1
Autoimmune anti-DNA antibodies predict disease severity
in COVID-19 patients
Claudia Gomes*
,1
, Marisol Zuniga*
,1
, Kelly A. Crotty
1
, Kun Qian
2
, Lawrence Hsu Lin
3
, Kimon
V. Argyropoulos
3
, Huilin Li
2
, Paolo Cotzia
3
, and Ana Rodriguez
#,1
*Both authors contributed equally
1
Department of Microbiology,
2
Division of Biostatistics, Department of Population Health,
3
Department of Pathology, New York University Grossman School of Medicine, 550 First Ave.
New York, NY-10016, US
#
Corresponding author. Email: ana.rodriguez@nyumc.org, Phone: (516) 701-5881
Abstract
COVID-19 can lead to severe disease and death, however the mechanisms of pathogenesis in
these patients remain poorly understood. High levels of autoimmune antibodies have been
observed frequently in COVID-19 patients but their specific contribution to disease severity and
clinical manifestations remain unknown.
We performed a retrospective study of 115 COVID-19 hospitalized patients with different
degrees of severity to analyze the generation of autoimmune antibodies to common antigens: a
lysate of erythrocytes, the lipid phosphatidylserine (PS) and DNA.
High levels of IgG autoantibodies against erythrocyte lysates were observed in a large
percentage (up to 41%) of patients. Anti-DNA antibodies determined upon hospital admission
correlated strongly with later development of severe disease,
showing a positive predictive value
of 89.5% and accounting for 22% of total severe cases. Statistical analysis identified strong
correlations between anti-DNA antibodies and markers of cell injury, coagulation, neutrophil
levels and erythrocyte size.
Anti-DNA autoantibodies may play an important role in the pathogenesis of COVID-19 and
could be developed as a predictive biomarker for disease severity and specific clinical
manifestations.
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted January 4, 2021. ; https://doi.org/10.1101/2021.01.04.20249054doi: medRxiv preprint
NOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice.
2
Introduction
Infections trigger immune responses that target pathogen antigens, but frequently they also
induce potent autoimmune responses that are characterized by high levels of antibodies
recognizing a variety of host antigens (1). Autoimmune antibodies have been characterized in
viral diseases such as hepatitis C, HIV, and arboviral infections, but also in bacterial and
protozoan infections like tuberculosis, and malaria. Infection-induced autoantibodies can
recognize a variety of self-antigens, including nucleic acids, lipids, and glycoproteins (1).
Autoimmune antibodies can contribute to systemic inflammatory responses and subsequent
tissue damage through different mechanisms, including immune-complex formation,
complement activation, formation of thrombi and/or lysis of uninfected cells (2).
In COVID-19, autoantibodies are found in high levels in a large proportion of hospitalized
patients with severe disease (3-5) and have been associated with the development of autoimmune
pathologies (6), such as thrombocytopenia (7), hemolytic anemia (8), Guillain-Barre (9), and
anti-phospholipid (10, 11) syndromes. Autoantibodies against type I interferons that neutralize
the anti-viral activity of these cytokines have also been identified and may explain disease
severity among a subset of COVID-19 patients (12).
In addition to acute respiratory distress and pulmonary edema, COVID-19 can cause multi-organ
widespread thrombosis and disseminated intravascular coagulation (13). To explore the relation
of autoimmune antibodies to COVID-19 clinical manifestations, we have determined IgG
autoantibody levels in COVID-19 patients’ plasma against a lysate of erythrocytes (as a general
measure of autoreactivity) and two specific antigens that are involved in the autoimmune
pathogenesis of other diseases: anti-phosphatidylserine (PS) (14) and anti-DNA (15). Our
analysis showed that some hospitalized patients present high levels of these autoantibodies and
that there is a strong correlation among these levels in patients, pointing to general autoreactivity
as a common phenomenon in COVID-19 patients. Among hospitalized patients, severity of
disease was strongly correlated with levels of anti-DNA antibodies and weakly with anti-PS.
High-throughput data analysis showed that anti-DNA antibodies correlate strongly with
parameters related to cell injury, coagulation, neutrophil levels and erythrocyte distribution
width, suggesting a role of these autoantibodies in exacerbating COVID-19 clinical course of
disease.
Methods
Bioethics statement: The collection of COVID-19 human biospecimens for research has been
approved by NYULH Institutional Review Board under the S16–00122 Universal Mechanism of
human bio-specimen collection and storage for research. This protocol allows the collection and
analysis of clinical and demographic data. The database used for this project is de-identified.
Study design and participants: This retrospective cohort study included 115 hospitalized cases
of SARS-CoV-2 infection at NYU Langone Health (NYULH) and 20 uninfected controls. All
COVID-19 cases occurred from April until June 2020, during the peak of the COVID-19
pandemic in this area. Control samples used were collected earlier than January 2020, to avoid
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted January 4, 2021. ; https://doi.org/10.1101/2021.01.04.20249054doi: medRxiv preprint
3
possible undiagnosed COVID-19 cases. Patients that needed intubation or required intensive care
monitoring were considered severe.
Sample collection and testing for COVID-19: Sample collection was performed as part of
routine clinical care for COVID-19 patients at NYULH at the time of hospital admission (day 0
or 1). Venous blood samples were collected in standard plasma separation tubes containing
heparin (16 USP units/ml) (BD Diagnostics). Plasma was recovered following centrifugation,
stored at 5ºC for 5 days (in case further clinical testing for the patient was required) before
aliquoting and storage at -80ºC. Assessment of SARS-CoV-2 infection was performed using
nasopharyngeal swabs in clinical PCR assays as described (16).
Clinical parameters analysis: Every patient sample is associated with a list of parameters,
including non-identifiable information (age, sex, race), clinical data and complete
immunological, hematological, and biochemical clinical characterization. The clinical
parameters used for the analysis had data from more than 30 patients (n > 30). The analysis was
performed to determine statistical correlations between the levels of autoantibodies, that were
determined on day 0-1 of admission and clinical parameters determined on day 0-3, to include
some tests that were performed a few days after admission. A parallel analysis was also
performed to analyze statistical significance in the relation between autoantibody levels and the
minimum or maximum value obtained for every laboratory test during each patient stay at the
hospital, with the aim to determine the predictive value of autoantibody levels for the different
clinical parameters.
Data analysis: SQL statements were generated programmatically to select specific records from
NYU deidentified COVID-19 database. WHERE clauses were generated so that only specific
patient records were selected. Records of interest were then exported to Excel using the HUE
SQL query tool. Excel worksheets were generated using the Infragistics Excel Engine™ software
library (Infragistics, Inc., Cranbury, NJ) which synchronized the data records obtained from the
NYU COVID-19 database with deidentified ID numbers which correspond to patient samples. A
simple algorithm was applied to the lab result records to obtain the minimum and maximum
values for each lab test and create new worksheets, designed to be consumed by statistical
analysis software.
Statistical analysis: Data were analyzed using GraphPad Prism v8. Two tailed Spearman
correlation was used to evaluate the association between the different autoantibodies and
between the autoantibodies and clinical parameters. False discovery rate (FDR) was controlled
for multiple comparisons within specific groups. The differences in autoantibody levels between
controls and COVID-19 patients were determined using Mann-Whitney test. The association
between living status and severity of disease with the autoantibodies was performed using
logistic regression, adjusting for age, race, and gender.
ELISA: To measure autoimmune antibodies, Immulon 2HB 96-well ELISA plates were coated
with lysates of human erythrocytes (2x10
3
cells/µl in PBS, from Interstate Blood Bank), PS
(Sigma) at 20
μ
g/ml in 200-proof Molecular Biology ethanol, calf thymus DNA (predominantly
double stranded, Sigma) at 10
μ
g/ml in PBS. Plates were allowed to evaporate for 16 h at 4°C.
PS-coated plates were let evaporate at RT until completely dry. Plates were washed 3 times with
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted January 4, 2021. ; https://doi.org/10.1101/2021.01.04.20249054doi: medRxiv preprint
4
PBS 0.05% Tween-20 and then blocked for 1 h at 37°C with PBS 3% BSA. Plasma from patients
or control was diluted at 1:100 in blocking buffer and incubated in duplicates for 2 h at 37°C.
Plates were washed again 3 times and incubated with a polyclonal goat anti-human IgG-HRP
diluted 1:500 (Invitrogen) for 1 h at 37°C. Plates were washed 4 more times and developed using
TMB substrate (BD Biosciences). The reaction was stopped using Stop buffer (Biolegend) and
absorbance read at 450 nm. The mean OD at 450 nm from duplicate wells was compared with a
reference positive control plasma sample, previously identified as high responder for IgG
autoantibodies, to calculate relative units (RU). To determine IC, plates were coated with C1q
polyclonal antibody (Invitrogen, 1μg/ml in carbonate-bicarbonate buffer) and the same protocol
was followed with the following modifications: plates were blocked for 1.5 h with PBS 0.1%
BSA, plasma and secondary antibody were diluted 1:80 and 1:5000, respectively. All steps were
performed at room temperature. Samples were considered positive for autoantibodies if the
relative units (RU) were greater than the mean plus 3 times the standard deviation of the
controls.
Role of the funding source: The funder had no role in study design, data collection, data
analysis, data interpretation, or writing of the report. The corresponding author had full access to
all data in the study and had final responsibility for the decision to submit for publication.
Results
Patients characteristics: 115 cases of hospitalized patients with SARS-CoV-2 infection with
different degrees of disease severity and 20 uninfected controls were used in this retrospective
study. Patients were stratified into three different groups: non-severe, severe patients that
survived and severe patients that died of COVID-19 (Table I).
Autoantibody levels are elevated in hospitalized patients with COVID-19 compared to
controls: The plasma of controls and COVID-19 patients on day 0-1 was used to determine the
levels of different IgG autoimmune antibodies. As a general measurement of the autoimmune
antibody response, samples were tested for reactivity to a lysate of human erythrocytes (RBCL).
Autoantibodies to the phospholipid PS and DNA were also determined, since they have been
involved in the pathogenesis of other diseases (14, 15).
We observed that COVID-19 patients had significantly higher average levels of circulating anti-
RBCL and anti-PS compared to controls, with 41% of patients testing positive for anti-RBCL
(Fig. 1A). The levels of autoantibodies increased with severity of disease, showing higher levels
in severe patients compared to non-severe or control (Fig. 1B).
In patients with autoimmune diseases, binding of autoantibodies to their antigens results in the
formation of immune-complexes (IC), which are deposited in various tissues frequently leading
to disease (2). The levels of circulating IC were determined in this cohort, finding no significant
differences between controls and COVID-19 patients (Fig. 2).
The levels of the three different autoantibodies that we examined are highly correlated with each
other (Table II), indicating that individual patients tend to have similar levels of different
autoantibodies and suggesting some patients are more prone to autoreactivity.
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted January 4, 2021. ; https://doi.org/10.1101/2021.01.04.20249054doi: medRxiv preprint
5
When the severity of disease or death occurrence were analyzed in regard to the levels of
autoantibodies, we observed a strong correlation of anti-DNA antibody levels with disease
severity (OR = 7.2, p = 0.006, after adjustment for age, race, and sex) (Table III).
We observed that a large proportion of patients that were positive for anti-DNA antibodies upon
hospital admission developed severe disease. Anti-DNA antibodies showed a positive predictive
value of 89% for COVID-19 severity (Table IV), which corresponds to 22% of total severe cases
in this cohort (17 out of 75).
Relationship between autoantibody levels and clinical parameters: We first performed a
statistical analysis comparing the levels of the three determined autoimmune IgG antibodies and
IC with values of laboratory tests assessed in the 115 patients hospitalized in the cohort. These
include data from 118 parameters that were assessed in the context of hematologic, metabolic,
immunological, and biochemical monitoring of these patients during their admission.
No significant correlations were found between the levels of any of the autoantibodies or IC with
any of the clinical parameters measured at day 0-3 (Table S1). However, significant correlations
(rho > 0.3 and p < 0.01) were found between the levels of different autoantibodies and the
maximum value of several laboratory tests performed during each patient stay at the hospital.
The correlations of each autoantibody type were distinct from the others and tended to be
clustered in groups of parameters related to specific disease processes (Fig. 3, Table S2).
Analysis of the minimum values of laboratory tests did not identify any significant correlations
with autoantibody levels (Table S3).
Anti-DNA antibodies correlated strongly with the maximum values of two markers of cellular
injury: lactate dehydrogenase (LDH), which is released by all cell types, and creatine kinase,
which is released specifically by striated muscle cells, suggesting a possible link between anti-
DNA antibodies and cellular lysis (Fig. 3). Significant correlations were also found between
anti-DNA antibodies and absolute numbers of neutrophils, total white blood cell counts and
markers of RBCs volume, such as mean corpuscular volume and red cell distribution width
(RDW). We also observed that maximum values of D-dimer concentration, a biomarker for
coagulation disorders and thrombus formation, correlated significantly with the levels of anti-
DNA antibodies.
Anti-RBCL and anti-PS antibodies correlated only with the total levels of protein in plasma and
parameters related to RBC and platelet levels, respectively. IC levels correlated with parameters
related to kidney function, as it is typical in autoimmune disorders.
It is important to note that no correlation was found between any of the autoantibodies and
cytokines or inflammatory mediators such as C-reactive protein, several interleukins (including
IL6), TNF or interferon-γ.
Discussion
Although autoantibodies have been observed in different viral and non-viral infections (1), the
high percentage of COVID-19 patients with autoantibodies (up to 33% positive for anti-RBCL)
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted January 4, 2021. ; https://doi.org/10.1101/2021.01.04.20249054doi: medRxiv preprint
