Type 1 interferon auto-antibodies are elevated in patients with decompensated liver cirrhosis

Abstract

Patients with decompensated liver cirrhosis, in particular those classified as Childs-Pugh class C, are at increased risk of severe coronavirus disease-2019 (COVID-19) upon infection with severe acute respiratory coronavirus 2 (SARS-CoV-2). The biological mechanisms underlying this are unknown. We aimed to examine the levels of serum intrinsic antiviral proteins as well as alterations in the innate antiviral immune response in patients with decompensated liver cirrhosis. Serum from 53 SARS-CoV-2 unexposed and unvaccinated individuals, with decompensated liver cirrhosis undergoing assessment for liver transplantation, were screened using SARS-CoV-2 pseudoparticle and SARS-CoV-2 virus assays. The ability of serum to inhibit interferon (IFN) signalling was assessed using a cell-based reporter assay. Severity of liver disease was assessed using two clinical scoring systems, the Child-Pugh class and the MELD-Na score. In the presence of serum from SARS-CoV-2 unexposed patients with decompensated liver cirrhosis there was no association between SARS-CoV-2 pseudoparticle infection or live SARS-CoV-2 virus infection and severity of liver disease. Type I IFNs are a key component of the innate antiviral response. Serum from patients with decompensated liver cirrhosis contained elevated levels of auto-antibodies capable of binding IFN-α2b compared to healthy controls. High MELD-Na scores were associated with the ability of these auto-antibodies to neutralize type I IFN signalling by IFN-α2b but not IFN-β1a. Our results demonstrate that neutralizing auto-antibodies targeting IFN-α2b are increased in patients with high MELD-Na scores. The presence of neutralizing type I IFN-specific auto-antibodies may increase the likelihood of viral infections, including severe COVID-19, in patients with decompensated liver cirrhosis.

Keywords: COVID-19; SARS-CoV-2; auto-antibodies; decompensated liver cirrhosis; type I interferon.

Graphical abstract

Figure 1.

Serum from patients with varying severity of decompensated liver disease does not alter viral infectivity in vitro. (A) Schematic representation of the viral pseudoparticle (PP) assay. (B) Infectivity of SARS-CoV-2 PPs in the presence of serum from patients grouped by Child-Pugh class and analysed using the Mann–Whitney test. (C) Spearman correlation analysis of MELD-Na score and SARS-CoV-2 PP infectivity. (D) Schematic representation of the live virus assay using modified A549 lung epithelial cells. (E) Infectivity of the SARS-CoV-2 BT20.1 strain in the presence of serum from patients grouped by Child-Pugh class and analysed using the Mann–Whitney test. (F) Spearman correlation analysis of MELD-Na score and infectivity of the SARS-CoV-2 BT20.1 strain. Data for the SARS-CoV-2 PP infectivity assay and the A549 clearance assay represent the average from two independent experimental repeats and are expressed as a % of infectivity/clearance observed in the presence of DMEM only. Horizontal lines indicate the group median. ns, not significant. Created with BioRender.com.

Figure 2.

Viral infectivity in vitro in the presence of patient serum is not associated with age or gender. (A) Spearman correlation analysis of chronological age and infectivity of SARS-CoV-2 PP in the presence of serum from patients with decompensated CLD. (B) Infectivity of SARS-CoV-2 PP in the presence of serum from patients grouped by sex and analysed using the Mann–Whitney test. (C) Spearman correlation analysis of chronological age and infectivity of the SARS-CoV-2 BT20.1 strain (measured as clearance of modified A549 cells) in the presence of serum from patients with decompensated CLD. (D) Infectivity of the SARS-CoV-2 BT20.1 strain in the presence of serum from patients grouped by sex and analysed using the Mann–Whitney test. (E) Spearman correlation analysis of infectivity of SARS-CoV-2 PP versus infectivity of the SARS-CoV-2 BT20.1 strain in the presence of serum from patients with decompensated CLD. Data for the SARS-CoV-2 PP infectivity assay and the A549 clearance assay represent the average from two independent experimental repeats and are expressed as a % of infectivity/clearance observed in the presence of DMEM only. Horizontal lines indicate the group median. ns, not significant.

Figure 3.

Inhibition of IFNα signalling is correlated with MELD-Na score but not Child-Pugh class. (A) Type I IFN activity following pre-incubation of IFNα2b (500 pg/ml) in the presence of a 1:40 dilution of serum grouped by Child-Pugh class, analysed using the Mann–Whitney test. (B) Spearman correlation analysis of type I IFN activity following pre-incubation of IFNα2b (500 pg/ml) in the presence of a 1:40 dilution of serum with MELD-Na scores. (C) Type I IFN activity following pre-incubation of IFN-α8 (50 pg/ml) in the presence of a 1:40 dilution of serum grouped by Child-Pugh class, analysed using the Mann–Whitney test. (D) Spearman correlation analysis of type I IFN activity following pre-incubation of IFN-α8 (50 pg/ml) in the presence of a 1:40 dilution of serum with MELD-Na scores. (E) Type I IFN activity following pre-incubation of IFN-β1a (100 pg/ml) in the presence of a 1:40 dilution of serum grouped by Child-Pugh class, analysed using the Mann–Whitney test. (F) Spearman correlation analysis of type I IFN activity following pre-incubation of IFN-β1a (100 pg/ml) in the presence of a 1:40 dilution of serum with MELD-Na scores. Horizontal lines indicate median. ns = not significant.

Figure 4.

Inhibition of IFNα signalling is due to IgG auto-antibodies present in patients with decompensation liver disease but is not associated with sex or age. (A) Type I IFN activity following pre-incubation of IFNα2b (500 pg/ml) in the presence of a 1:40 dilution of serum comparing two cohorts of healthy controls (Table 1) and the liver disease patient cohort, analysed using a Welch ANOVA test. (B) Type I IFN activity following pre-incubation of IFNα2b (500 pg/ml) in the presence of a 1:40 dilution of serum from liver disease patients grouped by sex and analysed using the Mann–Whitney test. (C) Spearman correlation analysis of type I IFN activity following pre-incubation of IFNα2b (500 pg/ml) in the presence of a 1:40 dilution of serum from liver disease patients with age. (D) Type I IFN activity following pre-incubation of IFNα2b (500 pg/ml) with varying dilutions of serum. Serum samples with evidence of inhibition of Type I IFN activity are black (n = 3) and unmatched health controls are grey (n = 3). (E) Comparison of Type I IFN activity following pre-incubation of IFNα2b (500 pg/ml) in paired samples of whole serum, purified IgG and IgG-depleted serum, analysed using the repeated measures one-way ANOVA test. For (A) and (B) the horizontal lines indicate the group median. For (D) and (E) the error bars denoted the standard deviation. **, P-value < 0.01; ***, P-value < 0.001; ns = not significant.

Figure 5.

IgG autoantibodies from patient serum directly bind recombinant IFNα2b. (A) Comparison of IFNα2b ELISA absorbances in serum comparing two cohorts of healthy controls (Table 1) and the liver disease patient cohort, analysed using a Welch ANOVA test. (B) Comparison of IFNα2b ELISA absorbances in serum from patients with liver disease grouped by Child-Pugh class, analysed using the Mann–Whitney test. (C) Spearman correlation analysis of the IFNα2b ELISA absorbance readings in serum from liver disease patients and age. (D) Comparison of IFNα2b ELISA absorbances in serum from patients with liver disease grouped by sex and analysed using the Mann–Whitney test. (E) Comparison of IFNα2b ELISA absorbances in serum samples with evidence of inhibition of Type I IFN activity (‘inhibitory’) versus serum samples lacking inhibition of Type I IFN activity (‘non-inhibitory’). Data were analysed using the Kruskal–Wallis test. (F) Spearman correlation analysis of the type I IFN activity in samples following pre-incubation of IFNα2b (500 pg/ml) in the presence of a 1:40 dilution of serum and IFNα2b ELISA absorbance readings. **, P-value < 0.01; ***, P-value < 0.001; ns = not significant.