Role of autoantibodies targeting interferon type 1 in COVID-19 severity: A systematic review and meta-analysis

Introduction Impairment of the type I interferon (IFN–I) signaling pathway is associated with increased severity of COVID-19 disease. Here we have undertaken a systematic review and meta = analysis on the association between the severity of COVID-19 and IFN-1 autoantibodies (AAbs; aIFN-1, aIFN-α, aIFN-ω, and aIFN-β). Methods Four databases, including Medline [PubMed], Embase, Web of Science, and Scopus, were systematically searched to find papers on the role of aIFN-1 and its subtype AAbs in the severity of COVID-19 infection. Data on the prevalence of aIFN-1, aIFN-α, aIFN-ω, and aIFN-β were pooled using random- or fixed-effects models. Subgroup analysis was performed based on disease severity. Odds ratios (OR) for COVID-19 disease outcome, including length of hospital stay, ICU admission and death, were calculated in relation to positive or negative plasma IFN-1 AAbs. Results A total of 33 studies with 13023 patients were included. The overall prevalence of circulating aIFN-1, aIFN-α, and aIFN-ω AAbs was 17.8 % [13.8, 22.8], 7.2 % [4.7, 10.9], and 4.4 % [2.1, 8.6], respectively, and the overall prevalence of neutralizing aIFN-1, aIFN-α, aIFN-ω, and aIFN-β AAbs was 7.1 % [4.9, 10.1], 7.5 % [5.9, 9.5], 8.0 % [5.7, 11.1] and 1.2 % [0.4, 3.5], respectively. Circulating aIFN-α (OR = 4.537 [2.247, 9.158]), neutralizing aIFN-α (O = 17.482 [8.899, 34.342]), and neutralizing aIFN-ω (OR = 12.529 [7.397, 21.222]) were significantly more frequent in critical/severe patients than in moderate/mild patients (p < 0.001 for all). Anti–IFN–1 was more common in male subjects (OR = 2.248 [1.366, 3.699], p = 0.001) and two COVID-19 outcomes including ICU admission (OR = 2.485 [1.409, 4.385], p = 0.002) and death (OR = 2.593 [1.199, 5.604], p = 0.015) occurred more frequently in patients with positive anti–IFN–1. Conclusion: aIFN-1 and its subtypes AAbs are associated with severe and critical COVID-19 disease and may be a predictive marker for a poor prognosis, particularly in men.


Introduction
Severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) began spreading in Wuhan, China, in late December 2019 and has caused nearly 7 million deaths by May 4, 2023 [1].SARS-CoV-2 infection has shown a broad spectrum of clinical presentations, ranging from asymptomatic to severe disease associated with multiple organ failure and death [2,3].Impairment of antiviral mechanisms such as interferon type 1 (IFN-1) has been associated with the severity of several life-threatening respiratory infections [4], including influenza pneumonia [5] and COVID-19 [6].IFN-1 belongs to a family of cytokines that trigger important innate immune defense mechanisms during viral infection [7].IFN-1 comprises several subtypes: IFN-α, IFN-β, IFN-ω, IFN-ε, and several other subtypes that are structurally related and have a variety of biological functions, including protection against pathogens, immune-regulation, differentiation, and development [8].Their deficiency can lead to viral spread, and their overproduction can cause excessive inflammation, leading to a poor clinical outcome.Impairment of the IFN-1 pathway can be caused either by inherited genetic defects or by the production of autoantibodies (AAbs) directed against IFN-1 [9].Several studies have shown that SARS-CoV-2 induces a low and delayed IFN response in severe and critical patients [2,10,11].Casanova and Anderson showed that inherited or autoimmune deficiencies of IFN-1 are responsible for 15 %-20 % of all critical COVID-19 patients [12].Recently, Goncalves detected anti-IFN-1 (aIFN-1) AAbs in 18 % of critically ill COVID-19 patients, whereas these AAbs were not detected in any of the patients with mild disease [9].
Detection of IFN-1 AAbs in the population infected with the SARS-CoV-2 virus has been proposed as a predictive marker for unfavourable COVID-19 outcomes; therefore, to test this observation, we systematically reviewed the available work on the association between COVID-19 severity and IFN-1 AAbs in the serum of COVID -19 patients.

Search strategy and study selection
This systematic review and meta-analysis were conducted in compliance with PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) protocols [13].The initial search strategy was developed in PubMed and PubMed's Medical Subject Headings (MeSH) and applied to each database searched.We identified articles by searching PubMed, Scopus, Embase, and Web of Science through May 10, 2023, using the following search terms: "COVID-19″ and "autoantibody".A complete list of search strategies for all databases is provided in Table S1.Studies reporting aIFN AAbs in patients with COVID-19 were extracted.We also searched Google Scholar and the reference lists of the included articles to identify further studies.We included only studies in English.

Inclusion and exclusion criteria
All included papers met the following criteria [1]: subjects being diagnosed COVID-19 based on a reverse transcription polymerase chain reaction (RT-PCR), serological tests or typical symptoms [2] at least one of the aIFN-1 or its subtypes AAbs had to have been measured.We excluded all case reports [14].

Eligibility and quality assessment
Four reviewers (A.A., M.A., N.N., and Z.S.) participated in selecting studies based on the articles' title, abstract and full text.In cases where agreement could not be reached, a fifth reviewer (A.S.) reviewed the eligibility to determine final inclusion.Two reviewers (M.A. and N.N.) independently reviewed included articles for quality.Included studies were assessed using the Joanna Briggs Institute (JBI) assessment tools, which provide an assessment tool for most study types, including observational studies [15].Disagreements were resolved by discussion between the investigators or a third reviewer (A.S.).

Data extraction
Study characteristics were extracted from the included articles, including first author's last name, study conduct date, title, study design, study location (country), patient demographics, COVID-19 confirmatory method, IFN autoantibody assay, and history of systemic autoimmune rheumatic disease.The number of individuals in whom circulating or neutralizing aIFN-1/α/ω/β was detected was also recorded.In addition, characteristics of COVID-19 patients in studies comparing seropositive patients with seronegative patients to IFN-1 or its subtypes, including duration of admission, demographics, intensive care unit (ICU)

Statistical analysis
The weighted pooled prevalence and 95 % confidence intervals (CIs) of aIFN-1, aIFN-α, aIFN-ω, and aIFN-β were calculated.The magnitudes of the estimated effects of AAbs on COVID-19 outcomes and the prevalence of AAbs in male patients were expressed as odds ratios (OR), with corresponding 95 % CIs in brackets.The prevalence of AAbs in critical/ severe or moderate/mild groups was compared using OR and 95 % CIs.Sensitivity analysis was done after removing a study conducted on children [16].Median values were converted to means using a formula described by Wan et al. [17].p < 0.05 was considered statistically significant.Heterogeneity was calculated quantitatively using the I2 statistical test.If heterogeneity was high (Cochran's Q < 0.05), the random-effects model was used; otherwise, the fixed-effects model was used.Potential publication bias was examined using funnel plots [18].All statistical analyses were done using Comprehensive Meta-Analysis Software (CMA v.3, Biostat Inc., Englewood, New Jersey, USA).

Circulating and neutralizing aIFN-1/α/ω/β
A total of 13023 patients (from 33 studies) diagnosed with COVID-19 and had at least one of the AAbs targeting IFN-1, IFN-α, IFN-ω, or IFN-β were included in this analysis.Seropositivity of patients with different stages of disease against IFN-1 and its subtypes is detailed in Table 2.
Fourteen studies compared patients with positive and negative AAbs to IFN-1 or IFN-α.Table 3 and Fig. 4 show the association between sex, death, ICU admission, and length of hospital stay with positive serum AAbs against IFN-1 or IFN-α.

Qualitative report of comparison of plasma aIFN-α/ω levels
Soltani-Zangbar et al. reported higher aIFN-α2 levels in severe than mild patients [42], though three other studies showed no difference [33,47,48].Three studies reported higher levels of aIFN-α2 among COVID-19 patients than in the control group [33,42,49], whereas two studies found no difference [47,48].Furthermore, Chen et al. demonstrated higher aIFN-ω levels in severe patients than in mild patients [49], whereas Smith found no difference [47].Only one study found no difference between aIFN-ω levels in COVID-19 patients and the control group [47].Table 1 provides information on the characteristics of the studies reported.

Publication bias
Publication bias was assessed by visual inspection of the asymmetry of the funnel plot.The figures show no evidence of publication bias;     A. Akbari et al.

Fig. 2. (continued).
A. Akbari et al.       A. Akbari et al. however, for some subgroup analyses, there were insufficient studies to present proper funnel plots.Funnel plots are provided for all forest plots.

Discussion
This meta-analysis of 33 studies analyzed data from 13023 COVID-19 patients with respect to aIFN-1 AAbs.We estimated the pooled rate of seropositivity of circulating aIFN-1, aIFN-α, and aIFN-ω AAbs among COVID-19 patients to be approximately 17.8 % [13.8, 22.8], 7.2 % [4.7, 10.9], and 4.4 % [2.1, 8.6], and the rates of neutralizing aIFN-1, aIFN-α, aIFN-ω, and aIFN-β AAbs were 7.1 % [4.9, 10.1], 7.5 % [5.9, 9.5], 8.0 % [5.7, 11.1], and 1.2 % [0.4,3.5], respectively.In contrast, the aIFN-α and aIFN-ω subtypes are present in about 0.3 % of the normal population [50].The discrepancies in seropositivity rates reported by different studies can be attributed to the stage and severity of the disease, the measurement method and the characteristics of the patients.We performed a subgroup analysis in which patients were divided by COVID-19 disease severity that revealed significantly higher rates of aIFN-1, aIFN-α, and aIFN-ω in severe patients.A subsequent analysis provided further confirmation when patients were divided into severe/critical group or mild/moderate group, showed that the odds ratio of being positive for aIFN-α was significantly higher in the critical/severe group than in the mild/moderate group.One explanation for the high aIFN rate in severe patients is that IFN blockade provides an optimal environment for viral replication and disease progression [51].Furthermore, patients with more severe disease are more prone to ICU admission and death, which we demonstrated that higher rates were associated with patients positive for aIFN-1 AAbs, though the association of IFN-1 AAbs with ICU admission and death may just be an epiphenomenon for disease severity.Our findings are supported by large cohorts and previous meta-analyses on this topic [50,52].It has been suggested that aIFN-1, aIFN-α, and aIFN-ω AAbs may be appropriate biomarkers of disease severity and that their elimination by plasma exchange may improve clinical status [53].However, IFN therapy has not been shown to improve mortality or length of hospital stay in COVID-19 patients [54].Furthermore, length of hospital stay was not associated with aIFN-1 AAbs seropositivity, with no difference between positive and negative groups for IFN-α AAbs.However, the small size of the available sample may have led to a type 2 statistical error and a false negative result.The analysis also showed that aIFN-1 AAbs were more frequent in male subjects, which may explain previous findings of an increased incidence of severe disease in male subjects [55,56].The recent meta-analysis by Wang et al. on this topic included 8 studies and concluded that AAbs against IFN-1 accumulate in 5 % of the total population and reach 10 % in severe patients.However, the prevalence of neutralizing and circulating AAbs was not reported separately [52].In agreement with them, we found that aIFN-I AAbs were more prevalent in male subjects.Furthermore, although AAbs can decrease circulating levels of IFN-1 and its subtypes, a meta-analysis by Pires et al. showed that there were no significant differences in peripheral IFN-α in patients with different disease severity [57].However, a significant difference was found between healthy control subjects and COVID -19 patients with mild disease.
Regarding the biomarker potential of aIFN AAbs, Beck et al. [58] suggested that an impaired IFN-1 response could be the hallmark of severe COVID-19.Zhang et al. also reported an association between SARS-CoV-2 infection and defects in genes regulating IFN-1 using extensive genetic sequencing [59].There are two main hypotheses regarding the cause of an elevated aIFN-1.Firstly, AAbs facilitate viral replication because the host response system is neutralized by AAbs [19].In support of this, Trouillet-Assant et al. [60] reported that a decreased plasma concentration of the IFN-1 cytokine predicts a poor prognosis and that patients with the lowest IFN-1 concentration have longer durations of admission.Secondly, viral evasion leads to the increased production of IFN-1, which contributes to the cytokine storm in COVID-19 disease [61], TNF-mediated inflammation [62], and high IFN-1 secretion impairs antibody production [63].Massive production of IFN-1 leads to tissue damage, and autoantibodies may provide impaired protection against the cytokine storm [64].In accordance with            A. Akbari et al. the second hypothesis, single-cell RNA-sequencing studies show hyperactivation of the IFN-1 pathway in severe patients, disrupting immune function, leading to overreaction and eventually resulting in organ damage [65,66].Post-mortem analyzes of COVID-19 patients reveal a high concentration of IFN-1 in lung tissue [67].Menezes et al. also demonstrated that higher levels of IFNB1 transcripts are found in the nasal mucosa of COVID-19 patients and predict a poor outcome [68].
IFN-1 are critical for controlling SARS-CoV-2 infection, which may explain the increased susceptibility of older adults with a reduced ability to secrete IFN-1 [59].The Bodansky et al. study showed a low frequency of positive AAbs against IFN-α2, 0.4 % in children and adolescents [16], and sensitivity analysis on removing this young population in the Bodansky study showed a rate of 7.9 % [5.2, 11.8], indicating that aIFN-α2-AAbs mainly occur in older patients.
Our study had some limitations.Firstly, limited analysis of IFN-ω or aIFN-β levels could be performed as few studies reported them, increasing the risk of a false negative result.Secondly, another limitation is the low sensitivity of the assays used in some studies to measure aIFN-1 and its subtypes, which may have led to underreporting.Assessment of aIFN-1 in the post-COVID-19 phase may provide insights into the role of aIFN-1 AAbs in patients with long-COVID-19 and post-COVID-19 syndrome, and future studies should also investigate aIFN-γ AAbs [48].

Conclusion
In conclusion, aIFN-1 and its subtypes AAbs are associated with severe and critical COVID-19 disease and may be a predictive marker for a poor prognosis, particularly in men.

A.
Akbari et al.

Fig. 4 .
Fig. 4. Association between sex, death, ICU admission, and length of hospital stay with positive serum autoantibodies against IFN-1 or IFN-α.

A
.Akbari et al.

A
.Akbari et al.

Table 1
Characteristics of the included studies.
(continued on next page) A.Akbari et al.admission, and mortality rate, were extracted.Data were stored in a Microsoft Excel spreadsheet (Redmond, WA).Data was collected on the plasma AABs levels in COVID-19 and non-COVID-19 patients or within COVID-19 patients according to disease severity.

Table 2
Seropositivity of patients with different severity of disease regarding aIFN-1 and its subtypes AAbs.

Table 3
Comparison of COVID -19 results in seropositive patients and serunegative patients.