference: PAGE 33 (2025) Abstr 11738 [www.page-meeting.org/?abstract=11738]
Poster: Drug/Disease Modelling - Infection