Abstract
Background Autoimmunity has been reported in patients with severe coronavirus disease 2019 (COVID-19). We investigated whether anti-nuclear/extractable-nuclear antibodies (ANAs/ENAs) were present up to a year after infection, and if they were associated with the development of clinically relevant post-acute sequalae of COVID-19 (PASC) symptoms.
Methods A rapid-assessment line immunoassay was used to measure circulating levels of ANAs/ENAs in 106 convalescent COVID-19 patients with varying acute phase severities at 3, 6 and 12 months post-recovery. Patient-reported fatigue, cough and dyspnoea were recorded at each time point. Multivariable logistic regression model and receiver operating curves were used to test the association of autoantibodies with patient-reported outcomes and pro-inflammatory cytokines.
Results Compared to age- and sex-matched healthy controls (n=22) and those who had other respiratory infections (n=34), patients with COVID-19 had higher detectable ANAs at 3 months post-recovery (p<0.001). The mean number of ANA autoreactivities per individual decreased between 3 and 12 months (from 3.99 to 1.55) with persistent positive titres associated with fatigue, dyspnoea and cough severity. Antibodies to U1-snRNP and anti-SS-B/La were both positively associated with persistent symptoms of fatigue (p<0.028, area under the curve (AUC) 0.86) and dyspnoea (p<0.003, AUC=0.81). Pro-inflammatory cytokines such as tumour necrosis factor (TNF)-α and C-reactive protein predicted the elevated ANAs at 12 months. TNF-α, D-dimer and interleukin-1β had the strongest association with symptoms at 12 months. Regression analysis showed that TNF-α predicted fatigue (β=4.65, p=0.004) and general symptomaticity (β=2.40, p=0.03) at 12 months.
Interpretation Persistently positive ANAs at 12 months post-COVID are associated with persisting symptoms and inflammation (TNF-α) in a subset of COVID-19 survivors. This finding indicates the need for further investigation into the role of autoimmunity in PASC.
Abstract
Circulating antinuclear autoantibodies were detected in COVID-19 patients up to 12 months post-recovery. The presence of these autoantibodies was associated with persisting symptoms and residual inflammation. https://bit.ly/3AduU02
Introduction
The majority of patients infected with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) virus recover; however, a significant subset report persistent symptoms (e.g. fatigue, dyspnoea and cognitive impairment) that do not resolve after infection [1]. This constellation of symptoms is called long COVID or post-acute sequelae of COVID-19 (PASC) and has been observed in 10–20% of convalescent patient cohorts [2]. SARS-CoV-2 infections are associated with the development of autoantibodies during the acute phase of disease, and these contribute to coronavirus disease 2019 (COVID-19) pathology [3–6]. Emerging evidence suggests that the failure to resolve these autoantibodies, or generating de novo pathogenic autoimmune responses post-recovery contributes to PASC with evidence of residual inflammatory cytokines [7–9]. Although it is not known if these autoantibodies are harbingers of emerging autoimmune disease, there have been many case reports of development of autoimmunity post-COVID with no prior personal or family history of autoimmunity [10]. To date, diverse autoantibody signatures including anti-nuclear/extractable-nuclear autoantibodies (ANAs/ENAs) have been reported in PASC patients as biomarkers, but there are no identified links with specific PASC symptoms [7, 8, 11, 12].
In our noninterventional, observational, longitudinal study, we utilised an extensive, clinically relevant ANA/ENA panel to serve as common rheumatological biomarkers of post-COVID trajectory of developing/sustaining PASC symptoms. We investigated circulating levels of ANAs in COVID-19 survivors with varying acute phase severities longitudinally, at 3, 6 and 12 months post-recovery. We further examined the temporal association between ANAs with COVID-19 pathology-associated inflammatory and vascular factors (e.g. tumor necrosis factor (TNF)-α, D-dimer), as well as commonly reported PASC symptoms of fatigue, cough and dyspnoea [1].
Methods
Study design and patient selection
This was a multicentre, multi-time-point observational study approved by the Hamilton integrated research ethics board (#11471, 13181) and the University of British Columbia clinical research ethics board (#H20-01239). Between August 2020 and September 2021, we enrolled 106 COVID-19 patients from St Joseph's Healthcare Hamilton (n=44; Hamilton, ON, Canada), Vancouver General Hospital (n=42; Vancouver, BC, Canada), and St Paul's Hospital (n=20; Vancouver, BC, Canada) for three study visits at 3, 6 and 12 months, post-recovery (deemed recovered as per public health guidelines). Consenting patients aged ≥18 years, with a positive PCR test for SARS-CoV-2 and no previous diagnosis of autoimmune disease were recruited via community self-referrals, physician referrals and hospital inpatient/outpatient post-discharge follow-ups. In order to compare whether autoantibodies differed between individuals who had COVID-19 compared to other respiratory infections, we enrolled 34 individuals with respiratory symptoms consistent with COVID-19 but did not have a test or did not become seropositive between 1 and 3 months post-infection [13]. 22 age- and sex-matched non-COVID, nonvaccinated healthy adults were recruited locally from Hamilton (ON, Canada) (figure 1). Criteria for recruiting healthy volunteers during the pandemic included never having had COVID-19, not yet vaccinated for COVID-19, never-smokers, with no history of respiratory/rheumatological disease. Serum was collected at each time point and stored at −80°C within 1 h of collection for fluid phase analysis.
Symptom assessment
In addition to patient demographics, the following symptoms were recorded by study coordinators via analogous research protocols at all recruitment sites for each time point post-COVID recovery: fatigue (patient-reported or fatigue assessment scale), cough (patient-reported) and shortness of breath (modified Medical Research Council dyspnoea scale).
Microarray autoantibody profiling
Serum IgG and IgM antibody reactivities against 102 autoantigens were analysed using a microfluidic antigen array developed at the Microarray and Immune Phenotyping Core Facility at the University of Texas Southwestern Medical Center, as described previously [14] for the 3-month time point for participants who consented to third-party off-site exploratory analysis. Serum samples from 22 healthy controls with no previous history of autoimmune disease were used to determine the cut-off threshold for IgG and IgM autoantibody reactivity, calculated via median plus three standard deviations, to each of the 102 common self/autoantigens on a microarray panel (supplementary figure E1). We utilised these cut-off thresholds to determine the number of autoreactive antibodies in serum of 36 convalescent post-COVID patients at 3 months post-recovery.
Detection and quantification of ANAs/ENAs in serum
An ANA/ENA line immunoassay (IMTEC-ANA-LIA-MAXX; Human Diagnostics, Germany) targeting 18 common self-antigens was used, as described previously [15] at disease-modifying titres of 1:100. Each strip was scanned (ChemiDoc MP Imaging System; Bio-Rad, CA, USA), and images were converted into eight-bit greyscale and inverted with ImageJ analysis software. A quantitative value was derived for each visible band and normalised to the cut-off control band to provide a mean quantitative value (MQV) with values ≥1.0 indicating positive reactivity. Validation of our quantification method was assessed using indirect immunofluorescence of HEp-2 cells (supplementary figure E2).
Serum molecular mediator analysis
Acute markers of inflammation (interleukin (IL)-1β, IL-6, IL-8, TNF-α) [16], coagulation mediators (D-dimer, E-selectin, intercellular adhesion molecule (ICAM)-1), vascular cell adhesion protein (VCAM)-1) and C-reactive protein (CRP) were assessed and quantitated using the Ella Automated Immunoassay System (Bio-Techne, MN, USA). Serum samples were diluted as per manufacturer's protocol for each mediator reported [13].
Statistical analysis
All experimental data were analysed and plotted using GraphPad Prism (version 9; La Jolla, CA, USA). After testing for normality, statistical comparisons between groups were performed by Mann–Whitney t-test (nonparametric unpaired analysis, two groups), Kruskal–Wallis test (nonparametric unpaired analysis, more than two groups) and Friedman test (nonparametric paired analysis with repeated measures, more than two groups); associations were determined by Spearman's rank correlation test, and categorical variables analysed using Chi-squared analysis. Receiver operating curves, multiple, and simple logistic regression using the “stats” package in the R software generated models to determine which autoantibodies or cytokines significantly predicted symptoms. p-values <0.05 were considered significant unless stated otherwise. The “pheatmap” package was used to produce the heatmaps of the estimated coefficients at the three time points.
Results
Study population
We recruited 106 convalescent COVID-19 patients (61 males, 45 females) with a mean age of 57 years and body mass index (BMI) of 27.2 kg·m−2 (table 1). 26 patients recovered from COVID-19 at home, 35 were admitted to the intensive care unit (ICU) and 45 were hospitalised, but not admitted to the ICU. 34 patients (10 male, 24 female) with a non-COVID-19 infection were recruited for comparison with a mean age of 46 years and BMI of 22.2 kg·m−2. Of these patients, 34 recovered at home, and one was admitted to the ICU. 22 healthy volunteers (11 male, 11 female) with a mean age 49 years and BMI 26.6 kg·m−2, were enrolled as a control population.
IgG/IgM autoantibodies in patients 3 months post-COVID
To determine whether circulating autoantibodies were higher in convalescent COVID-19 patients compared to uninfected controls we used an autoantigen microarray [17] that detects both IgG and IgM autoantibodies in 36 convalescent COVID-19 at the 3-month time point post-recovery and compared it to the 22 healthy donors. Heatmaps showing detected IgG and IgM autoreactivities are given in supplementary figure E1. While most of the healthy controls did not have IgG autoantibodies (20 out of 22, 91%), approximately one-third of the convalescent COVID-19 group had at least one autoreactive IgG (13 out of 36, 36%; p=0.03). Two or more autoantigens were found in 33% of COVID-19 convalescent patients (12 out of 36, 33%; p=0.002) (figure 2a). In contrast, the majority of healthy controls (21 out of 22, 95%) and convalescent COVID-19 patients (33 out of 36, 92%; p>0.05) did not have autoreactive IgM antibodies (supplementary figure E1b). IgG autoantibodies were detected against 21 (21%) of the 102 screened antigens, of which nine (43%) out of 21 were against ANAs with known pathogenic roles in various autoimmune diseases (e.g. anti-dsDNA in systemic lupus erythematosus, anti-SS-B/La in Sjögren syndrome) (figure 2b). Strong significant correlations were observed between the five most prevalent autoreactivities (ACE2, MDA5, CD255, SS-B/La and PM/Scl-75) including two ANAs or ENAs (figure 2c).
ANAs/ENAs in patients 3 months post-COVID-19
The convalescent COVID-19 patients had higher levels (p<0.05) compared to healthy controls for 16 out of 18 ANAs/ENAs and to the non-COVID-19 infection control group for 12 out of 18 ANAs at 3 months post-recovery (supplementary figure E3). We compared the number of positive ANAs (MQV ≥1.0) at 3 months post-recovery between the healthy (figure 2d) and the non-COVID-19 infection group (figure 2e) against convalescent COVID-19 patients recovered at home (PCI-home; n=26 (figure 2f), hospitalised non-ICU (PCI-hosp; n=45) (figure 2g) and those who were admitted to the ICU (PCI-ICU; n=35 (figure 2h). The prevalences of ANAs were not different between the healthy and non-COVID-19 respiratory infection groups. However, each had significantly fewer circulating ANAs/ENAs compared to PCI-hosp (p<0.0001) and PCI-ICU (p<0.0001) populations (figure 2i, supplementary figure E3). The PCI-home patients exhibited a higher number of ANA/ENA reactivities than the infection control group (p=0.047), yet also significantly fewer reactivities than the PCI-ICU group (p=0.004) (figure 2i). Patients with COVID-19 who had a more severe acute phase developed a stronger autoimmune response still evident at 3 months post-recovery (figure 2j).
Changes in circulating levels of ANAs/ENAs up to 12 months post-COVID-19
The majority of convalescent COVID-19 patients had two or more ANAs/ENAs at the 3-month (84 out of 106, 79%) and 6-month (76 out of 98, 78%) time points, and this was reduced to 41% by 12 months (34 out of 58, 41%; p<0.0001) (figure 3a). When stratified according to their acute-phase severities, this observation was consistent within the PCI-hosp (p<0.001) (figure 3c) and PCI-ICU (p<0.0001) (figure 3d) populations, but absent in the PCI-home group (figure 3b). Although we found no difference in MQVs between 3 and 6 months post-COVID-19 for all ANAs/ENAs, a significant attenuation of autoantibody levels at 12 months post-COVID-19 was observed for 13 out of 18 ANAs/ENAs. Although the overall number of detectable ANAs/ENAs declined by 12 months post-recovery (figure 3e–g), some remained detectable: anti-SmD1 (11%) (figure 4d), anti-PCNA (9%) (figure 4e), anti-SS-A/Ro60 (12%) (figure 4g), anti-SS-B/La (21%) (figure 4i), anti-U1-snRNP (30%) (figure 4l), anti-PM-Scl (21%) (figure 4o), anti-Ku (11%) (figure 4q) and anti-DFS70 (12%) (figure 4r). Furthermore, 12% of the positive ANA/ENA reactivities observed at 12 months were previously below cut-off threshold, underlining a potential de novo autoantibody production at this time (figure 3h).
Relationship between ANAs/ENAs and symptoms in post-COVID patients
At 3 months post-recovery, 36% presented with persistent fatigue, 21% with cough and 26% for dyspnoea (table 1). Although cough (6 months, 23%; 12 months, 22%) and dyspnoea (28%, 25%) remained consistent over time, the frequency of fatigue decreased over time (6 months, 40%; 12 months, 20%). However, in individuals who had at least one symptom, the ANA/ENA frequencies remained high throughout the follow-up period (3 months, 54%; 6 months, 77%; 12 months, 50%). Heatmaps of z-scores were generated from simple logistic regression analyses performed for individual ANAs/ENAs per symptom at each time point. The two most prevalent ANAs/ENAs at 12 months, anti-U1-snRNP (p=0.028) and anti-SS-B/La (p=0.003), both positively predicted persisting symptoms of fatigue and dyspnoea (anti-U1-snRNP p=0.02; anti-SS-B/La p=0.007) (figure 5d–f). Anti-U1-snRNP (p<0.007) (figure 5g) and anti-SS-B/La (p=0.002) (figure 5j) levels were higher in patients who reported fatigue compared to those who did not. ANAs/ENAs were unremarkable between patients with cough compared to those without (figure 5h,k). Although anti-U1-snRNP antibodies were slightly higher in patients with dyspnoea (p=0.09) (figure 5i), there was no difference in circulating anti-SS-B/La antibodies (figure 5l). There was a positive correlation between anti-SS-B/La and all three symptoms, as well as for anti-U1-snRNP with fatigue and dyspnoea (figure 5g–l). The presence of either of these two ANAs at 12 months post-recovery predict fatigue (92% specificity, 70% sensitivity), dyspnoea (97% specificity, 58% sensitivity) and overall symptomaticity (97% specificity, 58% sensitivity). We did not observe any statistically significant sex differences for autoimmunity or symptoms in our study cohort at 3 and 6 months. However, at 12 months, a larger proportion of females presented with fatigue compared to males (supplementary figure E4). We also did not observe any differences in autoimmunity or symptoms in patients with/without comorbidities (cardiovascular, respiratory, gastrointestinal, endocrine, renal) at 12 months (supplementary table E2).
The relationship between cytokines, ANAs/ENAs and symptoms in patients 12 months post-COVID-19
Positive correlations were found between various ANAs/ENAs and inflammatory mediators: CRP, ICAM-1, VCAM-1, IL-8 and TNF-α (table 2). Multiple regression analysis was performed on all cytokines for each ANA/ENA at 12 months. At a significance of p<0.01, we found that TNF-α positively predicted anti-U1-snRNP and anti-anti-SS-A/Ro60 reactivity, CRP positively predicted anti-PM-Scl and anti-SmD1 autoreactivities, IL-6 positively predicted anti-PCNA and VCAM-1 positively predicted anti-Ku (supplementary table E1).
Strong positive correlations were found between D-dimer and fatigue at 3 months (r=0.33, p=0.002), TNF-α and cough at 6 months (r=0.38, p=0.031), and TNF-α and fatigue at 12 months (r=0.42, p=0.004) (table 2). At 6 months, TNF-α, VCAM-1 and IL-6 showed the greatest association with symptoms (figure 6b). For 12 months, TNF-α, D-dimer and IL-1β had the strongest association with symptoms (figure 6c). Multiple regression analysis for symptoms demonstrated that D-dimer predicted fatigue (β=1.01, p=0.011) and dyspnoea (β=0.55, p=0.024) at 3 months, ICAM-1 predicted cough at 3 months (β=1.14, p=0.028), and TNF-α (β=4.65, p=0.004) predicted fatigue at 12 months (figure 6d–f). Subsequent regression analysis for general symptomaticity showed that D-dimer (β=1.08, p=0.013) and TNF-α (β=2.40, p=0.03) positively predicted symptomaticity at 3 and 12 months, respectively (figure 6g–i).
Discussion
We comprehensively profiled autoantibody signatures of 18 clinically relevant ANAs/ENAs in 106 convalescent COVID-19 patients at 3, 6 and 12 months post-recovery. First, we demonstrated that COVID-19 survivors had elevated levels of circulating ANAs/ENAs compared to the healthy and non-COVID infection groups at 3 months post-recovery. Among the COVID-19 survivors, the number of ANA/ENA reactivities at 3 months post-recovery proportionally increased with the severity of the patient's acute phase infection; however, this correlation was absent at later time points. Second, high titres of circulating ANAs/ENAs were maintained up to 6 months post-recovery, but were significantly attenuated by 12 months, although several pathogenic ANAs/ENAs are still detectable in up to 30% of COVID survivors at 12 months. Furthermore, for 12% of post-COVID patients, positive ANAs/ENAs were observed at 12 months afresh, that were otherwise below the cut-off threshold at the 3- or 6-month time points, underlining potential de novo autoantibody synthesis. Two of the most prevalent autoantibodies, anti-U1-snRNP and anti-SS-B/La, positively predict both persisting fatigue and dyspnoea symptoms in COVID-19 survivors. Finally, we demonstrated that TNF-α, a key cytokine associated with development/sustenance of autoimmune diseases, positively predicted the observed ANAs/ENAs as well as symptom scores at 12 months post-recovery. Taken together, we provide evidence of an ongoing autoimmune inflammation marked by detectable circulating ANAs/ENAs and elevated TNF-α, associated with persisting symptoms at 12 months post-recovery in individuals who were otherwise healthy before contracting COVID-19.
Although previous work has demonstrated the persistence of autoantibodies in post-COVID individuals [14, 18–21], to our knowledge, the current study is the first to track specific autoantibodies with confirmed/known clinical pathogenicity with commonly reported long COVID symptoms across three time points up to 1 year post-recovery. Transient increases in autoantibodies in response to viral infections is commonly seen in weeks following recovery; however, these generally resolve [22]. Consistent with this, there was a significant reduction in the mean autoreactivities at 12 months in our post-COVID cohort for most autoantigens. That said, several ANAs/ENAs remained detectable despite their statistically significant attenuation in some post-COVID patients, such as anti-U1-snRNP (30%), anti-SS-B/La (21%) and anti-PM-Scl (21%). Whether this is a harbinger of future autoimmunity is not known, but elevated anti-ribonucleoprotein and anti-SS autoantibodies after viral infections (e.g. Epstein–Barr virus, cytomegalovirus) are associated with the development of rheumatological diagnosis [23–25]. In fact, a number of cases of new-onset autoimmune diseases post-COVID have been reported including vasculitis [26, 27], arthritis [28], systemic lupus erythematosus [29] and myositis [30] in patients with no prior history of autoimmunity, irrespective of acute phase severity [31–34].
COVID-19 patients appear to have slower resolution of inflammation as evidenced by elevated IL-1β, IL-6, IL-8 and TNF-α, and this delay in resolution has been hypothesised to contribute to the development of PASC symptoms [7, 35, 36]. Indeed, TNF-α has been linked to fatigue in various diseases including chronic fatigue syndrome and rheumatoid arthritis. Thus, an incomplete mitigation of autoimmune responses/self-reactivities along with endothelial dysfunction (evident by elevated D-dimer) and residual type 1 inflammation may potentially streamline the trajectory towards persisting constitutional symptoms, chronic PASC and eventual development of rheumatological complications.
In a systemic review and meta-analysis conducted in January 2021, fatigue (58%) and dyspnoea (24%) were included within the five most commonly developed long-term symptoms in >47 000 post-COVID-19 patients [1]. We acknowledge that we did not comprehensively record all currently known long-COVID symptoms, and may have missed a subset of patients presenting with symptoms not included in the current study (e.g. joint pain, rashes, neurocognitive dysfunction). A significant subset of the patients was recruited at the early phase of the pandemic (August 2020) through patient referrals, community outreach and hospital recruitment; therefore, a confirmed PASC diagnosis could not be made as per the current guidelines. Symptomaticity, objectively measured, may fluctuate over time for an individual, and subject to recall bias. Hence, we have refrained from calling these patients to have confirmed PASC and aligned our analysis and conclusion with symptomaticity rather than PASC diagnosis.
A few limitations of our study merit consideration. First, given the study's focus on longitudinal observations, it would have been ideal to collect samples and symptoms from our non-COVID-19 infection control cohort at matching 6- and 12-months post-infection time points, similar to our post-COVID-19 population. The ever-changing pandemic landscape made it logistically difficult to allow the longitudinal recruitment of the non-COVID-19 participants. The reluctance of non-COVID-19 participants (infection control and healthy cohorts) to come into the hospital during the pandemic impacted study recruitment. Indeed, although we managed an age- and sex-matched cohort for the healthy participants, we could not do so for the infection control group. In addition, this mismatch was further impaired by the exclusion of five PCR-negative older individuals due to pre-existing rheumatological complications. A more proportional comparison including adequate hospitalised and ICU-admitted controls would shed more light on whether the development of autoantibodies is specific to SARS-CoV-2 infection or due to a general pathogenicity associated with severe viral infection. In addition, our convalescent COVID-19 patient sample size at 12 months totalled 58 patients, compared to 106 patients at 3 months and 98 patients at 6 months. We surmise that an increased attrition rate at later time points may be due to alleviated symptoms in study participants, leading to an enriched symptomatic population at 12 months. A balanced ratio of samples between time points could result in better statistical power for detecting relevant associations between output variables. However, given the topical scenario, we found merit in reporting our observations promptly. Finally, as we do not have pre-pandemic ANA values, we are currently unable to assess if the observed autoimmunity was prevalent pre-COVID, and whether causality exists with the observed symptoms. Though this is currently beyond the scope of the present study, a mechanistic investigation is underway in our ongoing longitudinal long-COVID trial (clinicaltrials.gov identifier NCT05459506).
In summary, ANAs/ENAs with known roles in autoimmune diseases were detected at elevated levels in patients at 3 and 6 months post-COVID-19. Attenuation in the frequency of these autoreactivities was observed by 12 months, despite anti-U1-snRNP and anti-SS-B/La antibodies remaining prevalent in up to 30% of post-COVID-19 patients. These autoreactivities strongly correlate with TNF-α, and both positively predict common PASC symptoms 1 year post-infection. The incomplete attenuation of clinically relevant autoreactivities 12 months post-COVID in one third of patients, associated with persisting symptoms and residual inflammation warrant long-term investigation of autoimmunity in PASC patients.
Supplementary material
Supplementary Material
Please note: supplementary material is not edited by the Editorial Office, and is uploaded as it has been supplied by the author.
Supplementary material ERJ-00970-2022.Supplement
Shareable PDF
Supplementary Material
This one-page PDF can be shared freely online.
Shareable PDF ERJ-00970-2022.Shareable
Acknowledgements
The authors would like to thank Indu Raman (University of Texas Southwestern Medical Center, Dallas, TX, USA) for assisting with the autoantigen microarray. The authors would also like to thank James Johnston, Alyson Wong and Aditi Shah (University of British Columbia, Vancouver, BC, Canada) for their initiatives with the UBC biobank and long COVID clinics.
Footnotes
Author contributions: Conceptualisation: Manali Mukherjee; recruitment (Hamilton): S. Svenningsen, D.M.E. Bowdish and I. Nazy; recruitment (Vancouver): C. Carlsten and C.J. Ryerson; patient care and referrals: P. Nair, T. Ho and S. Waserman; study coordination, sample processing, database management (Hamilton): C. Venegas, R. Jamil, Z. Patel, B. Cowbrough, K. Radford, C. Huang, M. Kjarsgaard, K. Miyasaki and J. Smith; study coordination sample processing, database management (Vancouver): A.C.Y. Yuen and K.S-K. Lau; molecular experiments: K. Son, R. Jamil, K. Miyasaki, Z. Patel and K. Radford; data analysis: K. Son, B. Salter and Manali Mukherjee; statistical analysis: A. Chowdhury, Manan Mukherjee, N. Balakrishnan and A. Dvorkin-Gheva; microarray: Q-Z. Li; original draft preparation: K. Son; manuscript review and editing: C.J. Ryerson, T. Ho, C. Carlsten, P. Nair, D.M.E. Bowdish, I. Nazy and Manali Mukherjee. Manali Mukherjee takes overall guarantee for the data integrity of the manuscript. All authors have read and agreed to the submitted version of the manuscript.
Conflict of interest: Manali Mukherjee is supported by early investigator award from Canadian Institutes of Health Research (CIHR) and Canadian Asthma Allergy and Immunology Foundation (CAAIF); and reports grants from CIHR and Methapharm Specialty Pharmaceuticals, personal fees from AstraZeneca and GlaxoSmithKline, consultant fees from Novartis, outside the submitted work. S. Svenningsen reports grants from Cyclomedica, personal fees from Arrowhead Pharmaceuticals, honorarium for lectures from AZ, Novartis and Polarean, outside the submitted work. S. Waserman reports grants and consulting fees from Alk Abello and CSL Behring, grants from Canadian Allergy, Asthma, and Immunology Foundation, Aimmune and Takeda, personal and consulting fees from AZ, GSK, Sanofi, Medexus, Miravo Health and Bausch Lomb, consulting fees from Novartis, Pfizer and AbbVie, outside of the submitted work; and is chairperson of an advisory board for Siolta, president for CAAIF, board of directors for Asthma Canada, and medical advisor for Food Allergy Canada. P. Nair reports grants and personal fees from AZ, Teva and Sanofi, personal fees from GSK, Equillium and Arrowhead pharma, grants from Foresee and Cyclomedica, outside the submitted work. N. Balakrishnan reports grants from Natural Sciences and Engineering Research Council of Canada, outside the submitted work. D.M.E. Bowdish reports grants from COVID-19 Immunity Task Force/Public Health Agency of Canada, grants from National Science and Engineering Research Council (NSERC), grants from Canadian Institutes of Health Research, personal fees from AZ Mexico, personal fees for invited presentations from academic institutions, outside the submitted work; and is on the board of directors for Lung Health Foundation, and has been an expert testimony witness for the Government of Canada. K. Son, R. Jamil, A. Chowdhury, Manan Mukherjee, C. Venegas, K. Miyasaki, K. Zhang, Z. Patel, B. Salter, A.C.Y. Yuen, K.S-K. Lau, B. Cowbrough, K. Radford, C. Huang, M. Kjarsgaard, A. Dvorkin-Gheva, J. Smith, Q-Z. Li, C.J. Ryerson, T. Ho, I. Nazy and C. Carlsten have nothing to report.
Support statement: Parts of the study were funded by an investigator-initiated grant from Cyclomedica (Canada), Weston Foundation, Michael Smith Foundation for Health Research, UBC Strategic Initiative Fund and COVID-19 Immunity Task Force (Emerging COVID-19 Research Gaps and Priorities Funding Opportunity). Funding information for this article has been deposited with the Crossref Funder Registry.
- Received May 10, 2022.
- Accepted August 4, 2022.
- Copyright ©The authors 2023.
This version is distributed under the terms of the Creative Commons Attribution Non-Commercial Licence 4.0. For commercial reproduction rights and permissions contact permissions{at}ersnet.org