TO THE EDITOR:
The direct antiglobulin test (DAT) detects immunoglobulin or complement bound in vivo to red blood cells (RBCs) and is widely used to diagnose immune-mediated hemolytic anemias. Positive DAT results, with or without clinically evident anemia, have been reported in a subset of patients with various viral infections.1 Very recently, a few cases with simultaneous onset of SARS-CoV-2 infection and autoimmune hemolytic anemia (AIHA) have been described.2,3
During the first weeks of the coronavirus disease 2019 (COVID-19) outbreak, we noticed an increasing frequency of DAT positivity at the Blood Center of Fondazione IRCCS Ca' Granda Ospedale Maggiore Policlinico (Milan, Italy). Therefore, we studied samples from 113 consecutive patients with confirmed COVID-19 that were sent to our laboratory for pretransfusion testing and/or AB0 and Rh typing during a single week (6-13 April 2020). All patients were hospitalized and receiving treatment with multiple drugs (including hydroxychloroquine, heparin, corticosteroids, anti–interleukin-1 biologicals, antivirals, antibiotics, vasopressors, and invasive or noninvasive ventilation). None of them received COVID-19 convalescent plasma. Red cell investigations were performed at the Immunohematology Reference Laboratory of the Department of Transfusion Medicine and Hematology of Fondazione IRCCS Ca' Granda Ospedale Maggiore Policlinico, which is certified by the American Association of Blood Banks. The study was approved by the Institutional Review Board of Fondazione IRCCS Ca' Granda Ospedale Maggiore Policlinico and conducted in accordance with the Declaration of Helsinki.
Red cell antibodies were determined by direct antiglobulin test and indirect antiglobulin test (IAT), using column agglutination technology (ORTHO BioVue system; Ortho Clinical Diagnostics, Raritan, NJ) and polyspecific antiserum (immunoglobulin G [IgG] plus C3d) and specific antiserum (IgG). DAT reactivity was confirmed, and antibody specificity was determined by microcolumn (IgG, IgA, IgM, C3c, C3d; DC-Screening I; Bio-Rad, Cressier, Switzerland). Reagent cells from patients and healthy donors were prepared from EDTA samples by washing the red cell fraction 3 times with saline. Specifically, we prepared 1 panel of 5DAT-negative patients with COVID-19 and 1 panel of 5 blood donors. A commercial panel of RBCs (SURGISCREEN; Ortho Clinical Diagnostics) was also used. Rapid acid elution (Gamma ELU-KIT II; Immucor Inc., Norcross, GA) was used for the recovery of antibodies bound to red cells, and the eluates were tested by IAT.
Fifteen consecutive samples were also tested by flow cytometry DAT. Briefly, packed RBCs were diluted with phosphate-buffered saline to reach a 1:80 suspension, and 10 µL were incubated with 50 µL of fluorescein isothiocyanate–conjugated F(ab′)2 goat anti-human IgG (Invitrogen, Carlsbad, CA) for 45 minutes at room temperature. After washing, samples were acquired using a BD FACSLyric flow cytometer, and the data were analyzed using FACSuite software (both from BD Biosciences, San Jose, CA). Each assay included a positive control (CHECKCELL; Immucor Gamma, Houston, TX) to validate the testing procedure. Results were expressed as median fluorescence intensity.
SARS-CoV‐2 RNA viremia was determined in a subset of consecutive patient samples (N = 10) at the Laboratory of Hematology/Oncology, Istituto Europeo di Oncologia. RNA was extracted from 1200 µL of plasma and 600 µL of packed cellular fraction using a QIAamp Circulating Nucleic Acid Kit (QIAGEN, Hilden, Germany) and tested by SARS-CoV‐2 RNA Droplet Digital PCR System (QX200; Bio-Rad) with Centers for Disease Control and Prevention primers and probes (2019-nCoV Kit; Integrated DNA Technologies, Coralville, IA), with a lower limit of detection of 0.109 copies per microliter.4
A positive DAT was found in 52 of 113 (46%) COVID-19 patients using a microcolumn screening assay. This prevalence of DAT reactivity was substantially higher than that observed at the Blood Center of Fondazione IRCCS Ca' Granda Ospedale Maggiore Policlinico among non–COVID-19 transfusion candidates (<10%; S.V., N.R., and D.P., unpublished data) and that reported by other investigators in hospitalized patients with acute illness (1-15%).1 Forty-six patients (88%) tested positive for IgG only, 4 (8%) tested positive for IgG plus C3d, and 2 (4%) tested positive for C3d only. Positive cases were confirmed by at least another microcolumn method and, in a subgroup, by flow cytometry, ruling out the hypothesis of a technical artifact. These results were obtained before patients received blood transfusions.
In contrast to what is commonly observed in cases of AIHA, no patient sera or eluates (ie, the solution of antibodies recovered from the RBC surface of DAT-positive patients) reacted by IAT against a commercial panel of reagent RBCs or with healthy donors’ RBCs. However, all eluates containing IgG tested positive with a panel of RBCs prepared from DAT-negative COVID-19 patients. These data suggest that the mechanism underlying DAT reactivity involves modifications of the erythrocyte surface during the course of the disease. A possible interpretation is that hyperinflammation in COVID-19 enhances the deposition of complement C3 and the binding of IgG autoantibodies to RBC membranes, which promotes the clearance of damaged RBCs by macrophages.5 In a subset of 78 patients, we could not demonstrate any association between DAT reactivity and the medications administered during hospitalization. Nevertheless, data on medications were incomplete and not prospectively collected; therefore, we cannot rule out the possible contribution of drug-induced mechanisms. We also considered the possibility that DAT reactivity occurs as a consequence of antibodies directed to viral proteins bound to the RBC membrane. However, this seems unlikely, because in all patients that we tested, SARS-CoV‐2 RNA was not detectable in plasma or RBCs using a highly sensitive droplet digital polymerase chain reaction protocol.
We next analyzed the possible relationship between positive DAT results and patients’ clinical and laboratory characteristics retrieved from the electronic clinical record (Table 1). Our data indicate that the presence of membrane-bound immunoglobulins was related to the severity of anemia, because DAT-positive patients had lower hemoglobin concentrations, greater anisocytosis, and needed more transfusions than DAT-negative patients (P < .01 for all). However, total bilirubin and lactate dehydrogenase (LDH) concentrations were not different between DAT-positive and DAT-negative cases, and the association with other hemolysis indicators could not be assessed, because reticulocyte count and haptoglobin are not routinely monitored in patients with COVID-19. Therefore, it is possible that anemia was not caused by extravascular hemolysis and, rather, can be interpreted as a marker of advanced disease. Although the numbers of patients requiring endotracheal intubation and mortality at 30 days were not significantly different in the 2 groups, definitive conclusions on disease outcome would require a prospective study design and perhaps a greater number of observations. It remains to be investigated whether RBC membrane injury could promote the thrombotic complications frequently observed in COVID-19 patients.6
. | All patients (N = 113) . | DAT-positive patients (n = 52) . | DAT-negative patients (n = 61) . | P* . |
---|---|---|---|---|
Age, mean ± SD, y | 63.5 ± 17.1 | 65.2 ± 16.2 | 62.1 ± 17.9 | .17 |
Males, n (%) | 60 (53.1) | 29 (55.8) | 31 (50.8) | .60 |
Length of hospital stay, d | 15.5 (10-26) | 19 (10.2-30) | 14 (9-22.2) | .06 |
Hemoglobin, g/L | 11.2 (8.6-12.7) | 9.8 (7.9-11.7) | 12.2 (9.4-13.1) | .001 |
Red cell distribution width, % | 14 (13.2-15.2) | 14.5 (13.5-15.9) | 13.6 (12.7-14.4) | .001 |
Platelet count, ×109/L | 260 (183-335) | 308 (207-395) | 237 (162-303) | .06 |
Mean platelet volume, fL | 10.7 (10.1-11.5) | 10.6 (9.9-11.5) | 10.9 (10.1-11.5) | .82 |
LDH, U/L | 229 (199.5-343.5) | 247 (203.5-351) | 227.5 (184.5-316) | .39 |
Total bilirubin, mg/dL | 0.51 (0.32-0.71) | 0.52 (0.36-0.80) | 0.48 (0.33-0.66) | .63 |
Serum ferritin, µg/L | 782 (452-1227) | 922 (526-1320.5) | 702 (359.5-1142.7) | .24 |
D-dimer, µg/L | 1269 (711-2427) | 1578 (803.5-3338.5) | 1109 (635-2074.5) | .11 |
C-reactive protein, mg/dL | 4.13 (1.42-10.32) | 6.27 (1.81-10.79) | 2.83 (1.12-9.11) | .12 |
Patients receiving ≥1 transfusion, n (%) | 44 (39) | 27 (51.9) | 17 (27.9) | .009 |
Units of RBCs transfused per patient, n | 0 (0-2) | 1 (0-2) | 0 (0-1) | .005 |
Patients on endotracheal intubation, n (%) | 18 (16) | 11 (22) | 7 (11.5) | .13 |
Mortality at 30 d, n (%)† | 14 (12.4) | 8 (15.4) | 6 (9.8) | .37 |
. | All patients (N = 113) . | DAT-positive patients (n = 52) . | DAT-negative patients (n = 61) . | P* . |
---|---|---|---|---|
Age, mean ± SD, y | 63.5 ± 17.1 | 65.2 ± 16.2 | 62.1 ± 17.9 | .17 |
Males, n (%) | 60 (53.1) | 29 (55.8) | 31 (50.8) | .60 |
Length of hospital stay, d | 15.5 (10-26) | 19 (10.2-30) | 14 (9-22.2) | .06 |
Hemoglobin, g/L | 11.2 (8.6-12.7) | 9.8 (7.9-11.7) | 12.2 (9.4-13.1) | .001 |
Red cell distribution width, % | 14 (13.2-15.2) | 14.5 (13.5-15.9) | 13.6 (12.7-14.4) | .001 |
Platelet count, ×109/L | 260 (183-335) | 308 (207-395) | 237 (162-303) | .06 |
Mean platelet volume, fL | 10.7 (10.1-11.5) | 10.6 (9.9-11.5) | 10.9 (10.1-11.5) | .82 |
LDH, U/L | 229 (199.5-343.5) | 247 (203.5-351) | 227.5 (184.5-316) | .39 |
Total bilirubin, mg/dL | 0.51 (0.32-0.71) | 0.52 (0.36-0.80) | 0.48 (0.33-0.66) | .63 |
Serum ferritin, µg/L | 782 (452-1227) | 922 (526-1320.5) | 702 (359.5-1142.7) | .24 |
D-dimer, µg/L | 1269 (711-2427) | 1578 (803.5-3338.5) | 1109 (635-2074.5) | .11 |
C-reactive protein, mg/dL | 4.13 (1.42-10.32) | 6.27 (1.81-10.79) | 2.83 (1.12-9.11) | .12 |
Patients receiving ≥1 transfusion, n (%) | 44 (39) | 27 (51.9) | 17 (27.9) | .009 |
Units of RBCs transfused per patient, n | 0 (0-2) | 1 (0-2) | 0 (0-1) | .005 |
Patients on endotracheal intubation, n (%) | 18 (16) | 11 (22) | 7 (11.5) | .13 |
Mortality at 30 d, n (%)† | 14 (12.4) | 8 (15.4) | 6 (9.8) | .37 |
Unless otherwise noted, data are median (interquartile range). Data for transfusion requirements were calculated during follow-up after DAT determination. Reference laboratory values were as follows: red cell distribution width, 11.5% to 14.5%; mean platelet volume, 9.5 to 13.1 fL; LDH, 135 to 214 U/L; total bilirubin, 0.12 to 1.1 mg/dL; serum ferritin, 15 to 150 µg/L; D-dimer <500 µg/L; and C-reactive protein <0.5 mg/dL. Numbers in bold type indicate statistical significance.
A χ2 test was used for discrete variables, and a Student t test or a Wilcoxon-Mann-Whitney test was used for continuous variables. Data were compared by generalized linear model (unadjusted).
Mortality at 30 days was calculated from the date of DAT determination. Other data refer to the day of DAT determination.
Finally, the high rate of DAT reactivity can have an impact on pretransfusion testing, because membrane-bound autoantibodies may mask the concomitant presence of RBC alloantibodies in patients who have been recently transfused, and it may complicate or delay the selection of phenotypically matched blood units.7
In conclusion, anti-RBC antibodies were detectable in almost half of the patients with COVID-19 referred to our blood center. The serologic features of DAT reactivity in COVID-19 patients are different from those generally observed in AIHA. Nevertheless, this condition was associated with an increasing frequency of anemia and greater transfusion requirements. Our data add more evidence to the importance of immune-mediated mechanisms in the pathogenesis of COVID-19.8-10
Data sharing requests should be sent to Daniele Prati ([email protected]).
Acknowledgments
This work was supported in part by grants from Fondazione IRCCS Ca' Granda Ospedale Maggiore Policlinico (Ricerca Corrente 2020 and COVID-19 biobank) (L.V. and D.P.) and by grants from Fondazione AIRC per la Ricerca sul Cancro and the Italian Ministry of Health (F.B.).
Authorship
Contribution: A. Berzuini, C.B., L.V., and D.P. wrote the manuscript; C.P., G. Gregato, A.C, L.P., N.R., and F.T. conducted laboratory experiments; and all authors contributed to the acquisition or analysis of data, critically revised the manuscript, approved the final version for publication, and agreed to be accountable for the results published.
Conflict-of-interest disclosure: The authors declare no competing financial interests.
Correspondence: Daniele Prati, Department of Transfusion Medicine and Hematology, Fondazione IRCCS Ca' Granda Ospedale Maggiore Policlinico, Via Francesco Sforza, 35, 20122 Milano, Italy; e-mail: [email protected].
REFERENCES
Author notes
A. Berzuini and C.B. contributed equally to this work.
L.V. and D.P. contributed equally to this work.
Comments
The deleterious effect of SARS-CoV-2 on red blood cells
(1) Berzuini et al.1 described the high frequency of erythrocyte antibodies and the special binding behavior of these IgG antibodies to red blood cells (RBCs) in COVID-19 patients and controls. The Coombs test results suggest modifications of the RBC membranes during the disease.
(2) The first reports regarding the association between AB0 blood groups and the different risk of SARS-CoV-2 infection came from hospitals in China. A few months later, various countries all over the world confirmed the involvement of ABO blood groups in the risk pattern for COVID-19.2 Up until today the underlying mechanism is matter of debate.
SARS-CoV-2 uses the angiotensin-converting enzyme 2 as a receptor for cell entry. Beside the receptor-binding domain of SARS-CoV-2 known to bind the ACE2 receptor, Gallaher3 located a second receptor-binding domain on the outer surface of the N-terminal domain of the spike protein. This binding domain shows sequence- and structural homologies with the lima bean lectin. The lima bean lectin was used for blood typing since it is an anti-A agglutinin which precipitates best with blood group A1 substances, moderately with A2 substances, very poorly with B substances and not at all with blood group 0.4
Due to Gallaher´s hypothesis the additional receptor-binding domain of SARS-CoV-2 selectively binds the N-acetylgalactosamine of blood group A glycoside. Given the expression of ABO antigens in various human tissues 5 the virus spreading might be basically influenced. Other effects of the receptor-binding domain are purely speculative, although the idea suggests itself that the modifications of the RBC membranes are attributable to a lectin activity of this domain. Therefore, the assignment of the Coombs test findings to the ABO blood groups would be of great interest.