We conducted a large prospective multicentre cohort study in which critically ill COVID-19 patients were screened for the presence of auto-antibodies neutralizing type I IFNs. The main results of our study are as follows: (1) auto-Abs were found in at least 10% of patients hospitalized in the ICU, consistent with previous studies; (2) critically ill patients who were found to be positive for auto-Abs against type I IFN did not have a different clinical presentation than others; and (3) they did not have a statistically different mortality at day 28 than patients without auto-Abs.
Our study included a large number of well-phenotyped critically ill COVID-19 patients and primarily aimed at studying the prognostic impact of auto-Abs neutralizing type I IFN. We found that 10% of patients had positive auto-Abs neutralizing type I IFN, consistent with several previous studies reporting positivity for neutralizing auto-Abs against type I IFNs, ranging from 3 to 19% in severe or critical cases [4, 5, 8,9,10,11,12,13, 15, 17, 19]. The number of auto-Ab-positive individuals might have been underestimated given the frequent use of corticosteroids that might lower the auto-Ab level or neutralization capacity. The majority of the patients of our study were positive for auto-Abs against IFN-α2 and IFN-ω, while patients with auto-Abs against IFN-β were less frequent. This result is consistent with previous studies showing that auto-Abs against IFN-β are scarce [11, 13]. Of note, in our study almost all positive patients against IFN-β were also positive for neutralizing auto-Abs against IFN-α2 and IFN-ω (n = 10/12, 83.3%). Such auto-Abs combination could be responsible for the lack of efficacy of sub-cutaneous IFN-β in severe COVID-19 .
We did not find any difference in terms of general characteristics and comorbidities between positive and negative patients. There was a trend for more positive patients being males but not to the extent to that reported in the study of Bastard et al. (94% positivity in males) .
Previous studies demonstrated that the detection of auto-Abs neutralizing type I IFNs in the overall population of COVID-19 patients is a risk factor for developing life-threatening COVID-19 pneumonia (i.e., requiring ICU admission) [5, 8, 9, 13]. We did not test the prevalence of auto-Abs positivity in healthy subjects of in mild or asymptomatic COVID-19 because of the inclusion criteria of our study, but Bastard et al. showed that anti-IFN auto-Abs are almost never found in these two populations with, respectively, 0.33% and 0% positivity rates . Several studies demonstrated that auto-Abs pre-exist the viral infection [5, 12, 28], which we could not confirm as pre-infection samples were not available in our patients. Our study revealed that positive patients required higher FiO2 levels at ICU admission. However, PaO2/FiO2 ratios and need for ventilator support did not differ between groups, suggesting there was no major difference regarding the severity of respiratory disease between groups. Such findings are conflicting with those of other studies focusing on ICU patients reporting more frequent organ failures in patients with positive auto-Abs [12, 15].
Regarding the association between auto-Abs and mortality, there has been discrepancies in the published studies related to the population studied. In the overall COVID-19 population, the auto-Abs positivity is associated with an increased mortality in the majority of studies [9, 10]. However, when focusing on critically ill patients, small studies reported an association between auto-Abs positivity and mortality , while others failed to replicate such results [11, 12]. With a prospective and multicentre design and including a large number of critically ill COVID-19 patients, our study did not find a prognostic impact of auto-Abs positivity neither on 28-day (primary outcome measure), nor on 90-day mortality. Such findings in this cohort of critically ill patients do not obviate the key pathogenic role of auto-Abs against type I IFN in severe COVID-19, but suggest their presence might be more critical during the early phase of SARS-CoV-2 infection (i.e., before ICU admission) than when severe infection is constituted. Indeed, several factors may account for the lack of outcome difference between patients with and without positive auto-Abs. First, key clinical determinants, including organ failures, age, gender, associated comorbidities, have been demonstrated to have a major impact on the outcome of critically ill COVID-19 patients , and may have blunted the deleterious impact of auto-Abs neutralizing type I IFN in critical patients. Second, in our study, critically ill COVID-19 patients who were not found to be positive for auto-Abs against type I IFNs had the same clinical profile in terms of demographics, underlying comorbidities, and organ support than those who were positive, and eventually had the same outcomes. We therefore speculate that the similar clinical picture between patients with positive and negative auto-Abs might be explained by a common mechanism, i.e., an impaired type I IFN production or response. SARS-CoV-2 infection induces a strong innate immune response associated with the production of type I IFNs, triggered by the interaction of pathogen-associated molecular patterns and pattern recognition receptors . However, compared to other severe viral infections (e.g., Influenza A H1N1 infections), severe COVID-19 was shown to be associated with a paradoxically lower type I IFN response. Indeed, severe COVID-19 patients have been characterized by type I IFN deficiency , associated with persistent viral load and a secondary exacerbated inflammatory response. The mechanisms underlying type I IFN deficiency in these critically ill patients who do not carry auto-Abs have to be investigated. Herein, we only studied the role of auto-Abs against type I IFN, but Zhang et al.  showed that 3.5% of critical COVID-19 patients carry significant inborn errors of type I IFN immunity related to loss of function variants, pointing out that a part of critical COVID-19 patients have an impaired IFN response unrelated to auto-Abs. A genetic assessment of all critically ill patients might lead to the identification of a higher number of patients with an identified type I IFN defect. SARS-CoV-2 variants have also been shown to sabotage the body’s IFN response through the production of immune-suppressive proteins . An impaired type I IFN response thus seems to be a crucial determinant of severity at least in the early phase of SARS-CoV-2 infection, but maybe at a lesser extent in the latter stage when the innate immune response has been activated. Indeed, in vitro viral replication was inhibited when cells were pre-treated with type I IFN, but was not modified when exogenous IFN was added after cell infection . This is consistent with the finding that treatment with IFN-β1a did not improve the course of the disease in hospitalized COVID-19 patients . Such therapeutic strategies should thus rather be tested in the early stage of SARS-CoV-2 infection.
Twenty-two percent of women had positive auto-Abs, higher than the 6% rate reported in the first study of Bastard et al. . Auto-Ab positive women displayed a peculiar phenotype, associating a younger age, more invasive mechanical ventilation requirement at ICU admission, and presented more frequently auto-Abs neutralizing IFN-β. Interestingly, they also had a more frequent auto-immune background. ANA have been reported in 14% of COVID-19 patients . Auto-Abs against IFN have previously been identified in the pathogenesis of systemic lupus erythematous  but determining whether ANA positivity is triggered by SARS-CoV-2 infection or if patients with auto-immune background are more prompt to develop auto-Abs against type I IFNs will require more studies. Based on our results, we hypothesize that patients with auto-Abs may have two phenotypes, a first one represented by men over 60 years old without any auto-immune background, as described in the studies by the COVID human genetic effort consortium [5, 8], and another one, which is described here, represented by young women with auto-immune background. In both cases, in the ICU setting, auto-Abs against type I IFN are causal of life-threatening disease and seem to be associated with more severe disease at ICU admission in women, but not with mortality.
Our study has several strengths. Its multicentre and prospective design allowed for studying a representative sample of COVID-19 patients with little missing data. We could precisely assess the clinical phenotype and outcome of patients, who were followed up until day-90 of ICU admission. Finally, auto-Abs have been tested using the previously described reference method .
Our study also has limitations. Patients have been included from the beginning of the pandemics to May 2021. During this time-period, ICU admission strategies and patients’ management varied between the successive COVID-19 waves, introducing potential bias. The SARS-CoV-2 Alpha variant appeared in France and became predominant during the study inclusion period (January–May 2021) . However, the distribution of positive patients did not change over time. We also did not record SARS-CoV-2 vaccination status as there was no anti-SARS-CoV-2 vaccine available at the time the study started. However, on May 1st 2021, when the inclusion period ended, less than 10% of the French population had been fully vaccinated (https://covidtracker.fr/vaccintracker/), implying that the proportion of vaccinated patients in this cohort of critically ill patients was very low. The existence of a control group consisting in COVID-19 negative ARDS patients would have reinforced the findings of the study regarding the prevalence of auto-Abs. Performing immunological tests to assess the relationship between auto-Abs and host response to SARS-CoV-2 infection would have allowed to further explore the pathophysiological role of auto-Abs against type I IFNs in vivo in critically ill COVID-19 patients. Finally, auto-Abs were screened on only one blood sample during ICU stay, potentially underestimating the rate of auto-Abs positivity.