Zika plasma viral dynamics in nonhuman primates provides insights into early infection and antiviral strategies

Abstract

The recent outbreak of Zika virus (ZIKV) has been associated with fetal abnormalities and neurological complications, prompting global concern. Here we present a mathematical analysis of the within-host dynamics of plasma ZIKV burden in a nonhuman primate model, allowing for characterization of the growth and clearance of ZIKV within individual macaques. We estimate that the eclipse phase for ZIKV, the time between cell infection and viral production, is most likely short (∼4 h), the median within-host basic reproductive number R₀ is 10.7, the rate of viral production is rapid (>25,000 virions d^-1), and the lifetime of an infected cell while producing virus is ∼5 h. We also estimate that the minimum number of virions produced by an infected cell over its lifetime is ∼5,500. We assess the potential effect of an antiviral treatment that blocks viral replication, showing that the median time to undetectable plasma viral load (VL) can be reduced from ∼5 d to ∼3 d with a drug concentration ∼15 times the drug's EC₅₀ when treatment is given prophylactically starting at the time of infection. In the case of favipiravir, a polymerase inhibitor with activity against ZIKV, we predict a dose of 150 mg/kg given twice a day initiated at the time of infection can reduce the peak median VL by ∼3 logs and shorten the time to undetectable median VL by ∼2 d, whereas treatment given 2 d postinfection is mostly ineffective in accelerating plasma VL loss in macaques.

Keywords: Zika virus; antiviral therapy; mathematical modeling; viral dynamics.

Fig. 1.

Best fit of model (Eq. 1, solid lines) to the observed plasma VL (black circles), with fixed k = 6 d⁻¹, c = 10 d⁻¹, V(0) = 10⁴ mL⁻¹, T(0) = 10⁵ mL⁻¹, and fixed ω = 0.4 for fitted parameters R₀, δ, and p (see text for further details). When virus was not detectable in a sample it is displayed at a value equal to the limit of detection, 200 RNA copies per mL. Subplot titles indicate animal ID.

Fig. 2.

The predicted effect of a constant concentration, C, of an antiviral drug initiated at the time of infection that reduces viral production rate p from infected cells by a factor 1 − ε [where ε = C/(EC₅₀ + C) is the drug effectiveness] on the (A) basic reproductive ratio, (B) peak VL, (C) time until the VL becomes undetectable (below 200 RNA copies per mL), and (D) log₁₀ VL AUC. At each value on the x axis, 5,000 viral dynamic parameter sets were sampled from the estimated parameter distributions as determined by mixed effects model fitting. Black lines represent the median value across simulations and gray lines represent the 5th and 95th percentiles. In A, the horizontal black dashed line represents R₀ = 1.

Fig. 3.

The predicted effect in male macaques of a 150 mg/kg twice daily regime of favipiravir, initiated at time of infection (red) or 2 d after infection (blue) on (A) median VL and (B) distribution of log₁₀ VL AUCs. In A, the horizontal black dashed line represents the limit of detection of the experimental assay, 200 RNA copies per mL. In B, median AUC values are 14.5 (no drug), 11.8 (2 dpi), and 4.7 (0 dpi). Data shown are from 5,000 repeated simulations, each including independent simulations of the PK profile and VL dynamics determined by parameters R₀, δ, and p selected at random from the population distribution predicted by mixed effects model fitting.