Skip to main content
SearchLoginLogin or Signup

Review of "Favipiravir,lopinavir-ritonavir or combination therapy (FLARE): a randomised, double blind 2x2 factorial placebo-controlled trial of early antiviral therapy in COVID-19"

Reviewer: Bumi Herman (Chulalongkorn University College of Public Health Sciences) | 📒📒📒 ◻️◻️

Published onJun 01, 2022
Review of "Favipiravir,lopinavir-ritonavir or combination therapy (FLARE): a randomised, double blind 2x2 factorial placebo-controlled trial of early antiviral therapy in COVID-19"
1 of 2
key-enterThis Pub is a Review of
Favipiravir, lopinavir-ritonavir or combination therapy (FLARE): a randomised, double blind, 2x2 factorial placebo-controlled trial of early antiviral therapy in COVID-19
Description

AbstractBackgroundEarly antiviral treatment is effective for COVID-19 but currently available agents are expensive. Favipiravir is routinely used in many countries, but efficacy is unproven. Antiviral combinations have not been systematically studied. We aimed to evaluate the effect of favipiravir, lopinavir-ritonavir or the combination of both agents on SARS-CoV-2 viral load trajectory when administered early.MethodsWe conducted a Phase 2, proof of principle, randomised, placebo-controlled, 2×2 factorial, double-blind trial of outpatients with early COVID-19 (within 7 days of symptom onset) at two sites in the United Kingdom. Participants were randomised using a centralised online process to receive: favipiravir (1800mg twice daily on Day 1 followed by 400mg four times daily on Days 2-7) plus lopinavir-ritonavir (400mg/100mg twice daily on Day 1, followed by 200mg/50mg four times daily on Days 2-7); favipiravir plus lopinavir-ritonavir placebo; lopinavir-ritonavir plus favipiravir placebo; or both placebos. The primary outcome was SARS-CoV-2 viral load at Day 5, accounting for baseline viral load. ClinicalTrials·gov: NCT04499677.FindingsBetween 6 October 2020 and 4 November 2021, we recruited 240 participants. For the favipiravir+lopinavir-ritonavir, favipiravir+placebo, lopinavir-ritonavir+placebo and placebo-only arms, we recruited 61, 59, 60 and 60 participants and analysed 55, 56, 55 and 58 participants respectively who provided viral load measures at Day 1 and Day 5. In the primary analysis, the mean viral load in the favipiravir+placebo arm had decreased by 0.57 log10 (95% CI -1.21 to 0.07, p=0.08) and in the lopinavir-ritonavir+placebo arm by 0.18 log10 (95% CI -0.82 to 0.46, p=0.58) more than in the placebo arm at Day 5. There was no significant interaction between favipiravir and lopinavir-ritonavir (interaction coefficient term: 0.59 log10, 95% CI -0.32 to 1.50, p=0.20). More participants had undetectable virus at Day 5 in the favipiravir+placebo arm compared to placebo only (46.3% vs 26.9%, odds ratio (OR): 2.47, 95% CI 1.08 to 5.65; p=0.03). Adverse events were observed more frequently with lopinavir-ritonavir, mainly gastrointestinal disturbance. Favipiravir drug levels were lower in the combination arm than the favipiravir monotherapy arm.InterpretationAt the current doses, no treatment significantly reduced viral load in the primary analysis. Favipiravir requires further evaluation with consideration of dose escalation. Lopinavir-ritonavir administration was associated with lower plasma favipiravir concentrations.FundingLifeArc, UK.

RR:C19 Evidence Scale rating by reviewer:

  • Potentially informative. The main claims made are not strongly justified by the methods and data, but may yield some insight. The results and conclusions of the study may resemble those from the hypothetical ideal study, but there is substantial room for doubt. Decision-makers should consider this evidence only with a thorough understanding of its weaknesses, alongside other evidence and theory. Decision-makers should not consider this actionable, unless the weaknesses are clearly understood and there is other theory and evidence to further support it.

***************************************

Review:

Overall: This study has a robust methodology, although some information is needed to clarify unclear issues, which might be essential to confirm some of my prior assumptions. The authors may disagree with some issues mentioned here and could provide rebuttals. 

1. The medication dose for each arm was mentioned in detail, with clear eligibility criteria. Furthermore, this study involved vaccinated individuals. The sample size estimation used a bigger effect size to estimate the sample size (0.9 difference of log copies), whereas the results showed a lower reduction. Therefore, it is essential to recruit more sample size.

2. Starting from line 90 regarding the participant's eligibility. I address a time-varying exposure where:

a. The symptomatic patients took the medication within seven days after ONSET, while

b. Asymptomatic patients tool the medication within 48 hours after onset.

However, only 1 participant was asymptomatic and allocated to the placebo group (Table 2). Perhaps this is insufficient to evaluate the effect of antivirus on all non-hospitalized patients. The finding of this study might be relevant to symptomatic non-hospitalized patients only. We could not also justify that symptoms are linear to the viral load. 

3. Did the study include any individual with suspected re-positivity or reinfection prior to enrollment? Or was it determined by the antibody found in the ELISA test? The author refers to paper no 20, and the antibody test was based on IgG. If the unvaccinated participants showed positive IgG antibodies, were these people included in the study? If yes, was there any effect of previous COVID care? 

4. Further reading indicates stratification or minimization for the allocation. Although minimization seems different from true randomization, table 1 illustrates a similar baseline in all arms. However, in Figure 2 supplementary figure 2, there was a slight difference in baseline viral load. In figure 2, either ITT or modified ITT population shows a lower value in the Favipiravir monotherapy group which is slightly advantageous. Nevertheless, the authors adjusted this issue to statistical analysis. 

5. Participants’ masking was performed, but I wonder what kind of place was given to the participants? 

6. Line 143 kindly explains the rationale for using saliva samples for viral load measurements instead of nasopharyngeal samples, aside from the reason it is more convenient to perform. The author should focus on the sensitivity and specificity of using saliva samples compared to other samples to detect viruses. Please consider some pertinent articles and add to the discussion.

7. As quantitative viral load measurement using viral culture and infectivity assay is limited due to logistic constraints, the cycle threshold was applied as a proxy to viral load for laboratory analysis. The first thing is based on which target gene the cycle threshold values were determined, N, S, ORF1b, RdRp? (I have limited knowledge of estimating copies from CT value using calibration curve conversion, whether using CT value of one target gene or all genes). Second, briefly justify that the CT value appropriately represents the actual viral load, irrespective of the heterogeneity of the procedure (as this is still the source of procedural and measurement bias). Third, what machines were used, and what was the limit of detection for the machine?

8. The statistical analysis plan is appropriate for measuring the effect adjusted by factors, including the possible variant and the antibody level. Supplementary figures also show the different baselines in a specific group. The presence of comorbidities should not be treated as a binary response. It is imperative to know whether these chronic diseases are well-controlled or not and adjust the analysis. Classification of symptom duration into five days could not identify the effect of receiving intervention each day. It is better if this study identifies the ideal day of receiving medication (whether it should be taken one day, two days after the onset, etc.). 

Comments
0
comment

No comments here

Why not start the discussion?