Developing SARS-CoV-2 viral dynamic model in patients with COVID-19 based on amount of viral RNA and viral titer
Daichi Yamaguchi (1), Ryosuke Shimizu (1), Ryuji Kubota (1)
(1) Clinical Pharmacology & Pharmacokinetics, Shionogi & Co., Ltd.
Introduction:
Target cell-limited models involve the process from viral infection to viral clearance in humans, and they are often used to characterize viral dynamics. This modeling approach has been also applied to describe the dynamics of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) [1-5], mainly used patients’ data in early phase of this pandemic. However, these models were developed using only the information on the amount of viral RNA, including both the infectious and non-infectious virus data, and the dynamics for only the infectious virus has not been modeled appropriately. In addition, the environment surrounding this infection has dramatically changed from early phase, such as most people have been vaccinated for coronavirus disease 2019 (COVID-19) [6-9] and the change of major SARS-CoV-2 variants [10]. We have obtained the amount of viral RNA and viral titer for only the infectious virus from patients with COVID-19 in the clinical trial for ensitrelvir (product code S-217622) [11-14]. The circumstance in our clinical trial would reflect that in current clinical practice during the Omicron epidemic. In this study, therefore, further detailed viral dynamic model for SARS-CoV-2 was developed based on both amount of viral RNA and viral titer from patients with COVID-19. We also evaluated the covariates affecting the dynamics.
Objectives:
- To develop SARS-CoV-2 viral dynamic model using observed amount of viral RNA and viral titer
- To evaluate the covariates that have an impact on the viral dynamics
Methods:
Clinical Trial Data
Our study includes consecutive 3330 amount of viral RNA and 3332 viral titer data from 464 patients with COVID-19 in the placebo group in the phase 2/3 study (phase 2a, 2b, and 3 parts) for ensitrelvir [12-14]. The patients with mild-to-moderate COVID-19 based on the US Food and Drug Administration guidance [15] and who tested positive for SARS-CoV-2 within 120 hours prior to randomization were collected in this study. The investigated intrinsic or extrinsic factors in the patients (e.g. age, body weight, history of vaccination, and SARS-CoV-2 variants) were also collected. The time from symptom onset to randomization was reported as categorical data every 24 hours, and the median value for each category was used in our analysis.
Viral Dynamic Model
Our model modified the previous target cell-limited model with immune function [1] to characterize SARS-CoV-2 dynamics and immune reactions for the infection. Both observed data were used for the dynamic parameter estimation. The time from infection to symptom onset (Tinf) was estimated in each patient by the viral dynamic model. The investigated intrinsic or extrinsic factors in the patients were evaluated to identify the covariates which affect the dynamics significantly. The selection of model structure and covariates were performed based on Bayesian information criterion and visual prediction accuracy. The parameter estimation was performed by the likelihood maximized using the stochastic approximation expectation–maximization algorithm implemented in NONMEM ver.7.4.4 [16].
Results:
The analysis population included patients aged 13 to 69 years old who were mainly vaccinated (90.1%) and infected with Omicron variant (91.6%). Our model assumed that “productive infected cells” produce only infectious viruses and then some infectious viruses transit to non-infectious viruses, unlike the previous model in which both infectious and non-infectious viruses were produced by “productive infected cells” based on a constant rate [1]. This model well described both dynamics of viral RNA and viral titer. Vaccination status was identified as a significant covariate for Tinf and the immune function parameter. Tinf of the unvaccinated patients was estimated as 2.8 days and shorter than that of the vaccinated patients (3.3 days). Immune strength for increasing the elimination rate constant of infected cells in the vaccinated patients was 1.1 times stronger than that in the unvaccinated patients.
Conclusions:
We characterized the SARS-CoV-2 viral dynamics using both the amount of viral RNA and viral titer in the population who mainly included vaccinated and Omicron-infected patients. This model allows to predict the viral dynamics in patients with COVID-19, including the time from infection to onset. Vaccination status was identified as a factor influencing the Tinf and the immune function.
References:
[1] Néant N et al. PNAS. (2021) 118, e2017962118.
[2] Goncalves1 A et al. CPT Pharmacometrics Syst. Pharmacol. (2020) 9, 509–514.
[3] Cao Y et al. Clin Trans Sci. (2021) 14, 2348-2359.
[4] Odaka M et al. Heliyon. (2021) 7, e08207.
[5] Kim KS et al. PLos Biol. (2021) 19, e3001128.
[6] Baden LR et al. N Engl J Med. (2021) 384, 403–16.
[7] Polack FP et al. N Engl J Med. (2020) 383, 2603–15.
[8] Voysey M et al. Lancet. (2021) 397, 881–91.
[9] Heath PT et al. N Engl J Med. (2021) 385, 1172–83.
[10] https://covariants.org/
[11] Unoh Y et al. J Med Chem. (2022) 65, 6499–6512.
[12] Mukae H et al. AAC. (2022) 66, e0069722.
[13] Mukae H et al. Clin Infect Dis. (2022) ciac933. doi: 10.1093/cid/ciac933
[14] Mukae H et al. Special lecture. JAID western branch annual conference 2023
[15] https://www.fda.gov/regulatory-information/search-fda-guidance-documents/assessing-covid-19-related-symptoms-outpatient-adult-and-adolescent-subjects-clinical-trialsdrugs
[16] Beal SL et al. 1989-2011. NONMEM Users Guides. Icon Development Solutions, Ellicott City, Maryland, USA.
