Modeling SARS-CoV-2 viral kinetics and association with mortality in hospitalized patients from the French COVID cohort

Abstract

The characterization of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) viral kinetics in hospitalized patients and its association with mortality is unknown. We analyzed death and nasopharyngeal viral kinetics in 655 hospitalized patients from the prospective French COVID cohort. The model predicted a median peak viral load that coincided with symptom onset. Patients with age ≥65 y had a smaller loss rate of infected cells, leading to a delayed median time to viral clearance occurring 16 d after symptom onset as compared to 13 d in younger patients (P < 10−4). In multivariate analysis, the risk factors associated with mortality were age ≥65 y, male gender, and presence of chronic pulmonary disease (hazard ratio [HR] > 2.0). Using a joint model, viral dynamics after hospital admission was an independent predictor of mortality (HR = 1.31, P < 10−3). Finally, we used our model to simulate the effects of effective pharmacological interventions on time to viral clearance and mortality. A treatment able to reduce viral production by 90% upon hospital admission would shorten the time to viral clearance by 2.0 and 2.9 d in patients of age <65 y and ≥65 y, respectively. Assuming that the association between viral dynamics and mortality would remain similar to that observed in our population, this could translate into a reduction of mortality from 19 to 14% in patients of age ≥65 y with risk factors. Our results show that viral dynamics is associated with mortality in hospitalized patients. Strategies aiming to reduce viral load could have an effect on mortality rate in this population.

The relative strength and timing of innate immune and CD8 T-cell responses underlie the heterogeneous outcomes of SARS-CoV-2 infection

“…We therefore added an estimated length of the prodromal period to the timepoints in order to compare our model predictions. We set this length to 4.8 d from the German transmission chain data (37, 38) (supplementary table S2), which is also consistent with other reported estimates (33). We estimated the viral load, , from our model predictions using the pseudo-steady state approximation, ≈ / , where is the per capita rate of viral production from infected cells and is the per capita rate of viral clearance.…”

Variant-specific SARS-CoV-2 within-host kinetics

“…A minority of samples originated from hospitals (6%). Most individuals (88%) were sampled twice and the median follow-up 2 duration was 8 days (IQR [6][7][8][9][10][11][12] days). Most samples originated from the Ile-de-France region (64%) and were caused by the V1 variant (73%), which reflects the state of the French epidemic at the time where the samples were collected [13].…”

“…Understanding the within-host kinetics of SARS-CoV-2 infections already already yielded original insights on infection virulence [10], or efficiency of screening strategies [11]. Here, we analyzed a large national dataset of longitudinal RT-PCR Ct values to test the hypothesis that epidemic rebounds associated with SARS-CoV-2 variants could be explained by phenotypic differences in the infections they cause.…”

“…In the case of SARS-CoV-2, numerous studies used Ct values as a proxy for virus load to report temporal variations within individuals. Some of these performed statistical analyses on longitudinal data from hospitalized patients [10], health workers [11], and experimental infections in non-human primates [12]. These studies show that viral kinetics are associated with the infection and the detection probability.…”

“…The copyright holder for this preprint this version posted May 29, 2021. ; https://doi.org/10.1101/2021.05.26.21257835 doi: medRxiv preprint A limitation of this analysis is that we do not have any indication regarding the date of the infection or of the symptom onset. This uncertainty prevents us from analyzing more mechanistic models with nonlinear mixed-effect models [10]. However, since we the nature of the virus causing the infection is unlikely to affect the number of days between infection and screening, we do not expect our assumption that the lowest Ct value corresponds to the peak viral load to introduce biases.…”