Study design

This randomized, controlled, open-label, phase 2 clinical trial ‘Pharmacokinetics, Tolerability and Safety of Favipiravir and Ribavirin for the Treatment of Lassa Fever (SAFARI)’ was conducted to assess favipiravir for the treatment of LF⁴². Favipiravir was evaluated as investigational drug, and ribavirin was used in the control arm as standard of care. Ribavirin treatment followed the so-called Irrua regimen. The Irrua ribavirin regimen (details provided below) was originally conceived at the Irrua Specialist Teaching Hospital (ISTH) and has since become the current standard of care in Nigeria and other endemic countries⁴³. It differs from the original so-called McCormick regimen by using a higher loading dose and subsequent lower total daily dose. Ribavirin is administered once daily in the Irrua regimen to reduce the biohazards for the healthcare personnel while providing constant drug exposure⁴⁴.

As per protocol, the primary endpoints were the description of classical pharmacokinetic parameters of favipiravir (for example, maximum plasma concentration, time to reach maximum plasma concentration, AUC, half-life, volume of distribution and so on) in patients with PCR confirmed LF and the proportion of drug-related adverse events and serious adverse events. The secondary endpoints were the mutagenicity of ribavirin and favipiravir measured via nucleotide exchange rate in individual virus genomes (deep sequencing of minor and major variants); the description of RNA concentrations, infectious titers and serological status during treatment; pharmacokinetic modeling and simulation of different loading regimens to characterize time to target concentration attainment; the assessment of covariates impacting drug exposure; and the relationship between drug exposure and viral elimination dynamics, length of hospital stay, mortality and blood component therapy use. Efficacy was included as a secondary endpoint in the study protocol as descriptive analysis of mortality in the respective treatment groups. Owing to the small sample size in phase 2, only preliminary observations can be made on the basis of the available data.

However, owing to limitations in the sensitivity of currently available protocols, no investigations of infectious titers, serological status during treatment or mutagenicity were performed. As neither ribavirin nor favipiravir are expected to induce dominant mutations but rather lead to an increase in the overall mutation rate, minor variant analysis is required. On the basis of the currently established sequencing protocols for LASV with the aim of analyzing the mutation rate, the current cut-off Ct value for the inclusion in the sequencing pipeline is the Ct value of 30. Samples with a lower RNA concentration are, at this stage, not analyzable for full genomes and will have insufficient sequencing depth to draw any meaningful conclusions about mutations occurring due to treatment. As the Ct values in the participant samples from this study were above the required threshold, the analysis of the drug-induced mutations has not been performed.

The two participating study centers ISTH, Edo, and Federal Medical Center of Owo (FMCO), Ondo, are located in the highest-transmission regions of Nigeria and serve as major national LF case management centers equipped with dedicated isolation wards for high-consequence pathogens. The study protocol was registered before study initiation (Clinicaltrials.gov NCT04907682 and PACTR 202010817169062) and has been included in the Supplementary Information. A DSMB composed of independent experts was involved in the oversight of this trial and advised for the safe conduct of the clinical trial. The study was conducted in compliance with the Declaration of Helsinki, the International Conference on Harmonization Good Clinical Practice guidelines and the Nigerian National Code for Health Research Ethics. Ethical and other applicable regulatory approvals were obtained before the start of the study from the National Health Research Ethics Committee, ISTH Health Research Ethics Committee, FMCO Health Research Ethics Committee, National Agency for Food and Drug Administration and Control and Ethik-Kommission der Ärztekammer Hamburg, Germany.

An amendment of the study protocol was submitted to and approved by the responsible ethics committees and regulatory bodies resulting from the DSMB’s recommendations following an unexpected case of PR prolongation in a participant treated with favipiravir. The amended version of the protocol was applicable from the fourth to the last participant. In this amended protocol, additional exclusion criteria were introduced, namely, medical history of clinically relevant cardiac disease, potassium level at screening −1 and PR interval ≥200 ms. Moreover, an ECG reading on day 3 for participants treated with favipiravir was added. As a precautionary measure, the administration of favipiravir had to be discussed with the medical monitor in case of PR interval measurements between 200 and 240 ms, and the drug had to be stopped if the PR interval was ≥240 ms. The amendment also included a reassessment of the risk profile of the high dose of favipiravir by the sponsor and the DSMB after ten treated participants or in the case of occurrence of an important AE. Finally, additional training on study-specific ECG readings, adequate potassium substitution in case of hypokalemia and management of potential cardiac conduction disorders were conducted. During the clinical trial, protocol deviations were identified and documented, and appropriate corrective actions were undertaken. These included the foremost inappropriate timing of procedures in individual participants, as described in Extended Data Table 7.

Eligibility criteria

After providing written informed consent, participants were screened for eligibility.

Inclusion criteria were as follows.

• Age ≥18 years

• LF confirmed by RT–PCR

• Written informed consent

Exclusion criteria were as follows.

• Inability to give consent (for example unconscious patients or cognitively impaired patients)

• Pregnancy/lactation (evidenced by negative urine pregnancy test in women of child-bearing potential)

• Women who plan to become pregnant within the upcoming 6 months

• Severe malnutrition (body mass index 

• Known intolerance to ribavirin or favipiravir

• History of hemoglobinopathies (that is, sickle-cell anemia or thalassemia major) and/or hemophilia

• Organ failure as indicated by:

  • Creatinine ≥3× upper limit of normal

  • AST (GOT) >150 IU/l

  • ACVPU scale score of V, P or U

  • Severe central nervous system features (for example, seizures, restlessness, confusion and coma)

  • O₂ saturation

  • Hematocrit

  • Severe anemia requiring blood transfusion

• Inability to take oral drug (for example, encephalopathy and severe vomiting)

• Patients who received ribavirin or favipiravir within the preceding 7 days

• A medical history of clinically relevant cardiac disease, potassium level −1 and PR interval ≥200 ms were added as exclusion criteria following a protocol amendment during the conduct of the clinical trial as described above

Study procedures

After confirming the eligibility to the study, participants were randomized to either the Irrua ribavirin regimen or oral favipiravir in a 1:1 ratio. The computer-generated randomization list was previously prepared by the sponsor with stratification per site, without restriction or blocking. Sealed opaque envelopes were prepared holding the treatment allocation. The envelopes were opened in ascending order upon confirmation of study eligibility and after recruitment into the clinical trial. No blinding of therapeutic regimens was performed after treatment allocation considering that study drugs presented distinct administration formulations.

Study participants were hospitalized in the isolation ward following RT–PCR diagnosis of LF until the 10-day treatment was completed. In the control group, participants received intravenous ribavirin (RIBAVIRIN, Jiangsu Ruinian Qianjin Pharmaceutical). The dosing schedule was 100 mg kg⁻¹ on day 1 (divided in two thirds initially and one third 8 h later, maximum 7 g day⁻¹), followed by 25 mg kg⁻¹ single dose on days 2–7 and 12.5 mg kg⁻¹ single dose on days 8–10. In the intervention group, participants received oral favipiravir formulated as 200-mg tablets (AVIGAN, Toyama Pharmaceuticals), in accordance with a previously reported regimen used for Ebola virus disease²¹. The dosing schedule was 6,000 mg on day 1 (divided in 3 doses, 2,400 mg, 2,400 mg and 1,200 mg), followed by 1,200 mg twice daily on days 2–10. Baseline data included demographic information, medical history, physical examination and signs and symptoms of LF. Sex was assigned following an external examination of body characteristics and self-report. Moreover, ECG was performed and repeated on days 0, 2, 4 and 10 and on day 3 for participants receiving favipiravir. Blood sampling for hematology, biochemistry, electrolytes and LASV RT–PCR with quantification of the Ct values were performed before treatment initiation and repeated every other day. ECGs were interpreted by the study physicians and reviewed by a cardiologist appointed by the trial sponsor. Ribavirin and favipiravir plasma concentrations were assessed by blood sampling before drug administration and on days 1, 2, 4, 6, 7, 8 and 10. Furthermore, plasma was collected after drug administration on day 1 (serial samples after 0.5, 1, 3, 5, 8 and 12 h) and on day 7 (serial samples after 1 and 4 h). Blood samples for phamacokinetic analysis were centrifuged, and plasma was stored at −80 °C within 2 h. Aliquots were shipped to Hamburg, Germany, according to UN2814 regulations, and specimens were inactivated in its biosafety level 4 laboratory for further pharmacokinetic analysis at the Department of Clinical Pharmacology at the Institute of Pharmacy, University of Hamburg. The schedule of study procedures is summarized in Extended Data Table 8.

The appearance of TEAEs, occurring after the first dose of the study drug, was systematically monitored and documented by the investigators during the study period. TEAE were assessed in terms of severity, seriousness, expectedness and association with the study drug. They were coded according to the Medical Dictionary for Regulatory Activities (MedDRA version 28.1) The assessment of association of event or laboratory test abnormalities was performed using the following criteria.

• Certain: plausible time relationship to drug intake; not explained by disease or other drugs; response to withdrawal plausible (pharmacologically and pathologically); event definitive pharmacologically or phenomenologically (that is medical disorder or recognized pharmacological phenomenon); rechallenge satisfactory, if necessary

• Probable/likely: reasonable time relationship to drug intake; unlikely to be attributed to disease or other drugs; response to withdrawal clinically reasonable; rechallenge not required

• Possible: reasonable time relationship to drug intake; could also be explained by disease or other drugs; information on drug withdrawal may be lacking or unclear

• Unlikely: time to drug intake that makes relationship improbable (but not impossible); disease or other drugs provide plausible explanations

Withdrawal from the clinical trial was done in case of withdrawal of consent or in case of important worsening of the health condition of the participant according to the judgment of the treating physician or in case of development of severe LF defined by the national guidelines. In this case, treatment was continued according to the standard of care with ribavirin. If a participant was still RT–PCR-positive at the end of the 10-day study follow-up period, treatment could be extended with ribavirin as recommended by the national guidelines⁴³.

Sample size determination

The sample size of 20 evaluable participants per arm as defined by the sponsor and the sampling schedule provided sufficient data points for the primary endpoint of characterization of the pharmacokinetics of investigated drugs. Sample size and sampling schedule calculations were performed using clinical trial simulations in NONMEM (v. 7.4, ICON Development Solutions). For pharmacokinetics, assuming prior information on ribavirin pharmacokinetics⁴⁵, the proposed sampling design allowed us to determine the structural pharmacokinetic parameters with low absolute relative bias (29,46,47. To evaluate the potential link between pharmacokinetics and pharmacodynamics (viral kinetics) in an exploratory analysis, the here-applied sampling schedule of viral load allowed to detect even weak exposure response relations (ribavirin-induced decline in viral load with a viral elimination half-life of 480 h compared with no effect assuming high interpatient variability in pharmacodynamic response of 70% and pharmacodynamic measurement error of 30%) at a statistical power of 99% with adequate accuracy (absolute relative bias

Statistical analysis

The statistical analysis was defined in a statistical analysis plan. Two study populations were defined. The per-protocol population included all participants randomized and followed-up according to the study protocol. This population was used for efficacy and pharmacometrics analysis. The modified intention-to-treat population included all participants that received at least one dose of the study drugs and was used for the analysis of safety and tolerability. Study data were aggregated for descriptive analysis according to the treatment arm. Quantitative data were described in medians and IQRs and categorical data as numbers and proportions. TEAEs whose association with the study drug was considered certain, probable/likely or possible were classified as drug-related TEAEs.

The objectives of the analyses on the population pharmacokinetics of ribavirin and favipiravir was predefined in the clinical trial protocol and statistical analysis plan following standard pharmacokinetic analysis processes, which were facilitated using NONMEM (Version 7.5, ICON) using first-order conditional estimation with interaction. For favipiravir, one- and two-compartment models with oral absorption and first-order or Michaelis–Menten elimination were tested as structural models. For ribavirin, one-, two- and three-compartment models were assessed as structural model. Interindividual variability and interoccasion variability was evaluated on the structural phamacokinetic parameters assuming a log-normal distribution. Body weight (including allometric scaling), body mass index, ideal body weight (equations of McCarron and Devine)⁴⁸, serum creatinine, bilirubin, uric acid and sex were evaluated as potential covariates on the phamacokinetic parameters. The pharmacodynamics of the study drugs on the LASV were described by a linear mixed-effects model. The increase of the Ct values of the GPC gene was modeled with an approach similar to Thielebein et al.⁴⁹. The parameter Ct₀ describes the baseline value at the time of the study inclusion, and α describes the increase of the Ct value per hour (Ct = CT₀ + α × t). Interindividual variation was evaluated on both model parameters (that is, Ct₀ and α) assuming a normal distribution. The M3 method^50,51 was used to handle Ct values above the upper limit of quantification. The potential influence of the categorical covariates sex, study drug (favipiravir or ribavirin) and study site on Ct₀ and α were tested, as prespecified in the statistical analysis plan. Further AUC, maximum serum concentration (C_max) and apparent clearance (CL/F) were tested on slope, using the pharmacokinetics of the drugs as continuous covariates. The same approach was used to model the Ct values of the L gene.

The (prediction-corrected) visual predictive checks, goodness-of-fit plots and the difference in the objective function value were used to guide the model selection. The likelihood ratio test (delta objective function value of −3.84, df of 1, α = 0.05) was used to compare nested models. For the comparison of non-nested models, the model selection was guided by the Akaike information criterion. For covariate testing on phamacokinetics and pharmacodynamics, the stepwise covariate modeling procedure was used using the likelihood ratio test (α = 0.05 during forward selection and α = 0.01 during backward elimination). Parameter uncertainty was quantified by a nonparametric bootstrap (n = 1,500 samples). The final favipiravir population pharmacokinetic model was used for Monte Carlo simulations assessing the time favipiravir exceeds the IC₅₀ (29.3 μM, equal to 4.60 mg l⁻¹), IC₉₀ (43.2 μM, equal to 6.79 mg l⁻¹) or IC₉₉ (69.6 μM, equal to 10.93 mg l⁻¹)²⁴. The study dosing regimen was simulated as well as different dosing regimen with loading doses ranging from 2,000 mg twice daily to 2,400 mg twice daily and maintenance doses ranging from 1,200 mg twice daily to 2,400 mg twice daily (Extended Data Table 5). The time above IC₅₀, IC₉₀ or IC₉₉ of unbound concentrations for each dosing regimen was calculated. Body weight was sampled from the mean and variance of the study population assuming a log-normal distribution. The fraction unbound was set to 0.456 (ref. ⁵²).

Missing data was treated as such in the safety analysis. However, in case of laboratory test results reporting measurements levels below the limit of detection, the imputation of half of the lower limit of detection was applied. In the case of results above the limit of detection, the imputation of upper limit of detection plus one was applied. In the case of missing values of the plasma concentration of the drugs, all nonmissing observations in the series were used. In the case that measured drug concentrations fell below the quantification limit of the bioanalytical methods, appropriate techniques, for example, the so-called M3 method, was used to handle these data points⁵⁰. In case of missing covariates, the population mean of this covariate was imputed for the analysis. Data collected from source documentation were recorded in REDCap (version 10.0.25). Data analysis was performed using STATA (version 17; StataCorp) and NONMEM (Version 7.5, ICON).

Ethics and inclusion statement

This study was conducted within the framework of a long-term partnership of the German and French institutions with the ISTH and the Federal Medical Centre Owo, Nigeria. These institutional collaborations have been established for more than 15 years and are based on an equitable partnership and clear capacity building plan. This equitable partnership forms the basis for the successful conduct of this first interventional clinical trial for next generation LF drugs since four decades. The governance of the clinical trial was shared between the Northern and Southern partners, with principal investigators and lead investigators represented by the Nigerian institutions. Both reference treatment centers are located in the hyperendemic LF regions of Nigeria; hence, they are the primary beneficiaries of the findings of this study, along with the population served by them. The study was conducted in compliance with the Declaration of Helsinki, the International Conference on Harmonization Good Clinical Practice guidelines and the Nigerian National Code for Health Research Ethics. Ethical and other applicable regulatory approvals were obtained from all study sites before the start of the study from the National Health Research Ethics Committee (NHREC/01/01/2007-14/09/2020), ISTH Health Research Ethics Committee (ISTH/HREC/20201711/128), FMCO Health Research Ethics Committee (FMC/OW/380/VOL.CX/131), National Agency for Food and Drug Administration and Control (NAF/DER/LAG/V&CT/RIBAVIRIN/2021/1) and Ethik-Kommission der Ärztekammer Hamburg, Germany (2020-10299-BO-ff).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.