Type I interferons (IFNs) constitute a major antiviral defense system of the body. They were discovered over 60 years ago by Isaacs and Lindenmann in the National Institute of Medical Research in London, initially as a biological activity in the supernatant of virally infected cells that could interfere with infection of new, previously uninfected cells (hence the name)[1], [2]. Their genes were cloned decades later [3], [4], and it’s now known that there are sixteen type I IFN subtypes in humans which include twelve IFN-α variants (encoded by 13 genes, with two genes encoding the same protein), IFN-β, IFN-ε, IFN-κ and IFN-ω [5]. A similar number of type I IFNs also exists in mice. IFN-α variants and IFN-ω are most similar biochemically and are longer-lived, low-affinity and found in the blood, while IFN-β is ubiquitous, shorter-lived, high-affinity and autocrine. IFN-ε and IFN-κ are predominantly expressed in reproductive and cutaneous tissues.

Type I IFN genes lack introns (with the exception of IFN-κ that has one) and are clustered on human chromosome 9 and mouse chromosome 4. All type I IFNs signal through the same heterodimeric receptor complex involving IFNAR1 and IFNAR2, and trigger the JAK/STAT downstream signaling pathway leading to the activation of the heterotrimeric transcription factor complex ISGF3, comprised of phosphorylated STAT1 and STAT2, and interferon regulatory factor 9 (IRF9), inducing the expression of hundreds of genes altogether termed IFN-stimulated genes or ISGs, some and perhaps many of which mediate cellular antiviral and more broadly antimicrobial defense mechanisms [5], [6]. In contrast, other ISGs regulate type I IFN activity. Type I IFNs have therefore taken center stage in studies addressing mechanisms of antimicrobial immunity, susceptibility to infection as well as immune-inflammatory responses to infection.

With the completion of the Human Genome Project, and the sequencing of the human genetic material, another type of interferon termed type III interferons or lambda interferons or interleukins(IL)− 28 and 29 were described, which exhibit remarkably similar properties to type I IFNs although they are expressed by different genes and signal through a different receptor [7], [8]. Thus, there are four type III IFN genes in humans encoding IFN-λ1 (IL-29), IFN-λ2 (IL-28A), IFN-λ3 (IL-28B) and IFN-λ4, and two in mice encoding IFN-λ2 and IFN-λ3. Of these, IFN-λ4 is a polymorphic pseudogene which, in some humans, encodes the interferon (IFN) lambda 4 protein whereas in others it has a frameshift mutation (the TT allele) that prematurely terminates the protein, producing non-functional truncated polypeptides [9]. Type III IFN genes are located on human chromosome 19 and mouse chromosome 7, and share a 5-exon gene structure with other IL-10 family cytokines with the exception of IFN-λ2 and IFN-λ3 that have an additional 6th exon.

Although type III IFNs share low homology with both type I IFNs and IL-10 cytokine family members, their induction mechanisms and downstream signaling mechanisms are reminiscent of those of type I IFNs [5], [10]. They are induced in response to infection and activation of pattern recognition receptors such as TLR3, TLR7 and TLR9 in the endosomes and RIG-I, MDA5 and cGAS in the cytosol, and trigger downstream signaling cascades involving the JAK/STAT pathway and leading to the activation of the ISGF3 transcription factor complex. This, in turn, triggers the induction of ISGs and antimicrobial immunity. This has complicated the unwinding of their unique non-redundant functions but emerging evidence now indicates important similarities but also functional differences between the two IFN systems.

Among the most obvious differences between type I and type III IFNs are their expression patterns. Although type I IFNs exhibit a promiscuous expression, with almost all cells of the body being capable of expressing type I IFNs and their receptor, the IFNλ system exhibits a much more restricted localization pattern, being mostly present at mucosal sites such as the respiratory, gastrointestinal and urogenital tract [10], [11]. Expression of IFNλs and their receptor among different cell types is also restricted, being mostly found in epithelial cells and some leukocyte populations. Interestingly, in the respiratory tract type III IFNs appear to be induced first upon infection, and to be the dominant IFNs in the upper airways [12].

In addition to this topological difference, there are important functional differences. Type I IFNs exhibit strong pro-inflammatory activity, activating neutrophils and driving inflammatory cytokine production, which is not seen by type III IFNs [12]. In side-by-side experiments, type I IFNs have been shown to induce pro-inflammatory neutrophil responses which are largely absent when type III IFNs are used [12]. Type III IFNs thus appear to be the front-line guardians of respiratory viral infections, whereas type I IFNs follow, strengthening antiviral responses at the expense though of collateral tissue damage. Type I IFNs also up-regulate DC antigen presentation and co-stimulation, effects that have not been described for type III IFNs, although both IFN systems can drive Th1 cell differentiation and T cell immunity [13], [14], [15]. Notably, type I IFNs exhibit many additional and broad effects in innate and adaptive immunity including macrophage and NK cell activation, B cell class switching and antibody production, and induction of T and B cell regulatory responses by virtue of their ability to signal in multiple cell types that express the type I but lack the type III IFN receptor [16], [17].

From their very early discovery type I IFNs raised hopes that they could be used as broad-spectrum antivirals, much in the same way broad-spectrum antibiotics are used to combat bacterial infections. However, this path to clinical development has faced many hurdles that have limited their use in this context. The emergence of the coronavirus disease 2019 (COVID-19) pandemic has renewed interest to the application of IFNs as antivirals and has led to important progress in that direction. Here, we review the basic biology of type I and type III IFNs in the context of COVD-19, focusing on the genetic and immunological evidence placing IFNs at the epicentre of susceptibility to SARS-CoV-2 infection, and we discuss the latest developments in their application in the clinic for treating SARS-CoV-2 infections but also potentially providing powerful weapons for any viral infections of epidemic or pandemic potential that may emerge in the future.

COVID-19 caused by SARS-CoV-2 infection has led to an unprecedented high incidence of pneumonia and acute respiratory distress syndrome (ARDS), and millions of deaths worldwide. However, it is only 10% of the cases that are hypoxemic (severe or critical), requiring oxygen support, and one third of them mechanical ventilation (critical). Efforts have therefore concentrated, from the beginning of the pandemic, at the unwinding of the mechanisms driving disease susceptibility. One of the most remarkable findings made in that direction was that in COVID-19 patients that develop severe disease type I IFN responses are impaired. There is deficient production and activity of type I IFNs early on which is in sharp contrast to their robust induction in patients with milder forms of the disease [18], [19], [20], [21]. Interestingly, plasmacytoid DCs which are distinguished for their capacity to produce copious amounts of type I IFN in the body in response to respiratory as well as other viruses [22], [23] also exhibit impaired responses [23].

The kinetics of type I IFN production matter. Robust but transient type I IFN responses close to disease onset, when SARS-CoV-2 infection and replication is still ongoing, are essential for antiviral protection while their induction later on is less, if at all, effective in that respect. Rather, delayed and/or persistent Type I IFN production may do more harm than good by boosting pro-inflammatory processes damaging the lung [24]. Support for that comes from experimental animal models where delayed type I IFN expression in influenza virus, SARS-CoV or MERS-CoV infection drives pro-inflammatory cytokine responses and exhibits a potent pathogenic role [12], [25], [26]. However, the extent to which this applies to humans is less clear. Although aberrant type I IFN responses driven by the activation of the cyclic GMP-AMP synthase (cGAS)–stimulator of interferon genes (STING) pathway have been implicated in lung destruction in COVID-19 patients [24], only a fraction of them exhibit increased type IFN levels at later timepoints of the disease process [19], [20], [27], [28], [29].

Similarly to type I IFNs, type III IFNs are also impaired in SARS-COV-2-infected patients with severe but not milder disease[19]. By comparison, in patients with flu infection with similar or even more severe clinical characteristics at hospital admission, type I and type III IFN levels are robustly produced, possibly accounting for the better disease outcome these patients exhibit [19]. Thus, although in severe COVID-19 patients type I and type III IFNs are impaired, there is robust production of pro-inflammatory cytokines and chemokines such as TNF, IL-1, IL-6, IL-7, IL-8, IL-10, CCL3 and CCL4, from the early to the late stages of the disease process [19], [20]. This indicates a paradox as pro-inflammatory responses normally follow the antiviral ones, a phenomenon that has led to the suggestion that inflammatory responses in COVID-19 are ‘untuned’, ‘imbalanced’ and ‘misfiring’ [18], [19], [20]. Untuned immune responses against infection lead to more extensive tissue damage and lung injury as previously demonstrated in experimental animal models [12], [30]. Type I and type III IFN deficiency, often followed by delayed type I IFN overproduction, is therefore central to the pathophysiology of COVID-19 (Fig. 1). There are several reasons for that as subsequently discussed. Fig. 2.

When defective type I and type III IFN responses are considered, the first thing that comes to mind is the possibility that SARS-CoV-2 expresses inhibitors that impair IFN production per se or suppress its effects. This is a well-known mechanism used by viruses to evade innate immunity and SARS-CoV-2 is no exception. Indeed, SARS-CoV-2 has developed multiple strategies to impede the IFN response and achieve efficient replication and spreading.

First, SARS-CoV-2, as other coronaviruses, has developed mechanisms to evade detection. It shields its viral dsRNA replication intermediates in double-membrane vesicles (DMVs) mediated by the virus proteins NSP3, NSP4, and NSP6 [31], [32], and modifies its viral RNA cap to a ribose 2′-O-methylation 5′-cap structure through the actions of NSP13, NSP12, NSP14, NSP10 and NSP16 [33], thus avoiding sensing by the PRRs RIG-I, MDA5 and TLR3. It also cleaves ISG15 and RIG-I through the proteolytic actions of NSP3 and NSP5, or the ubiquitination of RIG-I following the interaction of TRIM25 with the N protein of the virus [34], [35], [36], [37]. Second, SARS-CoV-2 antagonizes at various levels PRR downstream signaling which, following viral recognition, leads to IFN production. Several studies have thus reported inhibitory effects of NSP1, NSP3, NSP5, NSP6, NSP12, NSP13, NSP14, NSP15, ORF3a, ORF3b, ORF6, ORF7a, ORF7b, ORF8, ORF9b, N, and M in type I IFN induction [38], [39], [40], [41], [42].

Many of them target the IRF pathway. For example, NSP6 was found to bind TANK binding kinase 1 (TBK1) and suppress IRF3 phosphorylation, NSP13 to bind and block TBK1 phosphorylation, and open reading frame 6 (ORF6) to impair docking of karyopherin/importin complex, thereby disrupting nuclear translocation IRF3 transcription factors [39], [40]. M protein, in turn, was shown to interact with RIG-I, MAVS, and TBK1, to prevent the formation of the RIG-I, MAVS, TRAF3 and TBK1 multiprotein complexes, and thus inhibit the phosphorylation, nuclear translocation and activation of IRF3, thus reducing type I and III IFN production [43]. A similar effect was described for ORF9b [44]. Other viral proteins were found to antagonize type I IFN signaling by blocking of STAT1/STAT2 activation. These include NSP1, NSP6, NSP13, ORF3a, ORF6, ORF7b and M which were shown to inhibit STAT1 phosphorylation, NSP6, NSP13, ORF6, ORF7a, and ORF7b which inhibited STAT2 phosphorylation, and ORF6 and nucleocapsid (N) protein which suppressed both STA1 and STAT2 phosphorylation [40], [45], [46]. The NF-κB pathway may also be a target as ORF9b was shown to target NEMO and interrupt its K63-linked polyubiquitination, thereby inhibiting canonical NF-κB activation involved in IFN production [47]. There are therefore multiple mechanisms though which SARS-CoV-2 can inhibit IFN production, at least in vitro.

In addition to this, SARS-CoV-2 antagonizes the IFN response once induced, inhibiting IFN-α/β receptor signaling and ISG expression. This has consequences for its own production as well, as type I IFNs are known to boost their own gene expression through positive feedback mechanisms. Specifically, NSP14 promotes IFNAR1 lysosomal degradation, blocking the activation of the transcription factors STAT1 and STAT2 [48]. N protein, in turn, competitively binds to STAT1/STAT2 and interferes with their interactions with JAK1 and TYK2, thus blocking their phosphorylation and subsequent induction of ISGs [46]. ORF6 also blocks STAT1/2 nuclear translocation and ISG transcription of by interacting with the nucleopore complex Nup98 [45], [49]. ORF7 interferes with the host ubiquitin system to form K63-linked ubiquitin chains that inhibit STAT2 phosphorylation [50]. Thus, SARS-CoV-2 acts at multiple levels to suppress the induction of IFNs and reduce their biological effects [51].

Another possible explanation for the profound impairment of type I and type III IFN production in COVID-19 is the presence of genetic mutations or variants that reduce or abrogate their gene expression. There is strong precedent for that as such mutations, often referred to as inborn errors of immunity, have been shown to render otherwise healthy individuals particularly susceptible to viral or other microbial infections. In particular, loss-of-function mutations of TLR3 which senses viral or cellular double stranded RNA, IRF7 which is critical for basal and inducible type I IFN expression, and IRF9 which is an essential component of the interferon-stimulated gene factor 3 (ISGF-3) complex that mediates IFN signaling, have all been shown to affect type I IFN-dependent immunity and underlie life-threatening pneumonia caused by influenza virus, an RNA virus that bears similarities with SARS-CoV-2 [52], [53], [54].

This is also the case in COVID-19 as such loss-of-function variants have been strongly linked to critical or life-threatening pneumonia in SARS-CoV-2 infected individuals. These variants are functionally relevant as fibroblasts presenting autosomal dominant or recessive TLR3 deficiency or autosomal recessive IRF7 deficiency exhibited impaired type-I-IFN-dependent responses following SARS-CoV-2 infection in vitro [55]. Moreover, loss-of-function mutations of TLR7, an X-linked recessive deficiency, have been shown in about 1% of male patients with critical pneumonia, whereas they were absent in asymptomatic or mild COVID-19 patients. They confer a particularly elevated risk for severe or critical COVID-19 in male patients, with high penetrance [56]. Among the main functions of TLR7 is its ability to drive type I and type III IFN responses from plasmacytoid DCs (pDCs), the highest IFN producers in the body critically involved in sensing viral nucleic acids driving innate immune responses during infection. The observation, therefore, that in pDCs loss-of-function mutations of TLR7 severely impair IFN responses to SARS-CoV-2 virus in vitro suggests a major way by which such mutations drive disease susceptibility in COVID-19. Notably, residual IFN responses remain in TLR7-deficient pDCs, indicating that TLR9 is also involved as inactivating mutations of UNC-93B- and IRAK4, downstream components of both the TLR7 and TLR9 signaling pathways, have been shown to completely abrogate the ability of pDCs to produce IFNs [57]. Other genes that have been linked to IFN deficiency include TLR3, TICAM1, TBK1, IRF3, IFNAR1 and IFNAR2 [22]. Again, fibroblasts presenting autosomal dominant or recessive TLR3 deficiency, autosomal recessive IRF7 deficiency or autosomal recessive IFNAR1 deficiency displayed defective type-I-IFN-responses against SARS-CoV-2 in vitro [55].

In the same direction, large-scale GWAS on a large number of COVID-19 patients with severe disease have identified several genetic loci around IFN-related genes that are linked to severe COVID-19. These include 21q22.1 (rs2236757) which is within the IFNAR2 gene, and 19p13.2 (rs74956615) which is near the gene that encodes tyrosine kinase 2 (TYK2), an essential component of IFN signaling [58]. They also include 12q24.13 (rs10735079) that harbors IFN-inducible genes encoding for antiviral 2’,5’-oligoadenylate synthetase (OAS) enzymes, OAS1, OAS2, and OAS3 which generate short oligonucleotides that act as second messengers to activate the latent form of ribonuclease L (RNaseL) [58]. In its activated form, RNase L cleaves all RNA molecules in the cell leading to cell death and release of antiviral stress granules that further trigger IFN production and amplify the IFN response [59]. However, inherited defects of the OAS-RNASEL pathway do not seem to underlie COVID pneumonia (not even mild pneumonia) in adults whereas they are critically involved in driving multisystem inflammatory syndrome in children (MIS-C), a rare but severe condition that follows benign COVID-19 [60]. This is related to exaggerated type I IFN and pro-inflammatory responses unleashed in these patients. Altogether, these findings support a clear involvement of host genetic dysfunctions in the induction of the IFN pathway and its effects in COVID-19 [61].

Although viruses have evolved strategies to suppress IFN responses and thus evade antiviral immunity, this does not explain why older individuals, and individuals with comorbidities, are the most susceptible to severe COVID-19 while younger people are largely protected. Similarly, although a substantial number of genetic mutations that can inhibit or prevent type I and type III IFN production have been identified, and more will be recognized in the future, these can only account for a fraction of the individuals that present with deficient type I and/or type III IFN production, and again do not explain the high incidence of severe disease in the elderly population.

In a surprising development, it was discovered that at least 10% of patients with critical COVID-19 pneumonia had high titers of pre-existing neutralizing autoantibodies against type I IFN-α2 and IFN-ω [62]. This was widely replicated in various other populations [63], [64], [65], [66], [67], [68], [69], [70], [71], [72], [73]. These autoantibodies were not detected in asymptomatic or mild patients nor in healthy individuals, and could neutralize supraphysiological concentrations of 100 ng/ml of these cytokines in plasma. When this analysis was refined to neutralize lower more physiologically relevant concentrations of up to 1 ng/ml of IFN-α and/or IFN-ω in the blood [19], autoantibodies were detected in at least 15% of critical cases of COVID-19% and 20% of critical cases of patients > 80 years old [74]. The proportion of such patients increased after the age of 65 years and was greater in men than women. Moreover, another 1% of patients had auto-antibodies neutralizing 100 ng/ml IFN-β in plasma. Altogether, about 20% of all COVID-19 deaths, across all ages, were in patients that had IFN auto-antibodies [62], [74], [75].

It is noteworthy that neutralizing autoantibodies to type I IFNs were present in the general population. In a large study of more than 34,000 individuals from the general population aged 18 to 100 years, the prevalence of auto-antibodies neutralizing 100 ng/ml (or 1 ng/ml) of IFN-α or IFN-ω was higher in men than in women, and increased with age, with 0.17% (1.1%) of individuals being positive for these antibodies before the age of 70 years, and more than 1.4% (4.4%) positive after the age of 70 years [74]. On the contrary, auto-antibodies neutralizing IFN-β have a similar prevalence in all of the age groups tested. This notable distribution probably contributes to the higher risk of death from COVID-19 in the ageing population. Interestingly, autoantibodies neutralizing type I IFNs were also found to underlie over 20% of breakthrough cases of COVID-19 hypoxemic pneumonia in already vaccinated individuals, highlighting the major importance type I IFNs and innate immunity play [76]; however, their contribution to hypoxemic breakthrough cases in vaccinated people remains unknown.

Autoantibodies to IFN-λ also exist in both COVID-19 patients and the healthier population and are more prevalent than these against type I IFNs. However, they are rarely neutralizing, and do not appear to predispose individuals to severe COVID-19 pneumonia themselves [77]. Therefore, IFN-λ autoantibodies do not constitute a major risk factor for the development of severe COVID-19, in sharp contrast to type I IFNs. Nevertheless, they might operate as modifiers, worsening the clinical outcome of patients bearing stronger determinants of disease with incomplete penetrance (e.g. patients with neutralizing auto-antibodies to both type I and III IFNs).

Type I IFNs have a long history as potential therapeutics by virtue of the antiviral, antiproliferative and immunomodulatory activities they exhibit. Their ability, in particular, to interfere with viral infections non-discriminately raised hopes, soon after their discovery, that they could be used as antivirals for a wide range of infections. Although this did not materialize to the extent initially anticipated, several IFN formulations are now approved for viral infections such as hepatitis B and C as well as certain forms of cancers and autoimmune diseases. These are based only on a few subtypes of type I IFNs, mainly IFN-α2 and IFN-β, and include both shorter acting interferons which are often given several times a week, and longer acting formulations that are administered parenterally only once a week or even less. These are also made by a variety of drug companies and are often branded under several names.

IFN-α2 was the first Type I IFN subtype to be characterized in the early eighties, and exists in different allelic variants. Among them, IFN-α2a and IFN-α2b which differ from each other only at position 23 (lysine in IFN-α2a, arginine in IFN-α2b), are the main ones that have been used in pharmaceutical development [78], [79], [80].

Synthetic versions of IFN-α2a are approved for the treatment of chronic hepatitis B infection in adults, chronic hepatitis C infection in people above the age of 5, chronic hepatitis C and HIV co-infection in adults and chronic hepatitis C infection in combination with ribavirin [81], [82], [83]. They are also approved for the treatment of cancers such as hairy cell leukemia in adults, chronic phase, Philadelphia chromosome-positive chronic myelogenous leukemia (CML), follicular non-Hodgkin lymphoma, Kaposi Sarcoma, advanced renal cell carcinoma, and stage 2 malignant melanoma [80], [84], [85]. The most commonly used form of IFN-α2a is a longer acting pegylated version of IFN-α2 with a branched 40 kD PEG chain attached to it (Peginterferon alfa-2a; Pegasys®, Roche/Pharma& GmbH) [86], [87]. There has also been a shorter acting form of IFN-α2a (Roferon-A®, Roche) which has been discontinued.

Similarly, IFN-α2b is approved for chronic hepatitis B and hepatitis C infection, and chronic hepatitis C in adults in combination with ribavirin. It is also used for the treatment of hairy cell leukemia, AIDS-related Kaposi sarcoma, follicular non-Hodgkin lymphoma in combination with anthracycline chemotherapy, condylomata acuminata and adjuvant therapy for malignant melanoma [84], [85]. Similarly to IFN-α2a, IFN-α2b exists in different synthetic forms. The earlier synthetic versions of IFNα2b to be described are a pegylated one with a 12 kD linear PEG moiety covalently attached to it (Pegintron®, Biogen/Merck) [88], and a recombinant non-modified form (Intron-A®, Biogen/Merck) which have both been recently discontinued due to commercial reasons. However, there is also a third form of IFN-α2b termed Ropeginterferon alpha 2b (Besremi®; PharmaEssentia Corp) which is approved for the treatment of polycythemia vera, myelofibrosis, graft vs. host disease, and essential thrombocythemia [89]. In contrast to the other pegylated IFNα2 compounds which are mixtures of monopegylated positional isomers in which the polymers are covalently linked to one of several internal residues, Ropeginterferon alpha 2b is monopegylated, i.e. it consists of a single positional isomer resulting in an extended half-life and less frequent dosing.

Two synthetic forms of IFN-β, IFN-β-1a and IFN-β-1b, also exist and are approved for the treatment of multiple sclerosis [90]. IFN-β-1a (Avonex®, Biogen or Rebif®, Merck Serono/Pfizer) is a glycosylated preparation of the natural form of IFN-β preparations produced in Chinese hamster ovary (CHO) cells, while IFN-β-1b (Betaferon®/Betaseron®; Bayer Pharma or Extavia®, Novartis) is a non-glycosylated IFN-β Ser17 mutated version produced in Escherichia coli in which there is methionine-1 deletion and a cysteine-17 to serine mutation to prevent formation of intermolecular disulfide bonds and potential aggregation. IFN-β-1a also exists in a pegylated version (Peginterferon beta-1a; Plegridy®, Biogen), which has a 20 kDa PEG molecule covalently attached to the α-amino group of the N-terminus of the protein, resulting in increased half-life and reduced toxicity [91]. Moreover, there is an inhalable form of IFN-β-1a (SNG001; Synairgen) that has also been used in clinical studies [92], [93]. This involves ready-to-use aqueous solution of IFN-β-1a that is administrable through a common mesh nebuliser with adaptive aerosol delivery technology. This has the advantage of being administered locally in the respiratory tract, avoiding the serious adverse effects of systemic type I IFN administration, and therefore constitutes a promising drug candidate for the targeted treatment of respiratory viral infections.

Finally, a pegylated version of IFN-λ1, a type III IFN, has also been generated. This involves a slightly modified IFN-λ1 protein with a conservative substitution of serine for a cysteine at position 171, together with a short N-terminal deletion and addition of an N-terminal methionine, which is also pegylated with a 20 kDa N-terminal linear PEG. In a phase 1a, placebo-controlled, dose escalation study of single subcutaneous doses of PEG-IFN-λ in healthy subjects PEG-IFN-λ was well tolerated at pharmacologically active doses without the toxicities typically associated with subcutaneous pegylated IFN-α2 administration, and exhibited a half-life of 50 to 80 h [94]. Although this has not been approved yet for any indication, it has demonstrated efficacy in clinical studies for the treatment of hepatitis B and C, as well as hepatitis D [95], [96].

The use of IFNs for the treatment of respiratory viral infections has a long history. As IFNs exhibit broad antiviral activities against diverse viral infections, they have frequently been considered for the treatment of difficult to treat diseases or emerging viral strains for which other alternatives do not exist or even quite common viral pathogens that can infect a large proportion of the population. These include influenza viruses, respiratory syncytial viruses and rhinoviruses in which type I IFNs have been administered in prophylactic and/or therapeutic regimens, although efficacy has been mixed and contested [97].

Therefore, when SARS-CoV-2 emerged, type I IFNs were employed for the treatment of some of the first patients of COVID-19 and have actually been included to be part of initial guidelines proposed in China to confront this new disease [98].

Initially, studies where small, often not blinded, with heterogeneous groups of patients and a variety of different co-administered drugs including lopinavir/ritonavir and hydroxychloroquine which have since been shown to have no clinical benefit in people with COVID-19. They have also been performed before the widespread use of remdesivir, corticosteroids and other immunomodulators, limiting their informative value in terms of type I IFN treatment efficacy [99], [100], [101]. However, when later studies were performed that fulfilled these conditions, type I IFNs did not demonstrate any significant benefit, and rather suggested that type I IFNs can cause harm in patients with severe disease such as those who require high-flow oxygen, noninvasive ventilation or mechanical ventilation [102], [103]. Thus, in a large randomized controlled trial of hospitalized patients with COVID-19 that presented radiographic evidence of lower respiratory tract disease and/or required oxygen supplementation, the subcutaneous administration of interferon beta-1a together with remdesivir did not result in any clinical improvement compared to remdesivir alone [102]. Moreover, in the Solidarity trial run by the World Health Organization subcutaneous interferon beta-1a treatment of hospitalized patients, ∼50% of which were on corticosteroids, did not result in any additional benefit [103]. Other trials of smaller or moderate size of interferon alfa drugs administered systemically subcutaneously or intravenously have not demonstrated additional benefit either [104], [105], [106]. This may also be related to the timing of the treatment as many of these studies this was late.

Perhaps a promising exception has been the administration of an inhalable form of IFN-β. In a randomized, double-blind, placebo-controlled, phase 2 pilot trial of hospitalized COVID-19 patients, two thirds of which required oxygen supplementation at admission, daily administration of inhaled nebulized interferon beta-1a (SNG001) for 14 days led to greater improvement and more rapid recovery compared to placebo [93]. However, in a phase 3 study that followed, and in which new treatments such as corticosteroids and/or antivirals were implemented and some patients were vaccinated, such beneficial effects of the treatment were not observed although there was a tendency for improvement of certain parameters such as a reduced risk of progression to severe disease or death [107]. The changing population of patients which includes new variants, new treatments and vaccinated individuals is likely to account for such differences observed.

Interestingly, although trials with type I IFNs have not been very encouraging, this has not been the case with type III IFNs which were also proposed early on for the treatment of COVID-19 [108], [109]. In a randomized, double-blind adaptive clinical trial that enrolled a total of 1941 non-hospitalized patients with risk factors for severe COVID-19 patients, a single subcutaneous administration of pegylated IFN-λ1 demonstrated significant benefit over placebo [110]. Treatment was associated with a 51% decrease in patient visits to the emergency department for > 6 h or hospitalization, and a 39% decrease in patient hospitalization or death. Patients with a high baseline SARS-CoV-2 viral load who received interferon lambda were also more likely to have cleared the virus by Day 7 than those who received placebo [110]. Notably, 83% of these patients were vaccinated and side effects were not higher than placebo. This indicated that pegylated IFN-λ1 administration constitutes a powerful way to reduce emergency visits and hospitalizations of outpatients with COVID19, and is still effective in vaccinated individuals. The differential effects of type III IFNs to inflammation compared to type I IFNs may largely be responsible for the better efficacy to safety profile of pegylated IFN-λ1 administration [111].

Overall, these studies highlight the strong potential of using synthetic forms of IFNs for the treatment of COVID-19, and possibly other respiratory infections, especially IFN-λs and inhalable forms of type I IFNs. However, they also spotlight the need for better defining the disease course for different individuals, and identifying the right patient populations, timepoints and dose regiments to optimally adapt treatment to the ones that would mostly benefit, maximizing efficacy and limiting side effects.