Review: The Landscape of Antiviral Therapy for COVID-19 in the Era of Widespread Population Immunity and Omicron-Lineage Viruses

Abstract

The goals of coronavirus disease 2019 (COVID-19) antiviral therapy early in the pandemic were to prevent severe disease, hospitalization, and death. As these outcomes have become infrequent in the age of widespread population immunity, the objectives have shifted. For the general population, COVID-19–directed antiviral therapy should decrease symptom severity and duration and minimize infectiousness, and for immunocompromised individuals, antiviral therapy should reduce severe outcomes and persistent infection. The increased recognition of virologic rebound following ritonavir-boosted nirmatrelvir (NMV/r) and the lack of randomized controlled trial data showing benefit of antiviral therapy for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection for standard-risk, vaccinated individuals remain major knowledge gaps. Here, we review data for selected antiviral agents and immunomodulators currently available or in late-stage clinical trials for use in outpatients. We do not review antibody products, convalescent plasma, systemic corticosteroids, IL-6 inhibitors, Janus kinase inhibitors, or agents that lack Food and Drug Administration approval or emergency use authorization or are not appropriate for outpatients.

A detailed review of COVID-19 antivirals and immunomodulators focused on outpatients in the era of Omicron-lineage viruses and widespread population immunity, highlighting treatment considerations, such as the risk of virologic rebound and the limits of the current body of evidence.

DEFINING RELEVANT ENDPOINTS FOR EFFECTIVE COVID-19–DIRECTED ANTIVIRAL THERAPY IN THE ERA OF WIDESPREAD POPULATION IMMUNITY

The goals of antiviral therapy have shifted from prevention of hospitalization and death—outcomes that have become infrequent for immunocompetent, standard-risk adults with baseline immunity against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) (through vaccination and/or prior infection)—to more rapid reduction in the intensity and duration of coronavirus disease 2019 (COVID-19) symptoms and truncation of the period of infectiousness.

Outpatient trials of COVID-19 antivirals have adapted the influenza severity scale (FLU-PRO) to grade symptoms [1, 2]. These COVID-19 symptom-assessment tools have some evident problems. First, symptom profiles have changed. For example, altered taste and smell are far less and sore throat far more common for Omicron-lineage versus Delta and pre-Delta viruses [3–5]. Second, certain symptoms such as shortness of breath were more common for unvaccinated/immune-naive individuals [6]. Third, these tools may not assess the most relevant symptoms impacting functional status. For instance, acute perceived cognitive deficits (“brain fog”) is omitted [7]. While defining an appropriate symptom-assessment tool is beyond the scope of this review, the lack of one should be urgently addressed.

Additionally, effective approaches for the treatment of immunocompromised individuals are needed to prevent severe outcomes and death. Many trials excluded (or did not enroll) children, pregnant individuals, and immunocompromised adults, presenting challenges for the management of severe SARS-CoV-2 infections in these groups [8–10].

ANTIVIRAL AGENTS—DETAILED REVIEW AND LOOK AHEAD

RNA-Dependent RNA Polymerase Inhibitors

RNA-dependent RNA polymerase (RdRp), an error-prone enzyme, is indispensable for SARS-CoV-2 replication and is a key therapeutic target [11]. Two groups of RdRp inhibitors are currently available, including adenosine nucleoside analogs (remdesivir [RDV] and its derivatives] that block RdRp translocation and delay chain termination [12] and a cytidine analog [molnupiravir [MLP]) that induces lethal mutagenesis in SARS-CoV-2 [13].

Remdesivir

After being the first antiviral to demonstrate efficacy among hospitalized individuals, RDV (Gilead Sciences, Foster City, CA, USA) was also the first to show benefit in high-risk outpatients in PINETREE (Table 1) in the pre-Delta period [8, 9]. PINETREE included unvaccinated individuals at increased risk for severe COVID-19. The primary endpoint of COVID-19–related hospitalization or death from any cause by day 28 occurred in 2 of 279 (0.7%) patients receiving RDV and 15 of 283 (5.3%) patients receiving placebo (hazard ratio, .13; 95% confidence interval [CI], .03–.59; P = .008). A higher percentage receiving RDV had symptom alleviation by day 14 (36.1% vs 20.0%; rate ratio, 1.92; 95% CI, 1.26–2.94) (Table 1). Despite clinical benefit, nasopharyngeal viral load (VL) decline did not differ between the groups [9], possibly because RDV limits viral replication in the lower but not upper respiratory tract, a finding shown in rhesus macaques [28]. Whether the effect of RDV might be attenuated for Omicron-lineage viruses, which replicate less efficiently in the lungs compared with prior SARS-CoV-2 viruses, is unknown [29, 30]. The intravenous administration of RDV over 3 days makes it infeasible for broad use due largely to convenience and logistical issues related to outpatient infusion in the setting of acute COVID-19.

Table 1.

Summary of Small-Molecule Antiviral Agents and Immunomodulators in Nonhospitalized Populations: A Nonexhaustive List Focusing Mostly on Late-Phase Trials

Target	Antiviral	Trial [Ref]	Population	Endpoint	No.a	Eventb	Risk Reduction (95% CI) Unless Otherwise Specified	More Rapid Viral Load Decline Compared With Placebo Reported
RdRp	Remdesivir	PINETREE [9]	Unvaccinated, high-risk	1°: COVID-19–related hospitalization/all-cause death through day 28	562	0.7 vs 5.3	Absolute (%): ∼4.6 (1.7, 7.4); relative (%): 87 (41, 97)	No (day 7)
	Molnupiravir	MOVe-OUT phase 2 [14]	Unvaccinated, high-risk (75.2%)	1°: All-cause hospitalization/death through day 29	302	3.1 vs 5.4	Absolute (%): ∼2.3 (−3.5, 8.1); relative (%): ∼42 (NA, 85)	Yes (day 5, only ≤5 dpso)
		MOVe-OUT phase 3 [15]	Unvaccinated, high-risk (99.4%)	1°: All-cause hospitalization/death through day 29	1433	6.8 vs 9.7	Absolute (%): 3.0 (0.1, 5.9); relative (%): ∼30 (−2, 53)	Yes (day 3, 5, 10)
				2°: Median time to sustained alleviation of all 15 targeted signs and symptoms	1408 (mITT)	Median, 15 vs 16 days	P < .05
		PANORAMIC [16]	Vaccinated, high-risk (≥99%)	1°: All-cause hospitalization/death through day 29	26 411	0.8 vs 0.8	Absolute (%): ∼−0.06 (−0.2, 0.3); relative (%): −6% (−41, 19)	Yes (day 5)
				2°: Time to first reported recovery		Median, 9 vs 15 days	P Superiority > .99
		PLATCOV (refer to NMV/r section)		…	…	…	…	…
	VV116	Cao et al [17]	Vaccinated (75.7%) and unvaccinated, high-risk	1°: Time to sustained clinical recovery through day 28	822	Median, 4 vs 5 days in VV116 vs NMV/r	HR, 1.17 (1.02, 1.36; NI boundary, 0.8)	NA, compared to active treatment NMV/r
Mpro	NMV/r	EPIC-HR [10]	Unvaccinated, high-risk	1°: COVID-19–related hospitalization/all-cause death through day 28	2246	0.8 vs 6.3	Absolute (%): 5.6 (4.0, 7.2); relative (%): ∼88 (75, 95)	Yes (day 3, 5, 10, 10, 14, only ≤3 dpso)
					1379 (mITT, enrolled ≤3 dpso)	0.7 vs 6.5	Absolute (%): 5.8 (3.8, 7.8); relative (%): 89
		EPIC-SR [18]	Vaccinated and unvaccinated (no vaccination within 12 months of enrollment), low-risk	1°: Time to sustained alleviation of COVID-19 signs and symptoms through day 28	1296	Median, 12 vs 13 days	P = .6	Likely, pending formal report (day 3, 5)
				2°: COVID-19–related hospitalization/all-cause death through day 28		0.8 vs 1.7	Absolute (%): ∼0.97 (−0.25, 2.20); relative (%): ∼56 (−38, 88); P = .18
		PLATCOV [19]	Vaccinated, low-risk	1°: Rate of oropharyngeal viral clearance, measured by change in rate of clearance at day 7 (higher percentage indicated faster clearance)	209 (NmITT = 207)	Not reported	Relative (%) NMV/r vs placebo: 84 (54–119); MLP vs placebo: 37 (16–65); NMV/r vs MLP: 25 (10–38)	Yes
	Ensitrelvir	Mukae et al, phase 2b [20]	Mostly vaccinated (∼85%), mixed low- and high-risk	Co-1°: Change from baseline in SARS-CoV-2 culturable viral titer on day 4	428	−1.49 vs −1.08 log10 TCID50/mL	Viral titer absolute change: −0.4 (−0.5, −0.3)	Yes (day 2, 4, 6)
				Co-1°: Time-weighted average change from baseline up to 120 hours in the total symptom score		−5.4 (125 mg) vs −5.2 (250 mg) vs −5.1 (placebo)	Absolute: 125 mg vs placebo: −0.2 (−0.8, 0.3); 250 mg vs placebo: −0.04 (−0.62 to 0.53)
		SCORPIO-SR, phase 3 [21]	Mostly vaccinated (>90%), low-risk (∼75%)	1°: Time to resolution of 5 COVID-19 symptoms	1821 (NITT = 1030)	167.9 vs 192.2 hours	Absolute: −24.3 (−78.7, 11.7) hours	Yes (day 2, 4, 6)
Immunomodulator	Metformin	COVID-OUT [22–24]	Half-vaccinated (52.2%), overweight/obese	1°: Hypoxemia, ED visit, hospitalization, or death	1431 (metformin and placebo)	23.6 vs 27.4	Absolute (%): ∼3.8 (−1.7, 9.3); relative (%): 16 (−9, 34)	Yes
				2°: ED visit, hospitalization, or death		4.1 vs 7.3	Absolute (%): ∼3.2 (0.6, 5.8); relative (%): 42 (6, 65)
				2°: Incidence of long COVID	1126 (long-term follow-up)	6.3 vs 10.4	Relative (%): 41 (11, 61)
		TOGETHER [25]	Unvaccinated, high-risk	1°: Hospitalization (or transfer to a tertiary hospital) or an ED visit (observation for >6 hours)	418	15.8 vs 13.8	Absolute (%): ∼−2 (−9.4, 5.4); relative (%): −14 (−81, 27)	NA
	Pegylated interferon lambda	TOGETHER [26]	Mostly vaccinated (83%), high-risk	1°: Hospitalization (or transfer to a tertiary hospital) or an ED visit (observation for >6 hours)	1951	2.7 vs 5.6	Absolute (%): ∼2.9 (1.1, 4.7); relative (%): 51 (24, 70)	Yes, only baseline VL > 8.3 log10 copies/mL (day 7)
	Inhaled interferon beta 1a	ACTIV-2 [27]	Unvaccinated, mixed risk	1°: Time to symptom improvement of 13 targeted COVID-19 symptoms	221	13 vs 9	P = .17	No

Abbreviations: CI, confidence interval; Co-1°, co-primary; COVID-19, coronavirus disease 2019; dpso, day post–symptom onset; ED, emergency department; EPIC-HR, Evaluation of Protease Inhibition for Covid-19 in High-Risk Patients; EPIC-SR, Evaluation of Protease Inhibition for COVID-19 in Standard-Risk Patients; HR, hazard ratio; ITT, intention-to-treat; mITT, modified intention-to-treat; M^pro, main protease; NA, not available; NI, noninferior; NMV/r, nirmatrelvir-ritonavir; PANORAMIC, Platform Adaptive trial of NOvel antiviRals for eArly treatMent of COVID-19 In the Community; RdRp, RNA-dependent RNA polymerase; Ref, reference; TCID₅₀, 50% tissue culture infectious dose; VL, viral load; 1°, primary; 2°, secondary; ∼, indicates univariate estimation from the authors of this review based on the numbers provided in the original literature, not directly reported by the original studies.

^aNumber of participants undergoing randomization unless otherwise specified.

^bEvent, indicating percentage of participants who reached the primary endpoint in active treatment vs control/placebo groups unless otherwise specified.

Molnupiravir

The efficacy of MLP (Merck, Rahway, NJ, USA and Ridgeback Biotherapeutics, Miami, FL, USA) in unvaccinated, high-risk outpatients was established by the phase 2/3, randomized, double-blinded clinical MOVe-OUT trials in the Delta period [14, 15]. In MOVe-OUT phase 3, 1433 participants were randomized within 5 days of symptom onset [15]. The primary endpoint was hospitalization or death through day 29. In the modified intention-to-treat analysis 6.8% versus 9.7% were hospitalized or died (all-cause difference, −3.0%; 95% CI, −5.9% to −.1%) (Table 1) in the MLP and placebo groups, respectively [15]. Among those with positive baseline anti-nucleocapsid antibodies [31], MLP showed a trend towards higher risk of hospitalization or death (risk reduction, −2.3%; 95% CI, −7.1% to 1.7%) [15]. In a post hoc analysis of the study, MLP was associated with more rapid alleviation of symptoms (15 vs 16 days for median time to sustained alleviation of all symptoms) (Table 1) [32].

Platform Adaptive trial of NOvel antiviRals for eArly treatMent of COVID-19 In the Community (PANORAMIC), a large-scale, open-label, platform-adapted randomized controlled trial (RCT) assessing the efficacy of MLP in a high-risk, largely vaccinated population (>90% with ≥3 vaccine doses) during the Omicron era, found no difference in hospitalization or death (1% vs 1%) [16]. However, MLP was associated with accelerated symptomatic recovery on both days to first-reported recovery (9 vs 15 days; probability of superiority, >0.99) and days to sustained recovery (21 vs 24 days; probability of superiority, >0.99) [16]. In addition, MLP was associated with earlier reduction in upper respiratory tract SARS-CoV-2 VL, a finding consistently seen in trials enrolling unvaccinated and vaccinated participants [33, 34]. Paradoxically, there was a signal towards higher hospitalization or death for the immunocompromised subgroup receiving MLP (adjusted odds ratio, 1.89; 95% CI, .99–3.73). The significant heterogeneity of the group (people with human immunodeficiency virus [HIV], sickle cell disease, and stem cell transplant recipients) limits detailed analysis since levels of immunosuppression are associated with different viral decay and immune response [35, 36].

While several real-world studies have suggested that MLP is associated with a lower incidence of hospitalization and/or death in unvaccinated [37], vaccinated [38], and immunocompromised [39] populations, these studies may carry confounding factors and intrinsic biases even after adjustment and propensity-score matching.

Molnupiravir carries several potential risks. Mammalian cell culture–based mutagenesis assays demonstrated higher risk for DNA mutagenesis [40]. An analysis of global SARS-CoV-2 sequences also detected elevated MLP-associated mutational signature (G-to-A mutations) after MLP introduction [41]. The Food and Drug Administration recommends against MLP use during pregnancy due to potential teratogenesis [42].

VV116

VV116 (Junshi Bio, Shanghai, China) is a derivative of the RDV prodrug of GS-441524 (the main active metabolite of RDV) [43, 44]. A phase 3, observer-blinded, noninferiority RCT enrolled 822 participants (∼75% vaccinated) comparing VV116 or ritonavir-boosted nirmatrelvir (NMV/r) and found that the median time to sustained clinical recovery was 4 versus 5 days for VV116 and NMV/r, respectively (hazard ratio, 1.17; 95% CI, 1.02–1.36; noninferiority margin, >0.8) (Table 1) [17]. However, Evaluation of Protease Inhibition for COVID-19 in Standard-Risk Patients (EPIC-SR) found that NMV/r does not shorten symptom duration for standard-risk, vaccinated patients [18]. Additionally, only patients with very low baseline symptoms were included (median symptom score of 3 on a scale from 0 to 33) [17, 45]. The rapid time to sustained recovery in both groups (4–5 days) is far shorter than that seen in the NMV/r-treated participants of EPIC-SR (12 days). A subsequent study of VV116 in China randomized 1369 patients in a 1:1 ratio to receive VV116 or placebo (Citation: https://doi.org/10.1016/S1473-3099(23)00577-7). A 14-point COVID symptom scale was used (max score 42), with a median baseline score of 10, more suggestive of typical mild, uncomplicated COVID-19. The primary endpoint was the time to sustained clinical symptom recovery, defined as the first two consecutive days when the COVID symptom score was 0, and was met by VV116. The median time to sustained clinical symptom resolution was 10.9 days vs 12.9 days in the VV116 and placebo groups, respectively (stratified HR 1.17, 95% CI 1.04–1.33). VV116 is currently not available in the United States.

Obeldesivir

Obeldesivir (GS-5245, Gilead Sciences) is another orally administered prodrug of GS-441524 [46]. It has been shown to have efficacy in animal models [47]. Two phase 3, randomized, double-blinded, placebo-controlled trials evaluating obeldesivir in standard-risk (OAKTREE, NCT05715528) and high-risk (BIRCH, NCT05603143) populations are ongoing.

Protease Inhibitors

The 3CL protease (3CL^pro or M^pro), the main target of nirmatrelvir [48], is a cysteine protease that cleaves the 2 polyproteins encoded by the SARS-CoV-2 ORF1ab. Selection of viral rather than human proteins is possible because this enzyme cleaves after a glutamine residue, which is not known to occur among human proteases [48].

Ritonavir-Boosted Nirmatrelvir

A 5-day course of NMV/r (Pfizer, New York, NY, USA) showed benefit for high-risk, unvaccinated outpatients in Evaluation of Protease Inhibition for Covid-19 in High-Risk Patients (EPIC-HR) in the Delta era [10] (Table 1). An ongoing, UK-based platform clinical trial (PANORAMIC) is evaluating NMV/r versus placebo in high-risk, largely vaccinated individuals [49]. EPIC-SR testing of NMV/r in standard-risk, vaccinated and unvaccinated adults spanning the Delta and Omicron periods was ended early by the manufacturer [50] and did not meet the primary endpoint of time to sustained alleviation of COVID-19 symptoms by day 28 [18]. While vaccination was allowed, no vaccine doses within 12 months of enrollment were permitted. Time to resolution of symptoms was not different (12 vs 13 days, P = .603). A 0.737-log₁₀ copies/mL and 0.834-log₁₀ copies/mL greater reduction in VL for NMV/r-treated individuals was seen at day 3 and 5, respectively.

Virologic rebound is defined as a sustained significant increase in SARS-CoV-2 respiratory tract VL following an initial decline with or without associated worsening of COVID-19 symptoms [51]. Importantly, the incidence of virologic rebound can be highly affected by the way it is defined and the frequency of sampling [51, 52]. In the EPIC studies, virologic rebound was defined as “a 0.5 log10 copies/mL increase in VL” on day 10 or day 14 if only one viral sampling was available or day 10 and 14 if both were available. In EPIC-SR, virologic rebound was identified in 3 of 292 (1.0%) Delta and 1 of 62 (1.6%) Omicron NMV/r-treated and 4 of 232 (1.7%) Delta and 0 of 55 (0%) Omicron placebo-arm participants [53]. In EPIC-HR, VL rebound was documented in 23 of 990 (2.3%) NMV/r-treated and 17 of 980 (1.7%) placebo-arm participants, although infrequent sampling almost certainly undercounts the NMV/r-arm result in both studies [51, 54]. In the Post-vaccination Viral Characteristics Study (POSITIVES) cohort [54], VL was assessed 3 times weekly for the first 2 weeks and then weekly until undetectable, whereas in EPIC-SR and EPIC-HR it was assessed only on days 10 and 14. When POSITIVES included VLs only on days 10 and 14, 13 of 16 (81.2%) rebound cases were missed [54]. In clinical practice, virologic rebound may be identified when negative rapid antigen tests become newly positive after completion of a course of NMV/r [55].

Virologic rebound to more than 4.0 log₁₀ copies/mL after 5 days of symptoms may occur in approximately 10% of untreated individuals, although 95% are transient events, with very few having sustained rebound at multiple time points [52]. The degree to which untreated patients who rebound have prolonged infectiousness is not well defined, with culture data, to our knowledge, from a single individual with viable virus isolated through day 10 after diagnosis [54]. For context, in untreated, immunocompetent individuals, isolation of infectious virus has only rarely occurred beyond 10 days from symptom onset [56–58].

In treated patients, rebound occurs after completion of NMV/r and is associated with prolonged shedding of infectious virus [54, 55]. Among the NMV/r-treated individuals in POSITIVES, the median duration of shedding of infectious virus was far longer for those who rebounded (14 vs 3 days). PLATCOV, an open-label, phase 2, randomized platform trial enrolling vaccinated standard-risk participants found more rapid upper respiratory airway viral clearance for NMV/r (n = 58; viral decay half-life, 8.5 hours) versus MLP (n = 65; viral decay half-life, 11.6 hours) versus placebo (n = 84; viral decay half-life, 15.5 hours) [19]. An exploratory analysis found more viral rebound in the NMV/r group (10%) versus MLP (2%, P = .051) versus placebo (1%, P = .018), where viral rebound was defined as 2 consecutive days of VL of less than 100 copies/mL followed VL greater than 1000 copies/mL [19].

Rebound likely occurs due to transient suppression of viral replication with resurgence after withdrawal of antiviral pressure. Rebound does not seem to occur after RDV treatment, possibly because the long half-life of active metabolites of RDV versus NMV (>24 intracellularly vs ∼6 hours) [59, 60]. Additionally, NMV-resistance mutations have not been widely detected after rebound [52, 61, 62] unless in the setting of severe immunosuppression [63]. In contrast, emergent resistance was commonly seen in rebound with early protease inhibitors for hepatitis C [64]. Rebound does not occur after antiviral treatment of influenza, which has a far shorter period of infectiousness (∼3 days vs ∼10 days) [65, 66].

If insufficient treatment duration explains NMV/r-associated rebound, it might be more common for Omicron-lineage viruses since the incubation period has shortened and COVID-19 diagnosis occurs earlier (Figure 1) [57, 69]. Interestingly, POSITVES found rebound in 9 of 31 (29%), 6 of 36 (16.7%), and 0 of 5 (0%) of individuals starting NMV/r on day 0, 1, or 2 after diagnosis, respectively, with diagnosis a median of 1 day after symptom onset [54].

Figure 1.

SARS-CoV-2 viral load temporal trend in Delta and Omicron variants in immunocompetent individuals, with arrows indicating possible symptom onset. Of note, there is great variablility in terms of timing due to host immunity, comorbidities and subvariants. Abbreviation: SARS-CoV-2, severe acute respiratory syndrome coronavirus 2. The incubation time and relative peak viral load levels reflect findings from references [67] and [68].

Besides rebound with a true rate of likely 10–30% [52, 54, 70], numerous drug–drug interactions must be considered before NMV/r therapy and patients must be informed about metallic taste and other important potential side effects. One ongoing trial tests a repeat 5-day course of NMV/r for people with rebound (NCT05567952). Two modeling studies hypothesize that the timing of NMV/r initiation could lead to preservation of certain host cells typically depleted by the virus [71, 72], where viral replication resumes once antiviral pressure is removed. This suggests that timing of initiation may be a key variable determining the risk of virologic rebound.

Ensitrelvir

Ensitrelvir (formerly S-217622, Shionogi, Osaka, Japan) is another M^pro inhibitor [73]. A low-dose (125 mg) regimen received emergency regulatory approval in Japan based on trials conducted there [20, 74]. In a phase 2b trial in Japan and South Korea, 428 vaccinated participants were randomized to ensitrelvir low-dose (375 mg on day 1, then 125 mg on days 2–5), high-dose (750 mg on day 1, then 250 mg on days 2–5), and placebo [20]. Primary endpoints included SARS-CoV-2 viral titer and COVID-19 symptom score. Most participants (>80%) received at least 2 vaccine doses. Ensitrelvir use (both doses) was significantly associated with greater virus titer reduction by day 4 and faster improvement in certain symptoms (acute symptoms and respiratory symptoms), but not total symptoms, with a good safety profile. A subsequent phase 3 trial (double-blind, randomized, placebo-controlled) enrolled 1821 participants from Japan, Vietnam, and South Korea [21]. The primary endpoint was time to resolution of 5 COVID-19 symptoms (nasal symptoms, sore throat, cough, fever, and fatigue). Over 90% were vaccinated with at least 1 dose and 70% were deemed high risk. The low dose was associated with significant shortening of the 5 key symptoms in the intention-to-treat analysis (167.9 hours vs 192.2 hours in the low-dose vs placebo group; difference in median, −24.3 hours). Both ensitrelvir dose groups experienced more rapid decline of SARS-CoV-2 RNA and culturable virus (Table 1) [21]. A post hoc analysis of the SCORPIO-SR phase 3 study found virologic rebound in 7.8% in the ensitrelvir 125-mg group (n = 590) and 4.7% in the placebo group (n = 574) by day 21 after treatment [75]. Additionally, investigators found that early treatment with ensitrelvir was associated with a more rapid improvement in smell and taste symptom deficits. For instance, by day 6 after treatment, 39% fewer people receiving ensitrelvir 250 mg (n = 334) reported impaired taste or smell compared with those receiving placebo (n = 326) [76]. Furthermore, an exploratory analysis of the SCORPIO-SR phase 3 data found an attenuation in long-COVID symptoms among those treated with ensitrelvir [77]. Long COVID was defined as either persistence of any of 14 COVID-19 symptoms at mild or more severe levels during follow-up or 1 or more of 4 neurological symptoms associated with long COVID (difficulty with concentration or thinking, difficulty with reasoning or problem solving, memory loss, and insomnia). Among the subpopulation with a high baseline symptom score, the ensitrelvir 125-mg dose (n = 131 with a high baseline score of 232 total) had a 45% lower rate of persistent COVID-19 symptoms by day 169 compared with placebo (n = 118 with a high baseline symptom score of 218 total). Among those with a high baseline COVID-19 symptom score, the ensitrelvir 125-mg dose (n = 180) had a 33% lower and the ensitrelvir 250-mg dose (n = 148) had a 26% lower rate of 4 long-COVID neurological symptoms at day 169 compared with placebo (n = 175) [77]. Although not boosted, ensitrelvir is a moderate CYP3A inhibitor and drug–drug interactions need to be considered before treatment [78].

Immunomodulators

Metformin

COVID-OUT was a randomized, blinded, parallel-group, phase 3 trial testing dose-escalated metformin for 14 days, fluvoxamine, and ivermectin in outpatients in the pre-Delta, Delta, and Omicron periods [22, 23]. Important inclusion criteria included age 30–85 years, overweight or obese, no prior history of confirmed SARS-CoV-2 infection, and symptom onset within 7 days of randomization. The primary endpoint was a composite of hypoxemia of 93% or less measured using finger pulse oximeter at home, emergency department visits, hospitalizations, and death.

The primary endpoint occurred in 154 of 652 (23.6%) and 179 of 653 (27.4%) metformin and control participants (adjusted odds ratio, .84; 95% CI, .66–1.09), with a trend towards benefit for hospitalization or death occurring in 1.2% and 2.7% for metformin versus control (adjusted odds ratio, .47; 95% CI, .20–1.11) [22]. TOGETHER evaluated metformin for acute COVID-19 in the pre-Delta period and similarly found no benefit [25].

COVID-OUT found a significantly lower rate of long COVID among those who received metformin, with a cumulative incidence of 6.3% (95% CI, 4.2–8.2%) versus 10.4% (95% CI, 7.8–12.9%) and a hazard ratio of .59 (95% CI, .39–.89; P = .012) (Table 1) [23]. The benefit was more pronounced in those younger than 45 years, who were unvaccinated, and those starting within 3 days of symptom onset and was not seen in vaccinated individuals. “Long COVID” was diagnosed in 5 of 63 (7.9%) pre-Delta, 66 of 800 (8.3%) Delta, and 22 of 263 (8.4%) Omicron-era participants, although other studies have found that it may be less common following Omicron infections and less likely among vaccinated individuals [79, 80]. Metformin is currently being evaluated by the Accelerating COVID-19 Therapeutic Interventions and Vaccines (ACTIV-6) study (NCT06042855), which allows concurrent use of other antivirals.

Pegylated Interferon Lambda

TOGETHER was a randomized, blinded, controlled, adaptive platform trial testing pegylated interferon lambda in high-risk adults during the Delta and Omicron periods [26]. Inclusion criteria included age 18 years and older and presentation within 7 days of symptom onset. Participants received subcutaneous injection of pegylated interferon or placebo. The primary outcome was COVID-19–related hospitalization or emergency department visit for more than 6 hours within 28 days. Approximately 85% of trial participants received at least 1 COVID-19 vaccine dose.

The trial met the primary endpoint—2.7% versus 5.6% for treatment and placebo, respectively (relative risk, .49; 95% CI, .30–.76; probability of superiority, >99.9%) (Table 1) [26]. The effect was strongest for Omicron-lineage infections, perhaps due to the significant immune evasiveness of this variant [81–83]. Among the prespecified 15% with high baseline upper respiratory tract VL, those treated with pegylated interferon lambda had a greater reduction in VL by day 7 (median log₁₀ decline of 8.29 vs 5.16). Among those without high baseline VL, there was no difference in VL reduction by day 7.

From early in the pandemic, an attenuated interferon response was believed to contribute to severe COVID-19 pathophysiology [84]. While TOGETHER was open to immunosuppressed individuals, few, if any, appear to have been enrolled [26]. This agent is unlikely to be useful for standard-risk individuals, although it should be explored further in RCTs for immunocompromised individuals.

Inhaled Interferon Beta 1a

The Accelerating COVID-19 Therapeutic Interventions and Vaccines-2 (ACTIV-2/A5401) phase 2, platform RCT tested nebulized interferon beta 1a (Synairgen, Southamptom, UK) versus placebo in the pre-Delta and Delta periods [27]. Only 20% had received COVID-19 vaccines. The median time to symptom improvement was 13 versus 9 days for the agent and placebo, respectively (P = .17), with no faster VL decline. There was a trend towards fewer hospitalizations in the treatment arm (1% vs 6%, P = .07), all in unvaccinated participants.

The trend towards fewer hospitalizations in unvaccinated participants is consistent with the findings from TOGETHER [26]. This agent is unlikely to be useful for standard-risk individuals but could be explored for immunocompromised individuals and/or those with very high baseline VL.

RESISTANCE TO ANTIVIRALS

Although several in vitro studies have demonstrated pathways to RdRp inhibitor or M^pro inhibitor resistance [85–87], clinically meaningful resistance to these small-molecule antivirals is unusual in real-world settings. Remdesivir-resistance mutations remain rare based on the Global Initiative on Sharing All Influenza Data (GISAID) database [88].

Certain transmissible mutations in M^pro that confer reduced susceptibility to NMV/r and ensitrelvir existed prior to the introduction of M^pro inhibitors and remain quite rare, with uncertain clinical significance [89, 90].

Clinically meaningful resistance to RDV and NMV/r has largely been reported in immunocompromised individuals, especially in those with hematological malignancies, history of solid-organ transplantation, and/or B-cell deficiencies, with persistent COVID-19 and corresponding antiviral use [63, 91–93].

Some favorable outcomes have been reported in case reports and series evaluating combination antiviral therapies with or without monoclonal antibody products or convalescent plasma in immunocompromised individuals with persistent COVID-19, although publication bias is a significant concern [94–96].

Mechanistically, combination therapy is reasonable in highly immunocompromised individuals at risk of persistent COVID-19 and emergent drug resistance. To date, 1 trial evaluating NMV/r plus RDV compared with NMV/r monotherapy in immunocompromised individuals is ongoing (NCT05587894).

CONCLUSIONS

COVID-19 antiviral therapy in the age of widespread population immunity should decrease the severity of COVID-19 symptoms and reduce the period of infectiousness (Figure 2). There are major limitations to the currently available antivirals for standard-risk patients with baseline immunity in the era of Omicron-lineage viruses. Because of the shorter incubation period of Omicron-lineage compared with prior SARS-CoV-2 viruses, antivirals are initiated earlier in the course of viral replication. Consequently, rebound following NMV/r may be more common than it was previously. Additionally, too little is known on the impact of COVID-19 antiviral therapy on long COVID. Ongoing trials are therefore needed to identify safe, effective COVID-19 antivirals for endemic SARS-CoV-2 and to define optimal approaches for therapy for immunocompromised individuals and other special populations.

Figure 2.

Summary of the impact of current antivirals (excluding monoclonal antibody products) on risk of severe disease (hospitalization/death) and time to symptom resolution according to the populations tested (whether high-risk and whether individuals had baseline immunity from vaccination and/or prior SARS-CoV-2 infection) based on the findings from key trials. The y-axis indicates the relative effect size for reducing the risk of developing severe disease (calculated by [1 – hazard ratio, relative risk or odds ratio for active treatment] × 100%) or relative effect size for more rapid symptom resolution (calculated by [hazard ratio for active treatment – 1] × 100%). For statistically nonsignificant negative relative effect size, we truncate it to zero in this figure. *Original reports did not report hazard ratios or odds ratios for symptom resolution; however, based on the data presentation with numerically very close median durations and nonsignificant results, we labeled the relative effect size as near-zero. ^#Original reports did not provide sufficient data to calculate relative size. Abbreviations: EPIC-HR, Evaluation of Protease Inhibition for Covid-19 in High-Risk Patients; EPIC-SR, Evaluation of Protease Inhibition for COVID-19 in Standard-Risk Patients; MLP, molnupiravir; NMV/r, ritonavir-boosted nirmatrelvir; PANORAMIC, Platform Adaptive trial of NOvel antiviRals for eArly treatMent of COVID-19 In the Community; PEG-IFNλ, pegylated interferon lambda; RDV, remdesivir; SARS-CoV-2, severe acute respiratory syndrome coronavirus 2.

Notes

Financial support. Y. L. is currently supported by National Institutes of Health Rustbelt CFAR (Case Western Reserve University/University Hospitals Cleveland Medical Center and University of Pittsburgh, P30 AI036219), not related to the current work.