Abstract
Although the subject of intensive preclinical and clinical research, controversy on the protective vs. deleterious effect of interferon (IFN) in COVID-19 remains. Some apparently conflicting results are likely due to the intricacy of IFN subtypes (type I: IFN-alpha/beta, type III: IFN-lambda), timing and mode of administration (nebulized/subcutaneous) and clinical groups targeted (asymptomatic/mild, moderate, severe/critical COVID-19). Within the COntAGIouS (COvid-19 Advanced Genetic and Immunologic Sampling) clinical trial, we investigated endogenous type I and type III IFNs in nasal mucosa as possible predictors of clinical outcome in critical patients, as well as their correlation to SARS-CoV-2 viral load, using nCounter technology. We found that endogenous IFN-beta expression in the nasal mucosa predicts clinical outcome, independent of viral replication or Apache II score, and should be considered as a prognostic tool in a precision medicine approach of IFN therapy in COVID-19 clinical management.
INTRODUCTION
Although the subject of intensive preclinical and clinical research, controversy on the protective vs. deleterious effect of interferon (IFN) in COVID-19 remains. Some apparently conflicting results are likely due to the intricacy of IFN subtypes (type I: IFN-alpha/beta, type III: IFN-lambda), timing and mode of administration (nebulized/subcutaneous) and clinical groups targeted (asymptomatic/mild, moderate, severe/critical COVID-19).
Two recent phase-2 clinical trials1,2 reporting the use of type I and type III IFN achieved their primary clinical and virological outcomes in hospitalized and ambulatory COVID-19 patients, respectively. As set forth previously3, understanding the different kinetics of endogenous IFN production in mild and severe COVID-19 patients, relative to viral replication, will help identify the therapeutic window. Thus, endogenous IFN(s) add another layer of complexity to the COVID-19 IFN conundrum but have been understudied in critical (ICU) patients.
PATIENTS AND METHODS
The COntAGIouS trial (COvid-19 Advanced Genetic and Immunologic Sampling; an in-depth characterization of the dynamic host immune response to coronavirus SARS-CoV-2) proposes a transdisciplinary approach to identify host factors resulting in hyper-susceptibility to SARS-CoV-2 infection. Within this prospective clinical trial (NCT04327570), we investigated viral and host transcriptomes in the nasal mucosa of patients with COVID-19 critical disease, admitted to the intensive care unit (ICU) at University Hospitals Leuven, Belgium. Nasal swabs were collected from 57 critical COVID-19 patients on admission in ICU and/or first bronchoscopy. RNA was extracted and mRNA levels for IFN-beta (IFNB1 gene), IFN-alpha (IFNA2 gene) and IFN-lambda (IFNL2 and IFNL3 genes) were quantified by nCounter technology (Nanostring) as previously described4. Viral load (total SARS-CoV-2 transcripts corresponding to Surface glycoprotein, Nucleoprotein, Envelope protein, Membrane protein, ORF1AB, ORF3A and ORF7A) was quantified as described5, using digital transcriptomics (nCounter, Nanostring) previously validated in a large cohort of acute respiratory infection6. All ICU patients received standard-of-care treatment (corticosteroids, anticoagulants, vasopressors and/or antibiotics, in addition to ventilation/ECMO) but none received IFN treatment.
RESULTS AND DISCUSSION
As primary endpoint, we investigated length of stay in the ICU (ranging from 3 to 74 days) with type I and III IFN as predictors, using Kaplan-Meier curves. As shown in Fig. 1A-B-C, IFNB1 transcript levels (Hazard ratio (HR) 0.30 95%CI[0.16-0.56], p=0.0001) but not IFNA2 (HR 0.82 95%CI[0.46-1.50], p=0.53) or IFNL2/IFNL3 transcript levels (HR 0.67 95%CI[0.35-1.27], p=0.22), nor viral load (HR 1.13 95%CI[0.64-1.98], p=0.68, data not shown) predicted the length of ICU stay. Multivariable regression confirmed IFNB1 levels (β=0.45 [0.24-0.67], p=0.0002) and Acute Physiology and Chronic Health Evaluation (Apache) II score (β=1.06 [0.49-1.65], p=0.0009) as independent predictors, whereas viral load, age, gender, BMI or Charlson Comorbidity index were not. Moreover, IFNB1 levels also predicted worse clinical outcome measured by maximal WHO ordinal scale or maximal oxygen support (Mann-Whitney, p=0.027 and p=0.0068, respectively), as well as a composite score (discharge to rehabilitation centre, hospital stay >60 days or death; Mann-Whitney p=0.040). Noteworthy, 45% (5 of 11) of IFNB1-positive patients required ECMO vs. only 9% (4 of 46) of IFNB1-negative patients. The total days on ECMO was also higher in IFNB1+ patients (median 24.0 vs. 10.5 days, Mann-Whitney p=0.016). IFNB1 levels also predicted multi-organ involvement, another hallmark of critical COVID-19, as measured by Sequential Organ Failure Assessment (SOFA) score (median SOFA score 7 for IFNB1-negative vs. 12 for IFNB1+, Mann-Whitney p=0.0072). Surprisingly, IFNB1 levels were not correlated to viral load (Fig. 1D), in contrast to IFNA2 (r=0.45, p=0.0007) and IFNL2/IFNL3 (r=0.47, p=0.0003).
Kaplan-Meier curves of (A) IFNB1-positive vs. IFNB1-negative, (B) IFNA2-positive vs. IFNA2-negative, and (C) IFNL2/3-positive vs. IFNL2/3-negative ICU patients were compared using Log-rank test (***p=0.0001 for IFNB1, not significant for IFNA2, IFNL2/IFNL3 or viral load, not shown). (D) Viral load was correlated to IFNB1, IFNA2, and IFNL2/IFNL3 transcripts (Spearman correlation). Viral load and IFNL2/IFNL3 data were missing from 3 patients (n=54).
In conclusion, endogenous IFN-beta production in the nasal mucosa predicts clinical outcome, independent of viral replication or Apache II score, and should be considered as a prognostic tool in a precision medicine approach of IFN therapy in COVID-19 clinical management.
Data Availability
All data are available in the manuscript or upon request from the authors.