🧬 Aging Wiki

Autoantibodies neutralizing type I IFNs are present in ~4% of uninfected individuals over 70 years old and account for ~20% of COVID-19 deaths

12 min read

Autoantibodies neutralizing type I IFNs are present in ~4% of uninfected individuals over 70 years old and account for ~20% of COVID-19 deaths

TL;DR

In a multi-cohort study spanning 34,159 uninfected individuals from the general population plus 3,595 patients with critical COVID-19 (and 623 severe, 1,639 mild/asymptomatic), Bastard et al. (2021) established the population-level age-stratified prevalence of naturally occurring autoantibodies that neutralize type I interferons (IFN-α and/or IFN-ω). Overall prevalence in the uninfected general-population cohort at the high-concentration assay is 0.53% across all ages. Prevalence rises sharply after age 70: from 0.18% in those aged 18–69 years to 1.1% in those aged 70–79 and 3.4% in those aged >80 years (high-concentration assay; 10 ng/mL IFN-α/ω, plasma 1:10 dilution; abstract-level age bands). Finer age bands from the results section: 0.17% below 70 years, 0.9% at 70–75 years, 1.6% at 75–80 years, >4% at 80–85 years (with a decrease to ~2.6% after 85 years). Using a more sensitive lower-concentration assay (100 pg/mL), applied to a subsample of 10,778 uninfected individuals, overall prevalence is 2.3%; age-stratified: ~1% in those <70 years, ~2.3% in 70–80 years, and ~6.3% in those >80 years. These autoantibodies are pre-existing (predating SARS-CoV-2 infection), and account for approximately 20% of COVID-19 deaths and ~20% of critical COVID-19 cases in the over-80s. By contrast, autoantibodies neutralizing IFN-β do not increase with age. This paper is the canonical source for population-level age-stratified anti-IFN-I autoantibody prevalence figures in the type-i-interferon-signaling literature.

Context: companion to Bastard 2020 Science

This paper is a follow-up and large-scale population extension of bastard-2020-anti-ifn-covid-critical (Bastard et al. 2020, Science, doi:10.1126/science.abd4585), which established that ~10% of patients with life-threatening COVID-19 carry pre-existing anti-IFN-α/ω autoantibodies. That 2020 paper noted a modest age gradient within the COVID-19 cohort (8.5% in patients <65 years vs. 13.0% in those ≥65 years), but did not measure population-level prevalence in uninfected individuals across age bands. This 2021 paper fills that gap with a substantially larger general-population cohort and fine-grained age stratification.

The two papers together establish:

  1. 2020 Science: Anti-IFN-I autoantibodies cause life-threatening COVID-19 (causal inference via prospective pre-COVID samples) and affect ~10% of critical COVID-19 cases.
  2. 2021 Science Immunology (this page): Population-level prevalence of these autoantibodies rises monotonically with age, reaching ~4% (high-concentration assay) to ~6.3% (sensitive assay) in individuals over 80 years old.

Methods

Assay methodology

Anti-IFN-I autoantibodies were measured by functional neutralization assay, not simple binding ELISA. Specifically:

  • High-concentration assay (used for the uninfected general-population cohort): plasma diluted 1:10, incubated with recombinant IFN-α2 or IFN-ω at 10 ng/mL; residual IFN-I activity measured by ISRE-luciferase reporter assay — HEK293T cells transfected with (i) a plasmid containing five IFN-stimulated response elements (ISRE) driving firefly luciferase, and (ii) a constitutively-expressed Renilla luciferase normalization plasmid. Samples considered neutralizing if firefly luciferase induction (normalized against Renilla) fell below 15% of the median of non-neutralizing control samples. Initial screening by Gyros immunoassay (ELISA-like, bead-based high-throughput detection of IgG auto-Abs); all Gyros-positive and intermediate samples then retested by luciferase neutralization assay.
  • Lower-concentration assay (used for a subsample of 10,778): same ISRE-luciferase setup but recombinant IFN at 100 pg/mL (approximately physiological; corresponds to 1 ng/mL IFN in circulating blood at 1:10 plasma dilution), detecting autoantibodies with lower titer or lower avidity that are missed by the high-concentration assay. All available samples from the extended cohort were retested at this concentration.

This functional assay is more clinically relevant than a binding assay: it identifies only autoantibodies that can actually impair IFN-I signaling. The detection threshold difference between the two assays explains why the lower-concentration assay finds approximately 3-fold higher prevalence in all age groups. needs-replication — standardized clinical-grade assays for anti-IFN-I autoantibodies are not yet widely available; replication across independent labs requires assay harmonization.

Cohort composition

The study is a multi-cohort consortium effort spanning the COVID Human Genetic Effort (CHGE), the SFr1 cohort, and additional collaborating centers across Europe, Asia, and the Americas:

  • Uninfected general-population cohort (primary prevalence source): 34,159 individuals; age range approximately 18–99+ years; from blood banks, population biobanks, and pre-pandemic healthy controls. Fine-grained age-band breakdown was extracted from this cohort for the population-level prevalence figures.
  • Subsample (sensitive assay): 10,778 uninfected individuals from the above, tested with the lower-concentration assay.
  • Asymptomatic/mild COVID-19 controls: 1,639 individuals with proven SARS-CoV-2 infection (positive PCR and/or serology, or typical symptoms) who remained asymptomatic or developed only mild, ambulatory illness without pneumonia. These were used as controls for the COVID-19 severity comparisons (1% prevalence of anti-IFN-α/ω autoantibodies in this group by the 100 pg/mL assay).
  • Critical COVID-19 cohort: 3,595 patients enrolled with critical COVID-19 pneumonia (requiring high-flow oxygen >6 L/min, mechanical ventilation, cardiovascular shock, or ICU admission). Of these, 3,136 had samples available for the 10 ng/mL neutralization assay; all 3,595 were available for the 100 pg/mL assay.
  • Severe COVID-19 cohort: 623 patients enrolled with severe COVID-19 (low-flow oxygen <6 L/min); 522 had samples available for the 100 pg/mL assay (the number reported in the abstract for the 6.5% prevalence figure).
  • Deceased cohort: 1,124 patients who died of COVID-19 (aged 20 days to 99 years; mean age 71 years; 73% male). These patients are a subset of the critical COVID-19 cohort — they are not enrolled separately.

Total subjects across all cohorts: 40,016 (per Methods: 34,159 healthy controls + 3,595 critical COVID-19 + 623 severe COVID-19 + 1,639 asymptomatic/mild COVID-19). The 1,124 deceased are a subset of the 3,595 critical patients, not a separate group for enrollment counting. Not all enrolled subjects were available for both assays — the critical COVID-19 cohort available for the 10 ng/mL neutralization assay was 3,136 (of the 3,595 enrolled); the full 3,595 were available for the 100 pg/mL assay.

Key results

Population-level age-stratified prevalence (primary finding)

High-concentration assay (10 ng/mL IFN-α/ω; n=34,159 uninfected):

Overall prevalence across all ages: 0.53% (181 of 34,159 individuals neutralized 10 ng/mL IFN-α2 and/or IFN-ω).

Broad age bands (from abstract):

Age groupPrevalence (anti-IFN-α and/or -ω)
18–69 years0.18%
70–79 years1.1%
>80 years3.4%

Finer age bands (from results text; Figure 5C–F):

Age groupPrevalence (anti-IFN-α and/or -ω)
<70 years0.17%
70–75 years0.9%
75–80 years1.6%
80–85 years>4%
>85 years~2.6% (decrease from peak)

Note: the widely-cited “~4% over 70” approximation in the paper’s title and abstract conflates these finer bands. The prevalence peaks at 80–85 years (>4%) before declining after 85 years — a pattern the authors note may reflect healthy-survivor bias in blood donors aged >85.

Lower-concentration assay (100 pg/mL; subsample n=10,778 uninfected):

Overall prevalence across all ages: 2.3% (from Table S1).

Age-stratified:

Age groupPrevalence (anti-IFN-α and/or -ω)
<70 years~1%
70–80 years~2.3%
>80 years~6.3%

The increase is monotonic with age and accelerates after 70 years. Anti-IFN-β autoantibodies were also measured and did not show a similar age-associated increase — suggesting that the aging-related loss of self-tolerance is specific to IFN-α/ω and not a general type I IFN autoimmune phenomenon. needs-replication — the mechanistic basis for IFN-α/ω specificity vs. IFN-β sparing is not established.

Critical COVID-19 and mortality attribution

All figures below are for the lower-concentration assay (100 pg/mL, plasma 1:10) unless otherwise noted. The high-concentration assay (10 ng/mL) detected autoantibodies in ~9.8% (307 of 3,136) of critical patients tested and ~13.3% of deceased patients.

  • 13.6% of 3,595 patients with critical COVID-19 (N=489 positive; 100 pg/mL assay) had neutralizing anti-IFN-α/ω autoantibodies.
  • 21% of 374 patients >80 years with critical COVID-19 had these autoantibodies (100 pg/mL assay).
  • 6.5% of 522 patients with severe (non-critical) COVID-19 had these autoantibodies (N=34; 100 pg/mL assay).
  • 18.5% (rounded to “18%” in the abstract) of 1,124 deceased COVID-19 patients had these autoantibodies (100 pg/mL assay); 13.3% by the 10 ng/mL assay.
  • These autoantibodies account for approximately 20% of COVID-19 deaths and approximately 20% of critical COVID-19 cases in those over 80 years old.

Autoantibodies predate SARS-CoV-2 infection

For four unrelated patients who suffered critical COVID-19 (including one who died), samples collected before COVID-19 were available and tested positive for neutralizing anti-IFN-I autoantibodies. The autoantibodies were present at collection dates predating the pandemic. Additionally, two further pre-COVID-19 samples from previously described patients (from the companion Bastard 2020 paper) confirmed pre-existence. Together, six patients had confirmed pre-existing anti-IFN-I autoantibodies before SARS-CoV-2 infection. This is a critical confound-exclusion: the autoantibodies are not generated in response to COVID-19 disease, they are a pre-existing vulnerability.

IFN-β anti-autoantibodies: separate prevalence pattern

An additional 1.3% (N=23 of 1,773 tested) of critical COVID-19 patients carried anti-IFN-β autoantibodies neutralizing high concentrations of IFN-β (10 ng/mL). 0.9% of deceased patients also carried these. These were not found to increase with age in the uninfected population (n=9,583 tested for anti-IFN-β; overall prevalence 0.26%) — a specificity finding that helps narrow the mechanistic hypothesis (not a general type I IFN self-tolerance failure, but specifically IFN-α/ω). Anti-IFN-β did not vary significantly with age (Firth’s multivariable logistic regression P=0.68).

Aging-context interpretation

This paper establishes that a quantitatively significant fraction of the elderly population (~4–6% in those >70 years by sensitive assay) harbors pre-existing functional defects in the type I IFN axis — not due to reduced IFN-I production, but due to autoantibody-mediated neutralization of secreted IFN-I effectors. This is mechanistically distinct from the aging-associated chronic overactivation of type I IFN production described in the context of cGAS-STING and retroelement reactivation (see type-i-interferon-signaling § Aging-specific dysregulation).

Two paradoxical aging phenotypes thus coexist in the same pathway:

  1. Chronic tonic IFN-I production (from cGAS-STING activation by leaked mtDNA, nuclear chromatin fragments, and LINE-1 cDNA) — affects most aged individuals and drives inflammaging.
  2. Anti-IFN-I autoantibody neutralization of secreted IFN-I — affects ~4–6% of individuals aged >70 years, creating a functional IFN-I deficiency despite potentially elevated IFN-I production.

The coexistence of these two phenomena in the same individual is possible and could create a paradoxical state: high ISG transcription from intracellular IFN-I signaling loops (paracrine), but impaired extracellular IFN-I effector function (autocrine/paracrine IFN-α/ω neutralized by circulating autoantibodies). Whether these two states are mutually exclusive or can overlap is not established. no-mechanism

This finding is highly relevant to disabled-adaptive-immunity — the gradual loss of self-tolerance to ubiquitous self-antigens (including IFN-α/ω) represents a form of autoimmunity that mimics immunodeficiency in its functional consequences. The clinical implication is that a fraction of elderly individuals are constitutively immunocompromised against viral respiratory pathogens in a way that is invisible to standard immunological workups.

Extrapolation table

DimensionStatusNotes
Pathway conserved in humans?yesDirectly measured in humans
Phenotype conserved in humans?yesDirect human epidemiology; pre-pandemic samples confirm pre-existing nature
Replicated in humans?partialLarge consortium but mostly from European cohorts; replication in Asian/African population cohorts needed. Fernbach 2024 (JEM) independently confirmed age-related onset of anti-IFN-I autoantibodies in a longitudinal HIV cohort (median onset ~63 years) 1, supporting temporal emergence with aging.

Limitations and gaps

  • Assay standardization: The functional neutralization assay is not yet standardized for clinical use across laboratories. Different dilution factors and IFN concentrations yield different prevalence estimates (the 3x difference between the high- and low-concentration assays in this paper illustrates this problem). Prevalence figures are therefore assay-dependent. dose-response-unclear
  • Population representativeness: The uninfected cohort of 34,159 is large, but is predominantly European-ancestry and was assembled from blood banks and biobanks with possible healthy-volunteer bias. Prevalence in diverse global populations, in institutionalized elderly, or in those with comorbidities is unknown.
  • Mechanistic basis for age-related loss of tolerance: The paper establishes the epidemiology but does not identify why IFN-α/ω self-tolerance is specifically lost with aging. Thymic involution, regulatory T cell dysfunction, or accumulation of B cell clones with autoreactive receptors are candidate mechanisms, but are not tested here. no-mechanism
  • IFN-β sparing: The mechanistic reason why anti-IFN-β autoantibodies do not increase with age while anti-IFN-α/ω autoantibodies do is not explained. IFN-β shares the same receptor (IFNAR1/IFNAR2) but differs structurally. Whether this reflects different central vs. peripheral tolerance mechanisms, different protein abundance in thymus, or different structural similarity to foreign antigens is unknown. no-mechanism
  • Clinical consequence quantification: “~20% of COVID-19 deaths” is estimated from the proportion of deaths carrying the autoantibody (18.5% of 1,124 deceased patients by the 100 pg/mL assay; rounded to “18%” in the abstract), which is close to 20% attribution. However, this does not account for confounders (individuals with anti-IFN autoantibodies may differ in other ways from those without). The causal fraction estimate is compelling but not a formal causal analysis.
  • Clinical implications: Whether population-level screening for anti-IFN-I autoantibodies in elderly individuals would inform vaccination or antiviral prophylaxis decisions (e.g., antibody replacement therapy, type III IFN treatment which is not neutralized by these autoantibodies) is unresolved. long-term-unknown

Cross-references

  • type-i-interferon-signaling — canonical home for IFN-I pathway biology; this study provides the population-prevalence figures cited in § Anti-IFN-I autoantibodies in elderly
  • disabled-adaptive-immunity — age-related loss of self-tolerance is a form of immune dysfunction
  • chronic-inflammation — anti-IFN-I autoantibodies modulate but do not directly cause inflammaging; context for the IFN-I paradox
  • bastard-2020-anti-ifn-covid-critical (implicit stub — study page not yet seeded) — the 2020 Science companion paper establishing causality of anti-IFN-I autoantibodies in life-threatening COVID-19 at the cohort level
  • ifnar1 (implicit stub) — the IFNAR1 subunit is blocked by anifrolumab; anti-IFN-I autoantibodies act at the IFN-α/ω level upstream of IFNAR
  • stat1 (implicit stub) — downstream effector of IFN-I signaling; impaired activation in anti-IFN-I autoantibody carriers

Footnotes

Footnotes

  1. doi:10.1084/jem.20240365 · Fernbach N et al. · Journal of Experimental Medicine 2024 · n=1,876 HIV-positive individuals followed longitudinally over 35 years · observational (prospective cohort) · model: humans · anti-IFN-I autoantibodies emerged in approximately 1.9% of individuals as they aged; median onset approximately 63 years; once detected, autoantibodies persisted lifelong and titers increased over decades; age-related loss of self-tolerance preceded the triggering event; study population (HIV-positive) differs from the general uninfected population in Bastard 2021 but independently supports age-related emergence trajectory ↩

Graph View

Backlinks