Modeling Favipiravir Antiviral Efficacy Against Emerging Viruses: From Animal Studies to Clinical Trials.

In 2014, our research network was involved in the evaluation of favipiravir, an anti-influenza polymerase inhibitor, against Ebola virus. In this review, we discuss how mathematical modeling was used, first to propose a relevant dosing regimen in humans, and then to optimize its antiviral efficacy in a nonhuman primate (NHP) model. The data collected in NHPs were finally used to develop a model of Ebola pathogenesis integrating the interactions among the virus, the innate and adaptive immune response, and the action of favipiravir. We conclude the review of this work by discussing how these results are of relevance for future human studies in the context of Ebola virus, but also for other emerging viral diseases for which no therapeutics are available.

Antiviral efficacy of favipiravir against Ebola virus: A translational study in cynomolgus macaques.

BACKGROUND: Despite repeated outbreaks, in particular the devastating 2014-2016 epidemic, there is no effective treatment validated for patients with Ebola virus disease (EVD). Among the drug candidates is the broad-spectrum polymerase inhibitor favipiravir, which showed a good tolerance profile in patients with EVD (JIKI trial) but did not demonstrate a strong antiviral efficacy. In order to gain new insights into the antiviral efficacy of favipiravir and improve preparedness and public health management of future outbreaks, we assess the efficacy achieved by ascending doses of favipiravir in Ebola-virus-infected nonhuman primates (NHPs). METHODS AND FINDINGS: A total of 26 animals (Macaca fascicularis) were challenged intramuscularly at day 0 with 1,000 focus-forming units of Ebola virus Gabon 2001 strain and followed for 21 days (study termination). This included 13 animals left untreated and 13 treated with doses of 100, 150, and 180 mg/kg (N = 3, 5, and 5, respectively) favipiravir administered intravenously twice a day for 14 days, starting 2 days before infection. All animals left untreated or treated with 100 mg/kg died within 10 days post-infection, while animals receiving 150 and 180 mg/kg had extended survival (P < 0.001 and 0.001, respectively, compared to untreated animals), leading to a survival rate of 40% (2/5) and 60% (3/5), respectively, at day 21. Favipiravir inhibited viral replication (molecular and infectious viral loads) in a drug-concentration-dependent manner (P values < 0.001), and genomic deep sequencing analyses showed an increase in virus mutagenesis over time. These results allowed us to identify that plasma trough favipiravir concentrations greater than 70-80 µg/ml were associated with reduced viral loads, lower virus infectivity, and extended survival. These levels are higher than those found in the JIKI trial, where patients had median trough drug concentrations equal to 46 and 26 µg/ml at day 2 and day 4 post-treatment, respectively, and suggest that the dosing regimen in the JIKI trial was suboptimal. The environment of a biosafety level 4 laboratory introduces a number of limitations, in particular the difficulty of conducting blind studies and performing detailed pharmacological assessments. Further, the extrapolation of the results to patients with EVD is limited by the fact that the model is fully lethal and that treatment initiation in patients with EVD is most often initiated several days after infection, when symptoms and high levels of viral replication are already present. CONCLUSIONS: Our results suggest that favipiravir may be an effective antiviral drug against Ebola virus that relies on RNA chain termination and possibly error catastrophe. These results, together with previous data collected on tolerance and pharmacokinetics in both NHPs and humans, support a potential role for high doses of favipiravir for future human interventions.

Determining Ribavirin's mechanism of action against Lassa virus infection.

Ribavirin is a broad spectrum antiviral which inhibits Lassa virus (LASV) replication in vitro but exhibits a minor effect on viremia in vivo. However, ribavirin significantly improves the disease outcome when administered in combination with sub-optimal doses of favipiravir, a strong antiviral drug. The mechanisms explaining these conflicting findings have not been determined, so far. Here, we used an interdisciplinary approach combining mathematical models and experimental data in LASV-infected mice that were treated with ribavirin alone or in combination with the drug favipiravir to explore different putative mechanisms of action for ribavirin. We test four different hypotheses that have been previously suggested for ribavirin's mode of action: (i) acting as a mutagen, thereby limiting the infectivity of new virions; (ii) reducing viremia by impairing viral production; (iii) modulating cell damage, i.e., by reducing inflammation, and (iv) enhancing antiviral immunity. Our analysis indicates that enhancement of antiviral immunity, as well as effects on viral production or transmission are unlikely to be ribavirin's main mechanism mediating its antiviral effectiveness against LASV infection. Instead, the modeled viral kinetics suggest that the main mode of action of ribavirin is to protect infected cells from dying, possibly reducing the inflammatory response.

The paradox of highly effective sofosbuvir-based combination therapy despite slow viral decline: can we still rely on viral kinetics?

High sustained virologic response (SVR) rates have been observed after 6 weeks of anti-HCV treatment using sofosbuvir, ledipasvir and a non-nucleoside polymerase-inhibitor (GS-9669) or a protease-inhibitor (GS-9451) and after 12 weeks with sofosbuvir + ledipasvir. Here we analyze the viral kinetics observed during these treatments to decipher the origin of the rapid cure and to evaluate the possibility of further reducing treatment duration. We found that viral kinetics were surprisingly slow in all treatment groups and could not reproduce the high SVR rates observed. Based on experimental results suggesting that NS5A- or protease-inhibitors can generate non-infectious virus, we incorporated this effect into a mathematical model. We found that to predict observed SVR rates it was necessary to assume that ledipasvir, GS-9669 and GS-9451 rapidly reduce virus infectivity. We predicted with this model that 4 weeks of triple therapy could be sufficient to achieve SVR in patients with undetectable viremia at week 1, but would be suboptimal in general. In conclusion, the rapid cure rate achieved with these combinations is largely disconnected from viral loads measured during treatment. A model assuming that rapid cure is due to a drug effect of generating non-infectious virus could be a basis for future response guided therapy.

Individual Bayesian Information Matrix for Predicting Estimation Error and Shrinkage of Individual Parameters Accounting for Data Below the Limit of Quantification.

PURPOSE: In mixed models, the relative standard errors (RSE) and shrinkage of individual parameters can be predicted from the individual Bayesian information matrix (MBF). We proposed an approach accounting for data below the limit of quantification (LOQ) in MBF. METHODS: MBF is the sum of the expectation of the individual Fisher information (MIF) which can be evaluated by First-Order linearization and the inverse of random effect variance. We expressed the individual information as a weighted sum of predicted MIF for every possible design composing of measurements above and/or below LOQ. When evaluating MIF, we derived the likelihood expressed as the product of the likelihood of observed data and the probability for data to be below LOQ. The relevance of RSE and shrinkage predicted by MBF in absence or presence of data below LOQ were evaluated by simulations, using a pharmacokinetic/viral kinetic model defined by differential equations. RESULTS: Simulations showed good agreement between predicted and observed RSE and shrinkage in absence or presence of data below LOQ. We found that RSE and shrinkage increased with sparser designs and with data below LOQ. CONCLUSIONS: The proposed method based on MBF adequately predicted individual RSE and shrinkage, allowing for evaluation of a large number of scenarios without extensive simulations.

Improvement of ALT decay kinetics by all-oral HCV treatment: Role of NS5A inhibitors and differences with IFN-based regimens.

BACKGROUND: Intracellular HCV-RNA reduction is a proposed mechanism of action of direct-acting antivirals (DAAs), alternative to hepatocytes elimination by pegylated-interferon plus ribavirin (PR). We modeled ALT and HCV-RNA kinetics in cirrhotic patients treated with currently-used all-DAA combinations to evaluate their mode of action and cytotoxicity compared with telaprevir (TVR)+PR. STUDY DESIGN: Mathematical modeling of ALT and HCV-RNA kinetics was performed in 111 HCV-1 cirrhotic patients, 81 treated with all-DAA regimens and 30 with TVR+PR. Kinetic-models and Cox-analysis were used to assess determinants of ALT-decay and normalization. RESULTS: HCV-RNA kinetics was biphasic, reflecting a mean effectiveness in blocking viral production >99.8%. The first-phase of viral-decline was faster in patients receiving NS5A-inhibitors compared to TVR+PR or sofosbuvir+simeprevir (p<0.001), reflecting higher efficacy in blocking assembly/secretion. The second-phase, noted δ and attributed to infected-cell loss, was faster in patients receiving TVR+PR or sofosbuvir+simeprevir compared to NS5A-inhibitors (0.27 vs 0.21 d-1, respectively, p = 0.0012). In contrast the rate of ALT-normalization, noted λ, was slower in patients receiving TVR+PR or sofosbuvir+simeprevir compared to NS5A-inhibitors (0.17 vs 0.27 d-1, respectively, p<0.001). There was no significant association between the second-phase of viral-decline and ALT normalization rate and, for a given level of viral reduction, ALT-normalization was more profound in patients receiving DAA, and NS5A in particular, than TVR+PR. CONCLUSIONS: Our data support a process of HCV-clearance by all-DAA regimens potentiated by NS5A-inhibitor, and less relying upon hepatocyte death than IFN-containing regimens. This may underline a process of "cell-cure" by DAAs, leading to a fast improvement of liver homeostasis.

Favipiravir pharmacokinetics in Ebola-Infected patients of the JIKI trial reveals concentrations lower than targeted.

BACKGROUND: In 2014-2015, we assessed favipiravir tolerance and efficacy in patients with Ebola virus (EBOV) disease (EVD) in Guinea (JIKI trial). Because the drug had never been used before for this indication and that high concentrations of the drugs were needed to achieve antiviral efficacy against EBOV, a pharmacokinetic model had been used to propose relevant dosing regimen. Here we report the favipiravir plasma concentrations that were achieved in participants in the JIKI trial and put them in perspective with the model-based targeted concentrations. METHODS AND FINDINGS: Pre-dose drug concentrations were collected at Day-2 and Day-4 of treatment in 66 patients of the JIKI trial and compared to those predicted by the model taking into account patient's individual characteristics. At Day-2, the observed concentrations were slightly lower than the model predictions adjusted for patient's characteristics (median value of 46.1 versus 54.3 µg/mL for observed and predicted concentrations, respectively, p = 0.012). However, the concentrations dropped at Day-4, which was not anticipated by the model (median values of 25.9 and 64.4 µg/mL for observed and predicted concentrations, respectively, p<10-6). There was no significant relationship between favipiravir concentrations and EBOV viral kinetics or mortality. CONCLUSIONS: Favipiravir plasma concentrations in the JIKI trial failed to achieve the target exposure defined before the trial. Furthermore, the drug concentration experienced an unanticipated drop between Day-2 and Day-4. The origin of this drop could be due to severe sepsis conditions and/or to intrinsic properties of favipiravir metabolism. Dose-ranging studies should be performed in healthy volunteers to assess the concentrations and the tolerance that could be achieved with high doses. TRIAL REGISTRATION: ClinicalTrials.gov NCT02329054.

Persistence and clearance of Ebola virus RNA from seminal fluid of Ebola virus disease survivors: a longitudinal analysis and modelling study.

BACKGROUND: By January, 2016, all known transmission chains of the Ebola virus disease (EVD) outbreak in west Africa had been stopped. However, there is concern about persistence of Ebola virus in the reproductive tract of men who have survived EVD. We aimed to use biostatistical modelling to describe the dynamics of Ebola virus RNA load in seminal fluid, including clearance parameters. METHODS: In this longitudinal study, we recruited men who had been discharged from three Ebola treatment units in Guinea between January and July, 2015. Participants provided samples of seminal fluid at follow-up every 3-6 weeks, which we tested for Ebola virus RNA using quantitative real-time RT-PCR. Representative specimens from eight participants were then inoculated into immunodeficient mice to test for infectivity. We used a linear mixed-effect model to analyse the dynamics of virus persistence in seminal fluid over time. FINDINGS: We enrolled 26 participants and tested 130 seminal fluid specimens; median follow up was 197 days (IQR 187-209 days) after enrolment, which corresponded to 255 days (228-287) after disease onset. Ebola virus RNA was detected in 86 semen specimens from 19 (73%) participants. Median duration of Ebola virus RNA detection was 158 days after onset (73-181; maximum 407 days at end of follow-up). Mathematical modelling of the quantitative time-series data showed a mean clearance rate of Ebola virus RNA from seminal fluid of -0·58 log units per month, although the clearance kinetic varied greatly between participants. Using our biostatistical model, we predict that 50% and 90% of male survivors clear Ebola virus RNA from seminal fluid at 115 days (90% prediction interval 72-160) and 294 days (212-399) after disease onset, respectively. We also predicted that the number of men positive for Ebola virus RNA in affected countries would decrease from about 50 in January 2016, to fewer than 1 person by July, 2016. Infectious virus was detected in 15 of 26 (58%) specimens tested in mice. INTERPRETATION: Time to clearance of Ebola virus RNA from seminal fluid varies greatly between individuals and could be more than 13 months. Our predictions will assist in decision-making about surveillance and preventive measures in EVD outbreaks. FUNDING: This study was funded by European Union's Horizon 2020 research and innovation programme, Directorate-General for International Cooperation and Development of the European Commission, Institut national de la santé et de la recherche médicale (INSERM), German Research Foundation (DFG), and Innovative Medicines Initiative 2 Joint Undertaking.

Favipiravir Pharmacokinetics in Nonhuman Primates and Insights for Future Efficacy Studies of Hemorrhagic Fever Viruses.

Favipiravir is an RNA polymerase inhibitor that showed strong antiviral efficacy in vitro and in small-animal models of several viruses responsible for hemorrhagic fever (HF), including Ebola virus. The aim of this work was to characterize the complex pharmacokinetics of favipiravir in nonhuman primates (NHPs) in order to guide future efficacy studies of favipiravir in large-animal models. Four different studies were conducted in 30 uninfected cynomolgus macaques of Chinese (n = 17) or Mauritian (n = 13) origin treated with intravenous favipiravir for 7 to 14 days with maintenance doses of 60 to 180 mg/kg of body weight twice a day (BID). A pharmacokinetic model was developed to predict the plasma concentrations obtained with different dosing regimens, and the model predictions were compared to the 50% effective concentration (EC50) of favipiravir against several viruses. Favipiravir pharmacokinetics were described by a model accounting for concentration-dependent aldehyde oxidase inhibition. The enzyme-dependent elimination rate increased over time and was higher in NHPs of Mauritian origin than in those of Chinese origin. Maintenance doses of 100 and 120 mg/kg BID in Chinese and Mauritian NHPs, respectively, are predicted to achieve median trough plasma free concentrations above the EC50 for Lassa and Marburg viruses until day 7. For Ebola virus, higher doses are required. After day 7, a 20% dose increase is needed to compensate for the increase in drug clearance over time. These results will help rationalize the choice of dosing regimens in future studies evaluating the antiviral effect of favipiravir in NHPs and support its development against a variety of HF viruses.

Ebola Virus Infection: Review of the Pharmacokinetic and Pharmacodynamic Properties of Drugs Considered for Testing in Human Efficacy Trials.

The 2014-2015 outbreak of Ebola virus disease is the largest epidemic to date in terms of the number of cases, deaths, and affected areas. In October 2015, no antiviral agents had proven antiviral efficacy in patients. However, in September 2014, the World Health Organization inventoried and has since regularly updated a list of potential drug candidates with demonstrated antiviral efficacy in in vitro or animal models. This includes agents belonging to various therapeutic classes, namely direct antiviral agents (favipiravir and BCX4430), a combination of antibodies (ZMapp), type I interferons, RNA interference-based drugs (TKM-Ebola and AVI-7537), and anticoagulant drugs (rNAPc2). Here, we review the pharmacokinetic and pharmacodynamic information presently available for these drugs, using data obtained in healthy volunteers for pharmacokinetics and data obtained in human clinical trials or animal models for pharmacodynamics. Future studies evaluating these drugs in clinical trials are critical to confirm their efficacy in humans, propose appropriate doses, and evaluate the possibility of treatment combinations.

Ebola virus dynamics in mice treated with favipiravir.

The polymerase inhibitor favipiravir is a candidate for the treatment of Ebola virus disease. Here, we designed a mathematical model to characterize the viral dynamics in 20 mice experimentally infected with Ebola virus, which were either left untreated or treated with favipiravir at 6 or 8days post infection. This approach provided estimates of kinetic parameters of Ebola virus reproduction, such as the half-life of productively infected cells, of about 6h, and the basic reproductive number which indicates that virus produced by a single infected cell productively infects about 9 new cells. Furthermore, the model predicted that favipiravir efficiently blocks viral production, reaching an antiviral effectiveness of 95% and 99.6% at 2 and 6days after initiation of treatment, respectively. The model could be particularly helpful to guide future studies evaluating favipiravir in larger animals.

Extension of NPDE for evaluation of nonlinear mixed effect models in presence of data below the quantification limit with applications to HIV dynamic model.

Data below the quantification limit (BQL data) are a common challenge in data analyses using nonlinear mixed effect models (NLMEM). In the estimation step, these data can be adequately handled by several reliable methods. However, they are usually omitted or imputed at an arbitrary value in most evaluation graphs and/or methods. This can cause trends to appear in diagnostic graphs, therefore, confuse model selection and evaluation. We extended in this paper two metrics for evaluating NLMEM, prediction discrepancies (pd) and normalised prediction distribution errors (npde), to handle BQL data. For a BQL observation, the pd is randomly sampled in a uniform distribution over the interval from 0 to the probability of being BQL predicted by the model, estimated using Monte Carlo (MC) simulation. To compute npde in presence of BQL observations, we proposed to impute BQL values in both validation dataset and MC samples using their computed pd and the inverse of the distribution function. The imputed dataset and MC samples contain original data and imputed values for BQL data. These data are then decorrelated using the mean and variance-covariance matrix to compute npde. We applied these metrics on a model built to describe viral load obtained from 35 patients in the COPHAR 3-ANRS 134 clinical trial testing a continued antiretroviral therapy. We also conducted a simulation study inspired from the real model. The proposed metrics show better behaviours than naive approaches that discard BQL data in evaluation, especially when large amounts of BQL data are present.