Efficacy and Safety of Favipiravir in Moderate COVID-19 Pneumonia Patients without Oxygen Therapy: A Randomized, Phase III Clinical Trial
The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) that causes coronavirus disease 2019 (COVID-19) has spread rapidly since first described in December 2019. As of May 10, 2021, more than 157,000,000 cases have been reported, with the death toll exceeding 3,200,000 . Under such circumstances, many drugs have been investigated against this emerging viral infection. However, it is hard to say that the therapeutic option for COVID-19 is sufficient.
Favipiravir, approved for use in Japan for novel or re-emerging pandemic influenza virus infections, selectively inhibits viral RNA-dependent RNA polymerase. Favipiravir demonstrates antiviral activity against a broad spectrum of RNA viruses as well as influenza viruses [2, 3]. The drug’s inhibitory effects against SARS-CoV-2 were first reported in an in vitro study . Previous reports suggested favipiravir may exert its antiviral activity against SARS-CoV-2 through a combination of chain termination, retarding RNA synthesis, and inducing lethal mutagenesis [5, 6]. The clinical efficacy of favipiravir was first reported by Cai et al. They reported significant efficacy of favipiravir in the time to SARS-CoV-2 clearance and the time to chest imaging findings improvement compared to lopinavir/ritonavir ; however, these findings were obtained by a non-randomized and unblinded design. Thus, the clinical efficacy of favipiravir has not been confirmed in a well-controlled trial. In addition to this issue, it has also been unclear what characteristics of patients would be suitable for favipiravir treatment. We decided to confirm the clinical efficacy of favipiravir in a randomized, placebo-controlled trial, and to find suitable characteristics of patients with COVID-19 for favipiravir treatment.
Trial Design and Oversight
A randomized, single-blind, placebo-controlled, parallel-group comparison design was adopted to assess the safety and effectiveness of favipiravir in COVID-19 patients. Due to limited knowledge regarding this treatment at the time of planning, an adaptive design was incorporated to re-estimate the sample size based on the findings from an interim analysis. The observational study for compassionate use of favipiravir for COVID-19 had already been initiated in Japan at the time of this trial . Therefore, many COVID-19 patients were able to access favipiravir treatment. To minimize any disadvantages to patients assigned to the placebo group, the assignment ratio was set at 2:1 in favor of the favipiravir group. Investigators were permitted to switch patients to rescue treatments in the event of “lack of efficacy”, defined as marked deterioration in patients’ chest images and a continuous downward trend in oxygen saturation levels (SpO2) during the 12 h before and after imaging. In these cases, late administration of favipiravir was permitted as a treatment option for patients in the placebo group. Considering the above, a placebo-controlled trial was considered ethically permissible. This trial plan was filed with Japanese regulatory authorities (IND #2019-7269). The protocol, informed consent form, and all other required documents were reviewed by an Institutional Review Board (IRB) of each trial site before initiation of the trial at that site. The trial was approved by all IRBs (please see supplementary material) and was conducted by the Helsinki Declaration (1964 and its later amendments) and Good Clinical Practice.
Since favipiravir is an oral antiviral agent, it is unsuitable for patients with dysphagia such as patients requiring oxygen therapy or with a disturbance of consciousness. Patients with moderate illness requiring inpatient treatment were therefore targeted for this trial. The inclusion criteria were: (1) male or female aged 20–74 years; (2) positive for SARS-CoV-2 based on a nucleic acid amplification test of a respiratory tract sample taken at enrollment, pulmonary lesions confirmed by chest imaging, and fever ≥ 37.5 °C; and (3) written, informed consent obtained from the patient. The main exclusion criteria were: (1) 11 or more days since onset of fever of ≥ 37.5 °C; (2) the infection episode was a relapse or reinfection; (3) SpO2<94% without oxygen therapy. Further details were described in the protocol attached in the electronic supplementary material.
Investigators disclosed information regarding the allocated drugs to the enrolled patient only to the minimal necessary stakeholders of the trial site to minimize bias. The information on the allocated drugs was not disclosed to the patients or the Central Committee, and the blinding was maintained throughout the trial. Investigators observed and recorded the patient’s clinical symptoms and vital signs at least twice a day (morning and evening) until the time of discharge and/or Day 28. Virological examination of SARS-CoV-2 using nasopharyngeal specimens and chest X-rays were performed every 3 days until the time of discharge and/or Day 28. Chest X-rays and SARS-CoV-2 qualitative tests were able to perform arbitrarily when the investigator evaluated the primary endpoints. Investigators submitted the chest X-rays used for evaluation to the Central Committee. Viral samples for qualitative tests were evaluated at each trial site by the method recommended by the National Institute of Infectious Diseases.
After confirming trial eligibility, central randomization was conducted. Patients assigned to the favipiravir group received favipiravir at 1800 mg per dose twice a day on Day 1 followed by 800 mg per dose twice a day from Day 2 for up to 13 days. This dosage was higher than the approved dosage in Japan for influenza. This is because the EC50 of favipiravir for the influenza A (H5N1) virus by neutral red uptake assay ranged from 0.4–1.9 µg/mL , while the EC50, for SARS-CoV-2, was 9.72 µg/mL , which was higher. It has been reported that when the same favipiravir dose as in the present trial was administered to Japanese patients with severe fever with thrombocytopenia syndrome (SFTS), the mean trough concentration reached 40 µg/mL . Therefore, this dosage was adopted in the trial. Patients assigned to the placebo group received matching placebo tablets for up to 14 days. Inpatient management was mandatory throughout study drug treatment. The use of interferon-alpha, any drug reported to have an antiviral effect against SARS-CoV-2 (remdesivir, hydroxychloroquine sulfate, chloroquine phosphate, lopinavir/ritonavir, ciclesonide, nafamostat mesylate, camostat mesylate, nelfinavir mesylate, and ivermectin), and any type of hemofiltration therapy was prohibited.
At the time of this trial, the Japanese regulatory authorities stipulated that viral qualitative tests by nucleic acid amplification could be performed the day after improvement in fever and respiratory symptoms, and that, if negative results were obtained for two consecutive days, COVID-19 patients could be discharged. Concerning this criterion, the primary endpoint was a composite outcome defined as the time to improve in four clinical parameters (temperature, SpO2, findings on chest imaging, and viral clearance). SARS-CoV-2 qualitative testing using nucleic acid amplification of samples obtained at least 24 h after improvement in the other clinical parameters (temperature, SpO2, and findings on chest imaging) were documented. Improvement was defined as follows: (1) improvement in temperature was defined as axillary temperature falling to ≤ 37.4 °C and remaining at ≤ 37.4 °C for at least 24 h (temperature recordings taken within 4 h after use of an antipyretic were excluded); (2) improvement in SpO2 was defined as SpO2 remaining ≥ 96% for at least 24 h without the use of oxygen therapy; (3) improvement on chest imaging was defined as improvement in chest imaging findings taken at least 24 h after the previous image judged to be the worst, and (4) recovery to SARS-CoV-2-negative was defined as two consecutive negative results on qualitative tests by nucleic acid amplification separated by at least 24 h. If the above definition was met, the time to become negative for the first time in a qualitative test was defined as the improvement time in the primary endpoint. In other words, if all four clinical parameters were alleviated, the patient was judged to be able to be discharged.
In addition to the primary endpoint, several secondary endpoints were also evaluated. Regarding safety assessment, all adverse events (AEs) which occurred during 28 days and were reported by investigators were tabulated.
Based on the preceding reports , it was estimated that the time to symptom improvement would be 4 days shorter in the favipiravir group. Assuming that the survival function of days to improvement follows an exponential distribution, it was determined that 96 patients (64 favipiravir, 32 placeboes) would be sufficient to demonstrate the statistical superiority of favipiravir by the log-rank test with α = 0.05 (two-sided) and statistical power of 80%. The protocol stipulated that the sample size could be re-estimated based on the results of an interim analysis by the Central Committee. The interim analysis was held when data from ≥ 45 patients had been obtained, as stipulated in the protocol. The Committee biostatistician recommended that a sample size of at least 144 patients (96 favipiravir, 48 placeboes) be adopted, assuming a study dropout rate of 10% and patients' disease recovery rate before treatment initiation of 25%. The final target sample size was therefore set to 144 patients (96 favipiravir, 48 placeboes).
Independent Central Committee
Operational bias associated with a single-blind design was minimized by results being re-assessed by the Central Committee under blinded conditions. Regarding the re-assessment of chest imaging findings, the Committee identified the worst images from the extent and density of the lesions on chest X-ray images submitted by the investigator with the consensus of all members. In the event of a discrepancy between the Committee and the investigator, the opinion of the Committee was adopted. The Committee members included five external experts: four physicians including a radiologist, and one biostatistician. The Committee was assigned three duties: (1) re-assessing the primary endpoint; (2) assessing the validity of a judgment on the lack of efficacy; and (3) re-estimating the target sample size. To minimize evaluation bias, the re-estimated sample size was not disclosed to all investigators and the Committee members other than the biostatistician. The Committee gathering was held three times. Upon re-assessing the primary endpoint, the Committee members were not provided with data other than body temperature, SpO2, SARS-CoV-2 qualitative tests, and chest imaging findings related to the primary endpoint.
For the analysis of the primary endpoint, the log-rank test based on the weighted Z statistic  was used to control the Type 1 error before and after the interim analysis. Hazard ratios (HRs) and their 95% confidence intervals (CIs) were also calculated using the Cox proportional hazards model with sex and age at baseline as covariates. Lack of efficacy cases was censored on Day 28. Longitudinal data such as the patient status were assessed using a mixed-effects model for repeated measures. The differences in the least-square means between treatments and the 95% CIs were then calculated and compared. SAS software v.9.4 (SAS Institute, Cary, NC, USA) was used for these analyses. All statistical tests were two-tailed. p values less than 0.05 were considered significant.
A total of 156 patients were randomized (Fig. 1). All patients who received the study drug were included in the analyses. Eighty-five patients in the favipiravir group and 34 patients in the placebo group completed the study drug treatment, of which 81 patients in the favipiravir group and 33 patients in the placebo group completed the study. The investigator documented a lack of efficacy in 2 patients in the favipiravir group and 10 patients in the placebo group; however, the Central Committee overruled the judgment of a lack of efficacy in 1 patient in the placebo group. Seven patients in the placebo group were switched to treatment with favipiravir during Days 2–8 due to a lack of efficacy. Of the 156 patients, the proportion of males assigned to the favipiravir was relatively higher. Mean age did not differ between the two groups, but the proportion of patients aged ≥ 65 years was higher in the placebo group. Regarding NEWS, the proportion of high/medium-risk patients was higher in the favipiravir group. The two groups were otherwise well balanced (Table 1). The mean SpO2 before treatment initiation was 96% and more in both groups, as patients with COVID-19 pneumonia not requiring oxygen therapy were included.
Favipiravir appeared to shorten the time until clinical improvement by approximately 3 days in patients with moderate COVID-19 pneumonia presenting with SpO2 ≥ 94% as per the analysis plan. The result was mostly in line with the results in previous unblinded trials. Of the individual COVID-19 signs and symptoms, chest imaging, viral clearance, headache, myalgia or arthralgia, and fatigue or tiredness improved significantly earlier with favipiravir treatment. Furthermore, favipiravir showed significantly higher HRs in patients with early-onset or known risk factors for severe illness. At the same time, it should be noted that favipiravir treatment resulted in a significantly higher incidence of AEs. Although some limitations exist in this trial, the result indicates that favipiravir might be one of the options for moderate COVID-19 pneumonia treatment. The risk of AEs, including hyperuricemia, should be carefully considered when using it.