Research ArticleCORONAVIRUS

Potency and timing of antiviral therapy as determinants of duration of SARS-CoV-2 shedding and intensity of inflammatory response

See allHide authors and affiliations

Science Advances  20 Nov 2020:
Vol. 6, no. 47, eabc7112
DOI: 10.1126/sciadv.abc7112
  • Fig. 1 Mathematical model recapitulation of untreated SARS-CoV-2 kinetics.

    (A) Mathematical model schematic including infection of susceptible cells (S), production of virus (V) by infected cells (I), an early density-dependent immune response governed by exponent k, and a mounting T cell response with sequential populations of T cells (M1, M2, and E), which kill infected cells when above a certain threshold. (B) Model fit to individual data. Shapes are individual viral loads, and lines are model load projections. S, Singapore; G, Germany; K, South Korea; F, France.

  • Fig. 2 Early innate and late acquired killing rates of SARS-CoV-2–infected cells.

    Model projections of rates in 15 participants who cleared viral shedding. (A) Per-cell death rate mediated by innate responses (blue) and acquired immune responses (green). (B) Total death rate mediated by innate responses (blue) and acquired responses (green).

  • Fig. 3 Projected PK and PD of remdesivir therapy.

    (A) Complete model of remdesivir (RDV) including plasma levels of parent drug, intracellular levels of the active component (NTP), and antiviral efficacy of drug according to NTP concentration. (B) Projections of plasma remdesivir levels and intracellular NTP in PBMCs. Data points from NHP experiments are dots, while lines are model projections. (C) Simulated concentrations of the parent compound and intracellular levels of the active compound with a loading dose of 200 mg intravenously (IV) followed by nine daily doses of 100 mg intravenously. VL, viral load. (D) PD projections of antiviral efficacy according to drug concentration assuming different values for the in vivo EC50 of the drug. (E) Projected antiviral effects using combined PK and PD models at different assumed drug potencies.

  • Fig. 4 Treatment projections of a 10-day remdesivir course assuming different potency and timing of treatment.

    Each set of simulations is performed under assumptions of high, medium, and low potency (EC50 = 0.8, 8, and 80 μM, respectively). Treatment initiation at time points generally consistent with (A) hospitalization (day 10 after first positive sample), (B) first symptoms (day 5 after first positive sample), (C) presymptomatic post-peak phase (day 2 after first positive sample), and (D) presymptomatic pre-peak phase (day 0). Overall, early potent treatment limits duration of infection. Prolonged shedding is predicted as a possibility with subpotent, early initiation of therapy due to inadequate activation of immunity.

  • Fig. 5 Predictors of SARS-CoV-2 treatment outcomes with remdesivir.

    Heatmaps comparing variance in drug potency measured by in vivo EC50 (x axis) and timing of treatment initiation (y axis) for (A) shedding duration, (B) viral load AUC, and (C) extent of T cell response required for viral elimination. Potent therapy within the first 5 days of infection limits shedding duration and the extent of the T cell response. However, only extremely early therapy during the presymptomatic phase of infection markedly lowers viral AUC. Subpotent therapy given during the extremely early presymptomatic stage may extend shedding duration at lower viral loads by limiting the immune response.

  • Fig. 6 Projections of remdesivir drug resistance during therapy.

    Simulations are with assumed high potency (EC50 = 0.8 μM) and the assumption that mutants confer partial drug resistance. Treatment initiation is at time points generally consistent with hospitalization (day 10 after first positive sample), first symptoms (day 5 after first positive sample), presymptomatic post-peak phase (day 2 after first positive sample), or presymptomatic pre-peak phase (day 0). (A) Projections of no treatment, treatment with no assumed drug resistance, and treatment with assumed drug resistance. (B) Projections of assumed drug resistance with trajectories of sensitive strains, single mutants, and double mutants. Here, DOT represents the day of the start of the treatment.

  • Fig. 7 Projections of remdesivir drug resistance during therapy.

    Simulations are with moderate potency (EC50 = 8.0 μM) and the assumption that mutants confer partial drug resistance. Treatment initiation is at time points generally consistent with hospitalization (day 10 after first positive sample), first symptoms (day 5 after first positive sample), presymptomatic post-peak phase (day 2 after first positive sample), or presymptomatic pre-peak phase (day 0). (A) Projections of no treatment, treatment with no assumed drug resistance, and treatment with assumed drug resistance. (B) Projections of assumed drug resistance with trajectories of sensitive strains, single mutants, and double mutants. Here, DOT represents the day of the start of the treatment.

  • Fig. 8 Projected PK and PD of selinexor therapy.

    (A) Complete model of selinexor including plasma levels, PBMC levels, inhibition of XPO1-mediated cellular export of viral proteins, and inhibition of viral replication in a dose-dependent fashion. (B) Single-dose model drug levels (line) reproduce observed data points at different doses. (C) Projected PBMC levels assuming low and high selinexor absorption rates. (D) Predicted fold change in compensatory XPO1 mRNA expression after a single dose with model fits to data. (E) Projected plasma (black) and PBMC (red; dashed = high absorption and solid = low absorption) selinexor levels with three times weekly dosing for 4 weeks. (F) PD projections of antiviral efficacy according to drug concentration assuming different values for the in vivo EC50 of the drug. (G) Predicted fold change in compensatory XPO1 mRNA expression after multiple doses. (H) Projected antiviral effects using combined PK and PD models at different assumed drug potencies.

  • Fig. 9 Projected PK and PD of neutralizing antibody therapy.

    (A) Complete bicompartment model of bNAb therapy with lowering of viral infectivity according to antibody concentration. (B) Projections of plasma bNAb levels from simulations of VRC01 treatment. (C) PD projections of antiviral efficacy according to antibody concentration assuming different values for the in vivo EC50. (D) Combination simulations of PK and PD models demonstrating antiviral activity as a function of time.

  • Fig. 10 Treatment projections of cytolytic immunotherapy assuming different potency and timing of treatment.

    Each set of simulations is performed under assumptions of high, medium, and low potency based on multiplicative effect on infected cell death rate. Treatment initiation is at time points generally consistent with (A) hospitalization (day 10 after first positive sample), (B) first symptoms (day 5 after first positive sample), (C) presymptomatic post-peak phase (day 2 after first positive sample), and (D) presymptomatic pre-peak phase (day 0). Overall, early potent treatment limits duration of infection, but extremely early therapy fails. Prolonged shedding is predicted as a possibility with subpotent, early initiation of therapy due to inadequate activation of immunity.

Supplementary Materials

  • Supplementary Materials

    Potency and timing of antiviral therapy as determinants of duration of SARS-CoV-2 shedding and intensity of inflammatory response

    Ashish Goyal, E. Fabian Cardozo-Ojeda, Joshua T. Schiffer

    Download Supplement

    This PDF file includes:

    • Figs. S1 to S5
    • Tables S1 to S4

    Files in this Data Supplement:

Stay Connected to Science Advances

Navigate This Article