Adrien Mitard (1,2), Cécile Hérate (3), Nathalie Dereuddre-Bosquet (3), Romain Marlin (3), Flora Donati (4), Yannis Rahou (4), Sylvie van der Werf (4), Etienne Simon-Loriere (4,5), Roger Le Grand (3), Mélanie Prague (2), Jérémie Guedj (1)
(1) Université de Paris, IAME, INSERM F-75018 Paris, France. (2) Inria Bordeaux Sud-Ouest, Inserm, Bordeaux Population Health Research Center, SISTM Team, UMR 1219, University of Bordeaux, Bordeaux, France. (3) Université Paris-Saclay, Inserm, CEA, Center for Immunology of Viral, Auto-immune, Hematological and Bacterial diseases (IMVA-HB/IDMIT), Fontenay-aux-Roses & Le Kremlin-Bicêtre, Paris, France. (4) National Reference Center for Respiratory Viruses, Institut Pasteur, Paris, France. (5) G5 Evolutionary Genomics of RNA Viruses, Institut Pasteur, Université Paris Cité, Paris, France
Introduction: Despite unprecedented efforts to tackle a pandemic, SARS-CoV-2 is still circulating at a high rate and will most likely continue to do so. In the context of Omicron virus, vaccines still prevent severe disease but have poorer efficacy against infection. Part of the population, mainly immunocompromised individuals, remains at risk. Therefore, establishing Correlates of Protection against infection is crucial but challenging and can only be performed in a context of experimental challenges, where viral replication is measured frequently after infection.
This question can be addressed using non-human primate (NHP) models [1]. Here we use data generated in cynomolgus macaques, with different immunological backgrounds, infected with BQ.1.1 to develop a mathematical model of SARS-CoV-2 immunisation and infection.
Objectives:
- Develop a model of viral and immune dynamics considering the retroactive loop between virus and antibodies.
- With a simulation work, investigate neutralizing and binding antibodies levels as mechanistic correlates of protection against infection and transmission as high viral load is associated with higher transmission [2].
Methods:
Non-human primate data:
Our study includes 22 NHP infected with a BQ.1.1 isolate. Animals are infected via nasopharyngeal and intratracheal route. The study is composed of 4 groups:
- Control (6 NHP)
- Mono + Mono (6): Received 2 doses of the BNT162b2 vaccine, 4 and 3 months before challenge (human dose)
- Mono + Bi (6): Received 1 dose of the monovalent and 1 dose of the bivalent (Wuhan/BA.4-BA.5) vaccine
- Conv + Bi (4): Challenged by a previous omicron strain 1 year before (BA.2) and then received a dose of bivalent vaccine 3 months before the challenge
Genomic RNA, subgenomic RNA, infectious titers (TCID50, only nasopharyngeal) were followed in the nasopharynx and the trachea post challenge. Neutralisation (ID50, after challenge) and IgG binding (AU/mL, since vaccination) were measured in serum against various strains.
Viral kinetics and immune response model:
We used a target cell limited model to characterize the viral load of infected animals. Starting from a previous model [3], we added an antigen compartment filled either through vaccination or directly by the modelled viral load. This compartment stimulates the replication of B cells with a saturated speed. They produce binding antibodies that form immune complexes with free virions inducing a quicker elimination [4]. Our model also includes antibody-dependent cellular cytotoxicity (ADCC) and modelling of neutralisation has been tested.
Parameter estimation was performed in a non-linear mixed effect framework and the likelihood was maximized using the SAEM algorithm implemented in the Monolix software [8]. We selected the random effects using a downward approach with the BICc as indicator [6]. We tested for the need of different parameters between nasopharynx and trachea and addressed identifiability issues.
Results: We developed a model integrating viral load and antibodies dynamics. It replicates well the trajectories of the various biomarkers despite very different trends between groups without using this information as a covariate. The inter-individual variability in the elimination of infected cells is for example explained by ADCC. From the equations we can derive an intra host R0 (number of cells infected by viruses produced by one infected cell) depending on the binding level at baseline. It is as high as 19 for a protection of 26 AU/mL (the estimate for naive individuals). R0 is equal to 0.3 for a baseline of 5.5*105AU/mL, the mean protection of the Conv+Bi group. Based on this indicator, we could use the binding level as a correlate of protection against infection. We can in fact compute the probability that a given inoculum establishes an infection [10]. To have a protection of 90% against a human-like infection (10 infecting particles [8]), the binding level should be 2.19*104 AU/mL.
Through simulations of the infectious viral load after a human-like infection, we quantified the risks of transmission. This probability is considered very low if the infectious titer is below the limit of detection [9]. It corresponds to a baseline of 2.0*103 AU/mL.
Conclusion: Our model allows us to propose correlates of protection against infection and transmission in NHPs. After a translational work, we will need to verify that our threshold is coherent with binding levels in the general population [12],[13].
References:
[1] M. Alexandre et al., « Modelling the response to vaccine in non-human primates to define SARS-CoV-2 mechanistic correlates of protection », eLife, vol. 11, p. e75427, 2022.
[2] A. Marc et al., « Quantifying the relationship between SARS-CoV-2 viral load and infectiousness », eLife, vol. 10, p. e69302, 2021.
[3] A. Marc et al., « Impact of variants of concern on SARS-CoV-2 viral dynamics in non-human primates », PLOS Comput. Biol., vol. 19, no 8, p. e1010721, 2023.
[4] T. Igarashi et al., « Human immunodeficiency virus type 1 neutralizing antibodies accelerate clearance of cell–free virions from blood plasma », Nat. Med., vol. 5, no 2, p. 211-216, 1999.
[5] E. Kuhn et M. Lavielle, « Maximum likelihood estimation in nonlinear mixed effects models », Comput. Stat. Data Anal., vol. 49, no 4, p. 1020-1038, 2005.
[6] M. Delattre, M. Lavielle, et M.-A. Poursat, « A note on BIC in mixed-effects models », Electron. J. Stat., vol. 8, no 1, 2014.
[7] P. Czuppon et al., « Success of prophylactic antiviral therapy for SARS-CoV-2: Predicted critical efficacies and impact of different drug-specific mechanisms of action », PLOS Comput. Biol., vol. 17, no 3, p. e1008752, 2021.
[8] B. Killingley et al., « Safety, tolerability and viral kinetics during SARS-CoV-2 human challenge in young adults », Nat. Med., vol. 28, no 5, p. 1031-1041, 2022.
[9] G. Lingas et al., « Neutralizing Antibody Levels as a Correlate of Protection Against SARS-CoV-2 Infection: A Modeling Analysis », Clin. Pharmacol. Ther., vol. 115, no 1, p. 86-94, 2024.
[10] F. Carrat et al., « Heterogeneous SARS-CoV-2 humoral response after COVID-19 vaccination and/or infection in the general population », Sci. Rep., vol. 12, no 1, p. 8622, 2022.
[11] Q. Clairon et al., « Modeling the kinetics of the neutralizing antibody response against SARS-CoV-2 variants after several administrations of Bnt162b2 », PLoS Comput. Biol., vol. 19, no 8, p. e1011282, 2023.
Reference: PAGE 32 (2024) Abstr 11126 [www.page-meeting.org/?abstract=11126]
Poster: Drug/Disease Modelling - Other Topics