COVID-19: A need for new rather than repurposed antiviral drugs

Background SARS-CoV-2 infection, the causative agent of COVID-19, has resulted in over 2,500,000 deaths to date1. Although vaccines are becoming available, treatment options remain limited. Repurposing of compounds could reduce the time, cost, and risks associated with the development of new drugs and has been the focus of many clinical studies. Here, we summarise available evidence on 29 FDA-approved compounds, from in vitro results to clinical trials, focussing on remdesivir, galidesivir and favipiravir, and test 29 antiviral compounds’ activity in vitro. Methods A comprehensive search strategy was used to retrieve trials and publications related to antiviral compounds with potential efficacy to treat coronaviruses. These data were used to prioritise testing of a panel of antiviral drugs in vitro against patient isolates of SARS-CoV-2. An in vitro screen was carried out to determine the activity of 29 FDA-approved compounds. Results 625 clinical trials investigated 16 repurposed antiviral candidate compounds for the treatment of COVID-19. In vitro studies identified ten drug candidates with demonstrable anti-SARS-CoV-2 activity, including favipiravir, remdesivir, and galidesivir. To validate these findings, a drug screen was conducted using two cell lines and wildtype isolates of SARS-CoV-2 isolated from patients in the UK. While eight drugs with anti-SARS-CoV-2 activity were identified in vitro, activity in clinical trials has, as yet failed to demonstrate a strong effect on mortality. Conclusions So far, no repurposed antiviral has shown a strong effect on mortality in clinical studies. The urgent need for novel antivirals in this pandemic is clear, despite the costs and time associated with their development. Research in Context Evidence before this study Repurposing of existing compounds for the treatment of COVID-19 has been the focus of many in vitro studies and clinical trials, saving time, costs and risks associated with the research and development of new compounds. Added value of this study We reviewed the literature for 29 FDA-approved compounds with previously reported (or suspected) anti-SARS-CoV-2 activity and found 625 clinical trials that have been undertaken on 16 different drugs. We determined if repurposed antivirals are suitable for clinical trials based on previously published data, and conducted an additional in vitro screen using locally circulating strains in the UK (PHE2 and GLA1). We report the difference in IC50 from published data using Wuhan1/Wash1 strains with PHE2 and GLA1, including IC50 values below 100μM for galidesivir in wild-type virus. Given the limited success of repurposed compounds in the treatment of COVID-19, we comment on the urgent need for new antivirals specifically targeting SARS-CoV-2. Implications of all the available evidence Our data show that most prospective compounds for repurposing show no anti-SARS-CoV-2 activity, and antiviral activity in vitro does not always translate to clinical benefit. So far, no repurposed compound has shown a strong effect on mortality in clinical studies. Drugs, including monoclonal antibody therapies, that have been developed to target SARS-CoV-2 virus itself have shown most promise.


Introduction
The betacoronavirus SARS-CoV-2 is likely to be a significant public health threat to the global population in the medium to long term. This coronavirus family of RNA viruses are commonly associated with mild to moderate upper respiratory tract infections during the winter months 2,3 . However, the emerging coronaviruses MERS-CoV 4 , SARS-CoV-1 5 and SARS-CoV-2 6 have high mortality rates. In addition, long-term morbidity associated with SARS-CoV-2 infection, known as 'long COVID' 7 is a significant concern and highlights the need for intervention with antiviral therapies. The novel coronavirus SARS-CoV-2 originated in China in late 2019 and has spread across the globe, resulting in over 2,500,000 recorded deaths to date.
Repurposing of existing compounds for the treatment of COVID-19 has been the focus of many trials and studies, despite a lack of evidence regarding their effectiveness 8 . Repurposing of compounds is made possible by shared molecular pathways across diseases, which also overcomes bottlenecks in the developmental process by cutting the time, costs, and risks associated with drug development [9][10][11] . Previously developed antivirals may demonstrate activity against diverse viral families due to the presence of relatively conserved viral proteins such as DNA or RNA polymerases, integrase or viral proteases. Many viruses also require inactivation of host proteins, which could also provide targets for broad acting antivirals. Longstanding compounds have the advantage of pre-existing information regarding their pharmacokinetic and pharmacodynamic profiles and toxicity. However, repurposing drugs and compounds is not without caveats, and must occur in controlled, clinical trial settings. While a drug may be efficacious against the original target, the concentration required to combat another virus may be unachievable 12 due to toxicity or pharmacodynamic constraints in patients. Furthermore, there is a finite number of antivirals to trial, and emerging evidence is often conflicting. As a result, clinicians often make decisions in the absence of reliable data, based on preliminary reports and poorly conducted trials.
Herein, we summarise evidence on 29 FDA-approved candidate compounds, from in vitro results to clinical trials selected for potential SARS-CoV-2 antiviral activity. These potential antiviral candidates were next tested for activity against circulating SARS-CoV-2 variants in vitro.  Table 1). We then undertook a comprehensive search strategy to retrieve trials and publications related to antiviral drugs and compounds potentially effective in the treatment of coronaviruses. The method was adapted from that recommended in the Cochrane Handbook for Systematic Reviews of Interventions 13 for identifying trials. Searches for the below list of antiviral drugs and compounds were conducted between March 2020 and 28 January 2021 on ClinicalTrials.gov, the WHO International Clinical Trials Registry Platform (ICTRP) and the Cochrane COVID-19 Study Register. Search terms are shown in Supplementary Table 1. In addition, we searched the Chinese Clinical Trial Register and pre-print servers medRxiv, bioRxiv, and chinaRxiv, while relevant results through our academic network were also identified.
On the Cochrane COVID-19 Study Register, free text terms were used to search for the name of each drug or compound individually; no other search terms or limits were applied. This process was repeated by searching medRxiv and bioRxiv simultaneously through bioRxiv in the title or abstract fields, and for ChinaRxiv across all fields. For ClinicalTrials.gov and the Chinese Clinical Trial Register, each drug or compound was searched in the intervention field. Results were filtered for relevance before being exported and analysed in EndNote X9. Results were also screened to include only those published in English. No date or language limits were applied to the searches.
Weekly ICTRP reports were compiled and screened for trials mentioning COVID-19 or SARS-CoV-2 in their scientific or public titles, and their details were extracted using R version 4·0·0.

Cell lines
The methods of production and validation of the cell lines used in this study are fully described elsewhere 15 . In brief, VeroE6 (African green monkey kidney) cells obtained from Prof. Michèle Bouloy (Institut Pasteur, France) were induced to overexpress ACE2 using a modified lentiviral expression system (pLV-EF1a-IRES-Neo (Addgene, plasmid 85139)). A549 cells (human lung) expressing the NPro protein of bovine viral diarrhea virus (BVDV) were obtained from Prof. Rick Randall (University of St Andrews) and induced to overexpress ACE2 in the same manner as the VeroE6 cells.

Virus isolates and propagation
Two clinical isolates were used in the study, including the PHE2 isolate, provided by Public Health England, and a clinical isolate obtained from the MRC-University of Glasgow Centre for Virus Research, Scotland, UK (GLA1). GLA1 was obtained by inoculating a T-25 flask of VeroE6 cells with a bronchoalveolar lavage clinical sample and incubating the cells until CPE was observed. Both stocks were confirmed using metagenomic sequencing and expanded by inoculating a T-150 cm 2 of VeroE6 cells and incubating for 48 hours before clarifying the supernatant. A plaque assay was used to determine an accurate titre for each virus and ensure an accurate multiplicity of infection (MOI) for each isolate in subsequent experiments.

Endpoint Dilution Assay
An in vitro screen was performed to determine the activity of 29 FDA-approved compounds (see Table 1) on the ability of SARS-CoV-2 to infect and cause cytopathic effect (CPE) on the two cell lines derived from A549 and VeroE6 cells, as described above. A549-NPro-ACE2 and VeroE6-ACE2 cells were plated at a density of 1·25x10 4 cells/well in a 96-well plate in DMEM supplemented with 10% FCS the day before drug treating the cells. The media was replaced with 50ul DMEM-GlutaMAX (ThermoFisher, cat no 10566016) supplemented with 2% FCS (ThermoFisher, cat no A4766801) containing doubling dilutions of the various compounds at the stated concentrations. The cells were incubated with the compounds for three hours before the addition of 50ul of either SARS-CoV-2 PHE2 or GLA1 isolates at a MOI of 0·05 or 0·5. The concentrations presented in the results equate to the concentration of the compounds after the addition of the virus isolates. 72 hours post infection the cells were fixed with 8% (v/v) formaldehyde in PBS for one hour, washed with PBS and then stained with Coomassie solution (2ml Coomassie Brilliant Blue (ThermoFisher cat no. 20277), 900ml Methanol, 900ml water 200ml Acetic Acid) for 30 minutes. The plates were washed in water to remove the stain and dried overnight before using a Celigo Imaging Cytometer (Nexcelom Bioscience, UK) to measure the staining intensity. Percentage cell survival was determined by comparing the intensity of the staining to an uninfected well.

Statistical packages
All graphs were produced using Prism (GraphPad, version 8) with the IC50 values determined by a nonlinear regression using a sigmoidal 4PL equation.

Clinical trials with repurposed antivirals
The WHO's ITCRP database was reviewed for clinical trials examining the repurposing of 29 compounds for the treatment of COVID-19. Between March 2020 and January 2021, 655 such trials were registered, investigating sixteen drugs in mild to severe COVID-19 disease. For trials with registered details, median target size was 135 patients. Many trials were multi-national with multiple experimental arms. Iran had the highest number of clinical trials registered (n=145), followed by the US (n=72) (Figure 1). In Europe, Spain had the highest number of trials registered (n=40). 68 studies investigating drug repurposing in China, were initiated but concluded early due to a lack of patients. The number of trials remained low in Africa; however, one multicentre trial (PACTR202006537901307) investigated several antiviral compounds across 13 nations on the continent.
The most commonly repurposed inhibitor was HCQ (n=461, see Figure 2), a known anti-malarial and blocker of virus entry, which can be cheaply produced and dispensed around the globe. HCQ, lopinavir and ritonavir (the latter two are co-formulated) were investigated early in the pandemic (up to July 2020) but have now been removed from trials in hospitalised patients following data from the WHO's Solidarity trial in which they were shown to have no effect on mortality [16][17][18] .
Data on the antiviral drug remdesivir, previously tested in patients with Ebola virus disease were initially encouraging, with evidence that time to recovery was reduced in hospitalised patients 19 . However, these results were not replicated in the Solidarity trial 17,20 , which also found no effect of HCQ, lopinavir/ritonavir or interferon on overall mortality, initiation of ventilation or duration of hospital stay in hospitalized patients. These conflicting results may reflect the need to initiate antiviral treatment early before the onset of pathology driven by the immune response.
Favipiravir was the subject of 73 trials. Two trials have reported a faster time to viral clearance in favipiravir treated patients compared to controls ( 21 , FUJIFILM Toyama Chemical press release). Two trials are ongoing in the UK: Getafix 22 and Pioneer 23 . The first clinical trial assessing galidesivir in COVID-19 commenced in April 2020 14 (NCT03891420) and is expected to be completed by May 2021.
Galidesivir, remdesivir and HCQ were able to increase the survival of VeroE6-ACE2 when infected with the GLA1 and the PHE2 isolates (Supp figure 2, Supp figure 3). Lopinavir, ribavirin and favipiravir reduced cell death in both cell lines to a small degree at the highest concentrations, but this coincided with a cytotoxicity, with the latter two only showing inhibition at the highest concentrations of the drugs tested against an MOI of 0·05, with little effect on the virus at an MOI of 0·5. The same effect was seen with emetine and umifenovir with both increased cell survival but at concentrations that also lead to greater cell cytotoxicity than seen in lopinavir, ribavirin and favipiravir. The compounds were then tested on A549-NPro-ACE2 cells, which constitute a more physiologically relevant human cell line that is susceptible to SARS-CoV-2 virus infection. The results were very similar to those seen in VeroE6-ACE2 cells apart from favipiravir, which had no apparent effect ( Figures 4). Again, ribavirin showed an ability to increase cell survival against the PHE2 virus but not the GLA1 isolate (Supp figure 1 and Figure 4, respectively).
IC50 values were determined for the three most active compounds against the PHE2 and GLA1 viral isolates on t VeroE6-ACE2 and A549-NPro-ACE2 cells (  Prior to clinical trials, the efficacy of candidate therapies is ascertained by antiviral activity in vitro. Here, 29 FDA-approved nucleoside analogues, protease inhibitors, influenza antivirals and other compounds were investigated for anti-SARS-CoV-2 activity in vitro, of which three demonstrated significant antiviral activity. Both the EC50 (the concentration of a drug that gives half-maximal response) and the IC50 (concentration of an inhibitor where the response is reduced by half) have been widely reported for these compounds in the literature, as indicators of potency, but do vary in their values. These values depend on the conditions of the experiment, such as MOI, cell type and lineage as well as incubation time. On review of in vitro and clinical studies, antiviral activity in cells was not found to necessarily translate to benefit in patients. For example, while HCQ and lopinavir/ritonavir showed strong SARS-CoV-2 activity in cells, they failed to lower mortality rates in hospitalised COVID-19 patients in trials and their use has been discontinued for this patient group [16][17][18] . Furthermore, some contraindications will require a detailed patient history and continued monitoring, which in milder cases may not be possible 49 . For example, some drug candidates cause a prolonged QT interval and/or arrhythmias, or liver and kidney impairment. It is therefore crucial that such studies are carried out in robust blinded clinical trial settings to fully assess the impact of a given drug on COVID-19 outcomes.
A good example of this is HCQ which showed promise in early small-scale trials but was found to cause more harm than benefit, as well as no effect on COVID-19 outcomes 17,18 by large-scale trials. This may be related to difficulty in achieving adequate doses in vivo.
The anti-viral remdesivir have been shown to improve patient outcomes in one study, although evidence on the latter remains conflicting and no study has shown a significant effect on mortality 50 . Remdesivir (GS-5734) is an adenosine nucleotide prodrug and a broad-spectrum antiviral developed by Gilead, originally used for treatment of Ebola and Marburg virus infections. As an intravenous drug that interferes with viral RNA-dependent RNA polymerase, remdesivir is most effective during early stages of infection, like most antivirals 51,52 19 ). However, preliminary results from the WHO's Solidarity trial found no effect on mortality or hospital stay 56 . Similarly, a smaller randomised, double-blind, placebo-controlled multicentre trial study in China (n=237) found no benefits associated with remdesivir and reported that treatment was stopped early because of adverse events in 18 (12%) patients (NCT04257656) 57 .
Like remdesivir, galidesivir (BCX4430, BioCryst) is a novel synthetic adenosine analogue developed for the treatment of Ebola virus and Marburg virus disease. Galidesivir is a broad-spectrum antiviral given intravenously and has activity against several other viruses, including bunyaviruses, arenaviruses, paramyxoviruses, coronaviruses and flaviviruses. 42 Phase I studies found that it was generally safe and well-tolerated in healthy volunteers. Following promising results against coronaviruses in HeLa cells (MERS-CoV EC50=68·5 μM and SARS-CoV EC50=57·7 μM) 42 and a molecular docking study showing that galidesivir can bind tightly the RdRp of SARS-CoV-2 58 , a randomised, double-blind, placebo-controlled clinical trial has commenced (NCT03891420) in Brazil to consider galidesivir for treatment in COVID-19 and yellow fever. Recently, one study reported EC50 values under 100 μM for galidesivir, but the experiments used cell culture-adapted SARS-CoV-2 43 . Here, we report IC50 values for galidesivir against two isolates of SARS-CoV-2 demonstrating an IC50 of between 16·45 to 106.5μM, depending on the viral isolate, MOI and cell line, which is within the same activity range previous demonstrated for SARS-CoV and MERS-CoV.

The anti-influenza drug favipiravir (T-705) is also being investigated for treatment of SARS-CoV-2.
Favipiravir is a pyrazinecarboxamide derivative and a pro drug that is converted to a nucleoside analogue within the cell of interest, followed by a series of phosphorylations to produce the active ribofuranosyltriphosphate T-705RTP, which selectively inhibits viral RNA-dependent RNA polymerases 59 .

The way forward.
There is an urgent need for the development of novel antivirals to treat COVID-19 and tackle the pandemic. Since antiviral drugs are most likely to be effective early in the disease process, what are ideally required are antiviral therapies that can be taken orally and are well enough tolerated to be widely available for use in the community very soon after onset of symptoms. Currently the drug which has closest to this product profile is Favipiravir, which has had success in early clinical trials 60 but had no antiviral activity in our hands in vitro. A bespoke small molecule with a lower EC50 than Favipiravir might have a greater chance of success.
An alternative direction for the development of antiviral therapeutics is anti-SARS-CoV2 monoclonal antibodies (mAbs). For example, the S protein blockers casirivimab/imdevimab, also known as REGN10933/REGN10987 64 and Bamlanivimab (LY-CoV555) 65 which have both been issued with emergency use authorisation by the FDA, for use in non-hospitalised patients not requiring oxygen. Also under investigation in animal models, is CT-P59, which targets the receptor binding protein of SARS-CoV2 66 . Casirivimab/imdevimab showed significant reductions in virus levels and were associated with significantly fewer medical visits 64 . Bamlanivimab in combination with etesevimab decreased SARS-CoV2 log viral load by day 11 (BLAZE-1 study) 65 .
The EC50s for mAbs are promising (e.g. 31pM for casirivimab/imdevimab 67 , however, mAb preparations have some disadvantages that will limit their use. They have to be administered intravenously, and adverse events can include serious hypersensitivity reactions and anaphylaxis. These attributes preclude the possibility of using them as community-based therapies. Further, monoclonal antibodies tend to select for escape viral variants, which may limit their useful life span. Using monoclonal antibodies in combination is an attempt to limit escape viral variants. For example, escape variants have been documented when using either casirivimab or imdevimab individually, but so far not when used in combination 67 .
Repurposed compounds have been evaluated in hundreds of studies with limited clinical success, there remains an urgent need both for the development of novel antivirals to tackle severe disease and for a product that is safe enough and logistically practical for use in the critical early days of the disease process. It is time to think beyond the immediate emergency and fund the development of bespoke drugs for long-term solutions.

Acknowledgements
We

Declaration of interests:
The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. JTS is a chief investigator on the GETAFIX trial (ISRCTN31062548), other authors have declared no competing interests.

Supplementary Material
Supplementary Table 1 Terms used for searching the ICTRP database. Note that hydroxychloroquine (HCQ) and favipiravir were often misspelled, hence the more flexible terms.

Terms used in ICTRP database search
SARS-CoV-2 "covid19", "covid-19", "sars-cov 2", "sars-cov-2", "19-ncov" Drugs "atazanavir", "abacavir", "amantadine","AZT", "arbidol", "adefovir", "baloxavir", "danoprevir", "paritaprevir", "telbivudine", "sotetsuflavone", "sofosbuvir", "darunavir", "didonasine", "emetine", "emtricitabine", "favi*", "galidesivir","gemcitabine", "hydoxychloroquine", "hydroxychloroquine", "hcq", "fosamprenavir", "lamivudine", "lopinavir", "ritonavir", "oseltamivir", "remdesivir", "ribavirin", "saquinavir","tenofovir", "umifenovir" Should not be given with abnormally low neutrophil counts (<0.75 x 109/litre) or abnormally low haemoglobin levels (<7.5 g/decilitre or 4.65 mmol/litre). New born infants with hyperbilirubinaemia requiring treatment other than phototherapy, or with increased transaminase levels of over five times the upper limit of normal. Viiv, Retrovir, Summary of Product Characteristics Supplementary Figure 1 Assessment of SARS-CoV-2 inhibition by active FDA approved antiviral compounds. A549 cells, overexpressing Ace2 and BVDV Npro protein, were treated for three hours with varying concentrations of antiviral compounds before the addition of SARS-CoV-2 PHE2 isolate at an MOI of 0·05 and 0·5. The cells were fixed with 8% formaldehyde 72 hours post-infection and stained with Coomassie blue. Percentage cell survival was determined by comparing the intensity of the staining to uninfected wells. The data shown are representative figures of 3 independent repeats. The mean percentage cell survival of the DMSO control infected with an MOI of 0·05 and 0·5 was 51.08% (SD 1·117) and 47·35% (SD 0·764), respectively. The cytotoxicity of the compounds is depicted by the grey line and were determined by comparing the intensity of staining to untreated wells. Figure 2 : Assessment of SARS-CoV-2 inhibition by active FDA approved antiviral compounds. VeroE6 cells, overexpressing Ace2, were treated for three hours with varying concentrations of antiviral compounds before the addition of SARS-CoV-2 GLA1 isolate at an MOI of 0·05 and 0·5. The cells were fixed with 8% formaldehyde 72 hours post-infection and stained with Coomassie blue. Percentage cell survival was determined by comparing the intensity of the staining to uninfected wells. The data shown are representative figures of 3 independent repeats. The mean percentage cell survival of the DMSO control infected with an MOI of 0·05 and 0·5 was 50% (SD 0·502) and 43·45% (SD 0·642), respectively. The cytotoxicity of the compounds is depicted by the grey line and were determined by comparing the intensity of staining to untreated wells Supplementary Figure 3 Assessment of SARS-CoV-2 inhibition by active FDA approved antiviral compounds. VeroE6 cells, overexpressing Ace2, were treated for three hours with varying concentrations of antiviral compounds before the addition of SARS-CoV-2 PHE2 isolate at an MOI of 0·05 and 0·5. The cells were fixed with 8% formaldehyde 72 hours post-infection and stained with Coomassie blue. Percentage cell survival was determined by comparing the intensity of the staining to uninfected wells. The data shown are representative figures of 3 independent repeats. The mean percentage cell survival of the DMSO control infected with an MOI of 0·05 and 0·5 was 28·83% (SD 0·673) and 27.52% (SD 0·193), respectively. The cytotoxicity of the compounds is depicted by the grey line and were determined by comparing the intensity of staining to untreated wells