Ebola viral dynamics in nonhuman primates provides insights into virus immuno-pathogenesis and antiviral strategies

Abstract

Despite several clinical trials implemented, no antiviral drug could demonstrate efficacy against Ebola virus. In non-human primates, early initiation of polymerase inhibitors favipiravir and remdesivir improves survival, but whether they could be effective in patients is unknown. Here we analyze the impact of antiviral therapy by using a mathematical model that integrates virological and immunological data of 44 cynomolgus macaques, left untreated or treated with favipiravir. We estimate that favipiravir has a ~50% efficacy in blocking viral production, which results in reducing virus growth and cytokine storm while IFNα reduces cell susceptibility to infection. Simulating the effect of delayed initiations of treatment, our model predicts survival rates of 60% for favipiravir and 100% for remdesivir when treatment is initiated within 3 and 4 days post infection, respectively. These results improve the understanding of Ebola immuno-pathogenesis and can help optimize antiviral evaluation in future outbreaks.

Fig. 1

Survival and virological data. a Design of the four infected NHP experiments. b Survival curves. c Ebola virus plasma viral load. d Favipiravir plasma concentrations. Dosing regimen groups were untreated (black), 100 mg kg⁻¹ BID (green), 150 mg kg⁻¹ BID (blue), and 180 mg kg⁻¹ BID (red). Gray areas correspond to dosing periods

Fig. 2

Cytokine and immunological data. Plasma pro inflammatory cytokines Interleukin-6, Interferon α and Tumor necrosis factor α, and CD8 T cells expressing perforin in macaques left untreated (black, left panels), receiving favipiravir 150 mg kg⁻¹ BID (blue, middle panels) or 180 mg kg⁻¹ (red, right panels). Gray areas correspond to dosing periods

Fig. 3

Schematic and mathematical model of Ebola virus infection. As the virus early disseminate in blood and lymph circulation, the model assumes only one target cell compartment. After infection by free virus (V), target cells (T) enter an eclipse phase (I₁) before becoming productively infected (I₂), which activate macrophage and production of cytokines. Cytokine release, in particular IFNα, leads to an increase of cells that are refractory to infection (R), and triggers apoptosis of non-specific CD8 T cells (E₁), which creates room for proliferation of EBOV-specific CD8 T cells (E₂) that eliminate productively infected cells I₂ Polymerase inhibitors as favipiravir inhibit viral production with efficacy ε

Fig. 4

Model-based predicted survival rate (plain lines) and Kaplan–Meier survival curves and 95% confidence interval (shaded areas) for each dosing regimen group. Top left: cynomolgus experiment control pooled from the 4 experiments (N = 28), top right: favipiravir 100 mg kg⁻¹ BID (N = 6), bottom left: favipiravir 150 mg kg⁻¹ BID (N = 5), bottom right: favipiravir 180 mg kg⁻¹ BID (N = 5). In each panel, model-based predicted survival rate was calculated by simulating 1000 individuals for each dosing regimen and taking the medians of the predicted individual survival functions at each time

Fig. 5

Individual observation (dots) and median model predictions (line) in animals left untreated (black) or treated with 180 mg kg⁻¹ BID favipiravir (red). Top left: Ebola virus viral load; top middle CD8 T cells expressing perforin; top right: model prediction of productive infected cell elimination rate, increasing concomitantly to adaptive response; middle left: IFNα concentrations; middle: IL6 concentration; middle right: TNFα concentration; bottom left: predicted ratio of protected cells on the concentration of IFNα (R/F); bottom middle: model prediction of the ratio of infected on protected cells (R/I₂); bottom right: Kaplan Meier curves and survival rate prediction. Shaded areas represent the predictions within the 10th and 90th percentiles of 1000 simulations

Fig. 6

Viral load and survival rate predictions for various levels of drug efficacy and timing of treatment initiation, assuming no pharmacokinetic variability. Left: ε = 50% (eg, favipiravir), middle: ε = 90% (eg, GS-5734); right: ε = 99%. Dark blue: treatment initiation at D0, magenta: D2, light blue: D3, yellow: D4, green: D5, brown: D7. Results show the median of 1000 simulations