Nathanaël Hozé
Université Paris Cité, IAME, Inserm, F-75018, Paris, France,
Objectives:
Our understanding of the dynamics of SARS-CoV-2 infection within a host and how it relates to an individual’s ability to transmit the virus is currently limited. Some of the reasons are that the exact moment of transmission is not observable, viral kinetics varies greatly between individuals, and contacts are heterogeneous. Epidemiological studies of infection within households provide the ideal set up for studying the relation between viral load and transmission potential. Indeed, transmission between household members is high and these studies offer a relatively controllable environment where individuals that are involved are clearly identified and where frequent samples of the viral load can be acquired.
We propose here a new statistical framework that is based on a multiscale model that combines the within-host viral dynamics with a transmission model. We build the inference framework and answer three main questions: First, can we use household studies the relationship between the viral load at the time of transmission and the probability of transmission? Second, how to reconstruct the chains of transmission? Third, how can we optimize the design of household studies to establish the role of viral load in transmission?
Methods:
Within-host dynamics was modeled using a system of ordinary differential equations to describe the evolution of the population of target cells, infected cells that produce the virus, the virus particles, and the innate immune response. The model was calibrated on the viral load time-series of 4000 SARS-CoV-2 infected individuals [1]. Transmission in households is modeled as repeated contacts between all individuals. The transmission probability is described by a power law model to account for a non-linear relationship between the viral load and the instantaneous probability of infection at contact [2]. Household of different size were simulated for different values of the transmission parameter.
Results:
In a series of simulation scenarios, we varied the number and the size of households in the study, the sampling frequency, and the parameters relating viral load and transmission. We propose a two-step method to estimate the link between transmissibility and viral load. First, households are simulated and the individual parameters are estimated using a non-linear mixed effect model. Median estimates of the individual parameters were used to reconstruct the complete time-series of the viral load. Since the time of infection is not observed, we use the viral load and additional information on the timing of symptom onset for its estimation. A Markov Chain Monte Carlo method is used to estimate the transmission parameter based on the reconstructed viral load and time of infection. We show that a frequent measurement of the viral load is necessary to estimate the parameters, because it is required to provide an accurate estimate of the time of infection.
Conclusions:
Household studies provide an ideal setting to establish the relation between viral load and transmission. The study is designed for SARS-CoV-2 but could be extended to other respiratory viruses. Next steps of the analysis will include vaccination status to estimate the level of protection against infection conferred by the vaccine.
References:
[1] Conrado JD, et al. Viral kinetics in COVID-19 outpatients treated with CASIRIVIMAB+IMDEVIMAB combination. Conference on Retroviruses and Opportunistic Infections, 2022.
[2] Ke R, Zitzmann C, Ho DD, Ribeiro RM Perelson AS In vivo kinetics of SARS-CoV-2 infection and its relationship with a person’s infectiousness. Proceedings of the National Academy of Sciences, 2021, 118(49), e2111477118
Reference: PAGE 32 (2024) Abstr 11220 [www.page-meeting.org/?abstract=11220]
Poster: Methodology - New Modelling Approaches