In this study of outpatients with uncomplicated SARS-CoV-2 infection, a single subcutaneous injection of Lambda did not significantly reduce time to viral clearance or resolution of symptoms compared with placebo. We recruited participants within 72 hours of diagnosis, giving us an excellent opportunity to intervene early within the course of infection. We attained excellent follow-up and retention, with few missed visits (< 5%), and little missing data. Despite these strengths and compelling preclinical data–i.e., a plausible mechanism of action, the suppression of IFN activity by respiratory coronaviruses, and both in vitro and in vivo studies showing inhibition of SARS-CoV-2 replication by IFN-λ11,12--this phase 2 trial yielded little promise of efficacy at the tested dose and administration schedule. Lambda was well-tolerated, with few adverse effects, though asymptomatic liver transaminase elevations occurred more frequently in participants randomized to Lambda and are consistent with previous reports.21,26
The lack of effect of Lambda was surprising given recently described in vitro data and benefits seen in an in vivo model with early therapeutic and prophylactic administration12. There are several potential reasons for this lack of benefit. First, although we attempted to randomize participants as soon as possible after the COVID-19 diagnosis was made, the median symptom duration was 5 days at the time of randomization, and 40% of participants were already SARS-CoV-2 IgG positive at enrollment. It is possible that earlier administration, or prophylactic administration prior to established infection, would have been beneficial. Arguing against this, we observed no evidence of benefit among SARS-CoV-2 seronegative individuals, who presumably have been infected a shorter period of time. Second, a single, 180 mcg subcutaneous injection of Lambda may not achieve adequate therapeutic levels of drug in the upper respiratory epithelia. Consistent with this, a murine model of SARS-CoV-2 infection found that subcutaneous administration of Lambda did not result in significant reductions of SARS-CoV-2 viral titers in upper respiratory epithelium.12 It is possible that higher, or more frequent, dosing may have been beneficial. However, subcutaneous doses greater than 180 mcg in humans are limited by increasing drug toxicity, including significant liver transaminase elevations21. Finally, IFN-λ has been shown to disrupt the lung epithelial barrier in mice, leading to worsened disease course and increase susceptibility to bacterial superinfection27,28. This may negate any positive antiviral effects.
These data are in contrast to reports of benefits of Type I IFN in hospitalized COVID-19 patients. Subcutaneous IFN-β along with ribavirin and lopinavir/ritonavir was associated with shortened duration of symptoms and viral shedding in hospitalized patients in Hong Kong14, and a randomized clinical trial in England also suggest benefits of inhaled IFN-β for COVID-19.29 Although both type I and type III IFN activate the same dominant JAK-STAT signaling pathway6, inhibit SARS-CoV-2 in vitro10,11, and have receptors on respiratory epithelia16, in vivo activity and efficacy may differ6. A recent report found that inborn errors of Type I IFN immunity, including autosomal recessive IFNAR1 deficiency, were enriched in patients with life-threatening COVID-19 pneumonia30. Furthermore. patients with COVID-19 pneumonia were also more likely to have neutralizing auto-antibodies against type I, but not type III, IFNs31. These data suggest the possibility that type I IFN administration may be more beneficial than type III IFN in preventing adverse outcomes of SARS-CoV-2 infection.
Although there was some evidence that SARS-CoV-2 seropositivity at baseline modified the effect of treatment on shedding cessation, the effect modification was in the opposite direction than we had anticipated; Lambda appeared to prolong shedding relative to placebo among those who were seronegative at baseline, and to shorten the duration of shedding among those who were seropositive at baseline. Furthermore, Lambda also appeared to prolong shedding relative to placebo among those with high baseline viral RNA levels. These findings should be interpreted with caution for two reasons. First, these were exploratory analysis only and should be considered hypothesis generating. Second, these observations defy biological plausibility based on in vitro and animal model data.
The majority (62.5%) of participants in our study were Latinx, reflecting the high burden of COVID-19 among the Latinx community in our surrounding communities.32 Minority populations are disproportionately affected by COVID-19, with higher rates of infection and deaths observed due to a multitude of socioeconomic and demographic factors.33,34 Attention has recently been called to the relative absence of the most affected minorities in treatment trials35,36, and we prioritized recruitment efforts to the Latinx community in our study.
The study did have a few limitations. We recruited both symptomatic and asymptomatic patients. Asymptomatic patients contributed less to secondary outcomes since they presented with lower viral RNA levels and could not contribute to analyses of symptom alleviation. However, these patients represented < 10% of the enrolled cohort. Additionally, despite a reported median duration of symptoms prior to randomization of only 5 days, 40% of participants were already seropositive at enrollment. Unpublished data from a Regeneron outpatient monoclonal antibody study with similar study design (REGN-COV2) found similar rates of baseline SARS-CoV-2 IgG seropositivity (45%).37 These data suggest that enrolling COVID-19 outpatients early in the course of disease, before they develop an antibody response, may be challenging. Nonetheless, we found no suggestion of benefit of Lambda in seronegative individuals. Finally, the median time to cessation in the placebo arm was shorter than assumed in our sample size calculations, potentially due to less severe disease in this population. However, our original sample size estimates based on the number of events and median time to event were conservative; a shorter time to cessation, keeping all other assumptions the same, increases the power to detect differences between groups.
In conclusion, a single dose of subcutaneous Peginterferon Lambda-1a, while safe, neither reduced time to cessation of viral shedding nor symptom duration in outpatients with uncomplicated COVID-19 in this large, Phase 2, single-center study. Further investigation into the therapeutic utility of type III interferons for COVID-19 in patients with severe illness or as a prophylactic treatment are underway.