COVID-19 mRNA vaccine immunogenicity decay and breakthrough illness in adolescents and young adults with childhood-onset rheumatic diseases

Abstract

Objectives

To evaluate the humoral immunogenicity for 6 months after the two-dose coronavirus disease 2019 (COVID-19) mRNA vaccination in adolescents and young adults (AYAs) with childhood-onset rheumatic diseases (cRDs).

Methods

This monocentric observational study was conducted between August 2020 and March 2022. Humoral immunogenicity was assessed at 2–3 weeks after first vaccine dose and 1, 3 and 6 months after the second dose by the cPass™ severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) neutralization antibody (nAb) assay. An inhibition signal of ≥30% defined the seroconversion threshold and the readings were calibrated against the World Health Organization International Standard for SARS-CoV-2 antibodies.

Results. One hundred and sixty-nine

AYAs with cRDs were recruited [median age 16.8 years (interquartile range, IQR 14.7–19.5), 52% female, 72% Chinese]. JIA (58%) and SLE (18%) comprised the major diagnoses. After second vaccine dose, 99% seroconverted with a median nAb titre of 1779.8 IU/ml (IQR 882.8–2541.9), declining to 935.6 IU/ml (IQR 261.0–1514.9) and 683.2 IU/ml (IQR 163.5–1400.5) at the 3- and 6-month timepoints, respectively. The diagnosis of JIA [odds ratio (OR) 10.1, 95% CI 1.8–58.4, P = 0.010] and treatment with anti-TNF-α (aTNF) (OR 10.1, 95% CI 1.5–70.0, P = 0.019) were independently associated with a >50% drop of nAb titres at 6 months. Withholding MTX or MMF did not affect the vaccine response or decay rate. The COVID-19 breakthrough infection was estimated at 18.2 cases/1000 patient-months with no clinical risk factors identified.

Conclusion

Over half of AYAs with cRDs had a significant drop in SARS-CoV-2 nAb at 6-month despite an initial robust humoral response. JIA and aTNF usage are predictors of a faster decay rate.

Introduction

The coronavirus disease 2019 (COVID-19) pandemic continues to inflict significant mortality and morbidity worldwide [1]. Notably, patients with rheumatic diseases (RDs) are vulnerable due to the presence of comorbidities, use of immunosuppressive drugs and potential for disease flares during acute infection. A recent systematic review demonstrated a higher prevalence of COVID-19 infection and mortality rate in patients with RDs compared with the general population [2]. Despite a milder disease course in children afflicted with COVID-19, there are still prevalent concerns about its consequent complications, including the Multisystem Inflammatory Syndrome in Children [3–5].

Vaccination against the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has substantially mitigated the adverse impact of the COVID-19 pandemic [6, 7]. However, most vaccine trials have excluded patients with RDs, and subsequent data in patients with RDs have focused mainly on adults [8]. We had previously demonstrated a positive neutralizing response in 99% of our patients with childhood-onset RDs (cRDs) after the two-dose COVID-19 mRNA vaccination. Local reactions remained the most common adverse event, with disease flares occurring in only 2 and 7 patients in the cohort of 159 after their first and second vaccine dose, respectively [9].

Neutralizing antibody (nAb) response following COVID-19 vaccination has been maintained for up to 6 months in immunocompetent populations [10–12]. Limited longitudinal data exists on patients with RDs receiving immunosuppressants. Frey et al. [13] demonstrated high antibody titres in 80% of adult patients with RD 6 months after both doses of mRNA vaccines. However, there are currently no longitudinal data on the humoral immune response of adolescents and young adults (AYAs) with cRDs. This impedes efforts to optimize booster dosing intervals for AYAs with cRDs.

Herein, we aim to delineate the protective humoral neutralization response in our cohort of two-dose mRNA vaccinated AYAs with cRDs at 3 and 6 months post-vaccination and evaluate factors affecting antibody decay rate. Notably, we determined the SARS-CoV-2 nAb level against the spike glycoprotein S1 receptor binding domain (RBD) rather than the total spike glycoprotein S1 IgG titre as a stronger correlation has been observed between the former and breakthrough infections [14]. Importantly, we have also calibrated the nAb response against the World Health Organization (WHO) International Standard for SARS-CoV-2 immunoglobulin to harmonize the measurement for future cross-study comparisons [15, 16].

Methods

Participants and study approval

This is a monocentric observational study. Patients with cRDs registered in our REgistry for Childhood-Onset Rheumatic Disease (RECORD), KK Women’s and Children’s Hospital, Singapore, and had received two doses of Pfizer-BioNTech (BNT162b2; rINN tozinameran) or Moderna (mRNA-1273; rINN elasomeran) COVID-19 mRNA vaccines, were enrolled into the study after obtaining written informed consent, from 11 August 2020 to 18 March 2022. All patients had inactive or low disease activity and were on stable doses of immunosuppressants prior to vaccination. Patients with COVID-19 infection prior to the completion of primary vaccination series were excluded. Data captured included demographics, RD diagnosis, medication and vaccination details. Ethical approval was obtained from the SingHealth Centralised Institutional Review Board (CIRB: 2019-2961, 2019-2239). Blood samples were collected 2–3 weeks after the first vaccination and at 1, 3 and 6 months after the second vaccine dose. Blood samples were collected in serum-separating blood collection tubes and centrifuged at 1300g for 10 min to obtain the cell-free supernatants that were aliquoted into cryovials for storage at −80°C.

Humoral immune response to mRNA vaccines

The frozen serum was thawed in batches and studied with the commercial blocking ELISA, cPass SARS-CoV-2 nAb detection kit (GenScript USA, Inc., NJ, USA), according to the manufacturer’s protocol. This US Food and Drug Administration–approved kit measures the magnitude of antibody-mediated blockade of angiotensin-converting enzyme 2 receptor protein’s interaction with the SARS-CoV-2 RBD of the spike glycoprotein S1, a crucial initial step for viral entry into susceptible host cells. Contrastingly, ELISA kits that assay anti-spike glycoprotein S1 IgG titres measure both neutralizing and non-neutralizing antibodies and are thus of lesser functional relevance as compared with nAb quantification by cPass.

The cPass percent signal inhibition was calculated as [1 − (Optical Density, OD, of sample/OD of negative control)] × 100%, with a signal of ≥30% used as the cut-off for positive detection of SARS-CoV-2 nAbs. The inhibition signal from the cPass assay was converted to the WHO International Units (IU) using an Excel-based conversion tool available online (https://github.com/Lelouchzhu/cPass-to-IU_Conversion) based on previous calibration of this neutralization assay against the WHO International Standard for SARS-CoV-2 immunoglobulin [15–17]. A cPass inhibition signal of 30% corresponds to a cut-off of 28 IU/ml for serum samples. This positive cut-off strongly correlates with a plaque reduction neutralization test of 90% (PRNT₉₀) [18], a current gold standard for SARS-CoV-2 nAb characterization.

Outcome measures

The primary outcome measures were nAb titres at 1, 3 and 6 months after the second vaccine dose to assess decay rate. Secondary outcomes included proportion of patients with >50% decrease in nAb titres at month 3 and 6, and rate of breakthrough infections.

Statistical analysis

Non-parametric analyses were used to describe data and are shown as median [interquartile range (IQR) or range] for continuous variables and percentages for categorical variables. χ²/Fisher’s exact tests and Mann–Whitney U/Kruskal–Wallis analyses were used to compare groups where appropriate. Predictors of vaccine outcomes and rate of decay were assessed using logistic regression. Variables that gave a P < 0.1 in the univariate logistic regression were entered into the multivariate logistic regression by a backward conditional method and confirmed with a forward conditional method yielding odds ratios (ORs) and 95% CIs for predictor variables. All analyses were performed using SPSS version 23.0 (IBM Corp., NY, USA) and GraphPad Prism V.9.3.1 (GraphPad Software, Inc., CA, USA) with statistical significance set at P < 0.05.

Results

Our cohort comprised 169 AYAs with cRDs (52% female, 72% Chinese) with median age at first vaccination of 16.8 years (IQR 14.7–19.5) and disease duration of 5.0 years (IQR 2.7–8.0) (Table 1). All patients received an mRNA-based COVID-19 vaccine, with the majority (n = 157, 93%) inoculated with the Pfizer-BioNTech mRNA vaccines. JIA was the predominant diagnosis (57%), followed by SLE (18%). Diagnoses with five or fewer patients were collectively grouped under ‘others’ (Supplementary Table S1, available at Rheumatology online). For medications, the most common biological DMARDs used were anti-TNF-α (aTNF). As for conventional DMARDs, HCQ, MTX and SSZ were among the most prescribed.

SARS-CoV-2 neutralization response to COVID-19 mRNA vaccination in patients with cRDs

After two-dose COVID-19 mRNA vaccination, 169 patients had blood drawn at a median of 4.7 weeks (IQR 4.1–5.7) (Post-Vac2), with a further 78 (46%) and 43 (25%) samples collected at the 3-month (M3) and 6-month (M6) timepoints, respectively, after the second vaccine dose (Table 1). However, one and five samples were excluded at M3 and M6, respectively, due to the onset of COVID-19 breakthrough infection prior to blood sampling, leaving 77 samples at M3 and 38 samples at M6 for analyses. The nAb response after the first and second COVID-19 mRNA vaccinations of 159 patients of this cohort, with a median sampling duration of 4.6 weeks (IQR 3.9–5.1) from the second vaccine dose, has been reported previously [9].

Strong humoral immunogenicity in AYAs with cRDs after two-dose mRNA COVID-19 vaccination

A robust humoral immune response after the second vaccine dose (Post-Vac2) was demonstrated in our cohort (n = 169) with cPass ≥30% (positive cut-off) in 99% (n = 168) and ≥90% inhibition in 82% (n = 139). The median cPass nAb inhibition was 96.0% (IQR 92.2–97.2). After calibration against the WHO International Standard for SARS-CoV-2 immunoglobulin, the median standardized nAb titre was 1779.8 IU/ml (IQR 882.8–2541.9) (Supplementary Table S2, available at Rheumatology online). Notably, 98% and 96% of our patients had a Post-Vac2 nAb titre above the putative protective thresholds against the prior SARS-CoV-2 Alpha and Delta variants of 82 and 170 IU/ml, respectively, when estimated with a similar assay [16].

SARS-CoV-2 nAb decayed over time in AYAs with cRDs, resulting in reduced protection

At the M3 post-vaccination mark, the nAb titre significantly decayed (P < 0.001) to a median of 935.6 IU/ml (range 10.2–2364.1) (Fig. 1). The nAb titre further decayed at M6 post-vaccination to a median of 683.2 IU/ml (range 2.9–2393.6) but was not significantly different from that of M3 (P > 0.999), suggesting a faster decay rate in the first 3 months after two-dose vaccination with subsequent deceleration. Indeed, the median titre drops were 47.5% (IQR 32.1–69.4) from Post-Vac2 to M3 and 28.4% (IQR 28.4–62.2) from M3 to M6 (P = 0.024), despite a shorter time interval between Post-Vac2 to M3 as compared with M3 to M6.

Seven percent and 14% of patients at M3 and 13% and 26% of patients at M6 had nAb titres below the putative SARS-CoV-2 Alpha and Delta variant protective thresholds of 82 and 170 IU/ml, respectively. Unfortunately, the nAb protection thresholds for the newer SAR-CoV-2 variants, including the current prevailing Omicron variant, have not been determined.

Rapid SARS-CoV-2 nAb titre decay rates are associated with JIA and aTNF treatment

Next, we compared the clinical characteristics and treatments at M6 between the rapid-decay (dropped >50% of Post-Vac2 levels) and slower-decay (≤50% drop) groups (Table 2). The rapid-decay group had a higher proportion of patients with JIA (P = 0.007) and on aTNF (P = 0.015) but a lower proportion of patients taking HCQ (P = 0.002). With multivariate logistic regression, JIA (OR 10.1, 95% CI 1.8–58.4, P = 0.010) and aTNF therapy (OR 10.1, 95% CI 1.5–70.0, P = 0.019) were independently associated with rapid nAb decaying rate (defined as >50% drop from Post-Vac2 levels). The Post-Vac2 nAb titres, age at vaccination, withholding of MTX or MMF after vaccination, and use of other DMARDs were not associated with the rapid decay of nAb in our cohort (Table 3).

Effect of immunosuppressive treatments on the immunogenicity of the COVID-19 mRNA vaccine

JIA patients (58%) comprised the majority of our cRDs vaccination cohort and contributed 97, 46 and 19 samples at Post-Vac2, M3 and M6, respectively (Supplementary Table S3, available at Rheumatology online). Among them, 88% and 85% withheld MTX after their first and second vaccine doses, respectively. As shown in Fig. 2, patients receiving aTNF had significantly lower Post-Vac2 and M6 nAb titres relative to those without DMARDs. Stratified analysis by MTX for all four time points (2–3 weeks after the first vaccine dose, Post-Vac2, M3, M6) did not demonstrate an effect of MTX on nAb titres after the second dose despite a significantly lower response after the first (Supplementary Fig. S1, available at Rheumatology online).

Among 30 childhood onset SLE (cSLE) patients with Post-Vac2 samples, the median nAb titres were 2470.0 IU/ml (IQR 1473.8–2459.3) for HCQ monotherapy, 1133.8 IU/ml (IQR 858.2–1409.5) for MMF monotherapy, 1153.4 IU/ml (IQR 260.7–2519.2) for HCQ and MMF, and 2142.5 IU/ml (IQR 598.4–2652.0) for HCQ and AZA combination therapy (Supplementary Fig. S2, available at Rheumatology online). Stratified analysis by MMF for all four time points did not demonstrate an effect of MMF on nAb titres after the second dose despite a significantly lower response after the first (Supplementary Fig. S3, available at Rheumatology online).

Breakthrough COVID-19 infection in vaccinated patients with cRDs

Twenty-five of 169 cRDs patients (Table 4) had breakthrough COVID-19 infection after completing the two-dose vaccination with an estimated incidence rate of 18 cases/1000 person-months with a median follow-up of 8.8 months (IQR 7.2–9.1). The median duration after the second vaccination to breakthrough infection was 5.4 months (IQR 3.7–6.7). None of the clinical parameters predicted our cohort’s risk of breakthrough infection.

Of the five infected patients with M6 samples, three patients had M3 (pre-infection) samples with a median nAb of 1103.1 IU/ml (range 588.6–1371.1) and M6 (post-infection) median nAb of 2233.8 IU/ml (range 2063.3–2478.1). One cSLE patient on HCQ and MMF therapy had a breakthrough COVID-19 infection 3 days after his M3 nAb blood sample collection. His nAb level increased from 497.6 to 1842.1 IU/ml in 3 days.

Discussion

We had previously shown a robust humoral neutralization response after the two-dose COVID-19 mRNA vaccination in AYAs with cRDs [9]. In this study, we standardized and calibrated the cPass nAb results against the WHO International Standard for anti-SARS-CoV-2 immunoglobulin [16]. Additionally, we demonstrated a significant SARS-CoV-2 nAb decay over time, with a median titre drop of 47.5% 3 months after the second dose. Furthermore, half of the patients had a >75% titre drop 6 months from the second dose (Supplementary Table S4, available at Rheumatology online), which is concerning and warrants a revision in vaccination strategy for AYAs with cRDs.

Understanding the SARS-CoV-2 antibody dynamics after two-dose mRNA vaccination is imperative to sustaining protection. A SARS-CoV-2 IgG antibody kinetics study following two-dose BNT162b2 vaccination reported a decrease in titres by up to 40% each subsequent month after the second dose. At 6 months, 16% of the subjects had antibody levels below the seropositive threshold [19]. Naaber et al. [12] described a similar decline in anti-SARS-CoV-2 spike protein RBD IgG titres in most vaccinated subjects, with a 45% decrease between 1 and 6 weeks after the second BNT162b2 dose. In another longitudinal study involving vaccinated healthcare workers, nAb titres decreased by almost 4-fold after 6 months [11]. Antibody persistence after vaccination is less studied in patients with RDs on immunosuppressive agents. Interestingly, Frey et al. [13] detected a smaller reduction in antibody titres for patients with RDs—antibody titres decreased by 2.8 times between 1 and 6 months, and titres remained above the predicted neutralizing threshold.

In AYAs with cRDs, the interplay between a more robust immune response at a younger age and the presence of immunosuppressants with underlying disease remains uncertain. Our study shows that AYAs with cRDs have a significant decay in antibody titres of comparable magnitude to healthy vaccinated adults. Studies have attempted to estimate nAb half-life post-vaccination, but most of these studies deployed an exponential decay model, which assumes a steady decay rate over time [10, 20, 21]. In our study, antibody titre decay was, in fact, higher during the first 3 months and continued at a slower pace. Levin et al. [11] and Frey et al. [13] reported a similar trend of greater reduction in antibody titres during the first 3 months, prompting the need for more accurate half-life estimation. Our study further illustrated a significant difference in the proportion of patients with a >75% drop in nAb levels between M3 and M6 (17% vs 50%, P = 0.0002), suggesting that nAb levels continue to decline instead of plateauing over time, albeit at a lower rate.

The decline in antibody titres post-vaccination disputes the applicability of national vaccination regimens to AYAs with cRDs. While evidence supports a robust immune response to two-dose mRNA vaccination in AYAs with cRDs and that a three-dose primary series may not be needed, the significant drop in nAb titres highlights the importance of a booster vaccine dose [9]. In our study, one-tenth and one-fourth of the patients at 6 months had nAb titres below the putative protective threshold for Alpha and Delta variants of 82 IU/ml and 170 IU/ml, respectively [16]. Although a nAb protection threshold has not been determined for the currently dominant Omicron variant, the Omicron variant (B.1.1.529) is far more mutated at the spike protein with an unprecedented ability to escape immunity [22]. Current primary vaccination regimens also showed a precipitous decrease in neutralizing titres against the Omicron variant [23]. In addition, having an additional booster vaccination afforded better protection against the Omicron variant than without a booster in a meta-analysis (OR 0.60, 95% CI 0.52–0.68), implying that a higher nAb threshold is required for protection against this variant [24]. As newer SARS-CoV-2 variants emerge, it is likely that maintaining a high nAb titre with booster doses is needed for protection against these variants.

Similar to immunosuppressed solid organ transplant recipients, most patients with autoimmune diseases on immunosuppressants showed an augmented humoral response following booster vaccination [25]. However, the optimal timing of the booster vaccine remains elusive with the lack of longitudinal nAb data in AYAs with cRDs. Accounting for heterogeneity in age, cRDs diagnosis and immunosuppressive therapy, selected groups of patients may benefit from shortened booster dose intervals. Notably, our study showed that the diagnosis of JIA (OR 10.1, 95% CI 1.8–58.4) and aTNF therapy (OR 10.1, 95% CI 1.5–70.0) are independently associated with a more rapid nAb decay rate. Specifically, the latter finding underscores the detrimental effect of aTNF on nAb titre persistence and sustained protection despite good responses after two-dose primary vaccination [9]. The effect of aTNF on vaccine response has been documented in a few studies. Ruddy et al. [26] demonstrated that all patients with RDs on aTNF had positive antibody responses to two-dose COVID-19 mRNA vaccination. A similarly high percentage of vaccine responders in patients on aTNF monotherapy was found in a study by Syversen et al. [27]. In contrast, a recent study on patients with inflammatory bowel disease reported a greater reduction in nAb titres and shorter nAb half-life estimates in patients treated with infliximab (aTNF) when compared with those receiving vedolizumab (targets α4β7 integrin) [28]. Among our JIA patients, those taking aTNF appear to have lower nAb levels and a faster nAb decay rate compared with those on MTX. Our study is the first to show an association between JIA and faster nAb decay via multivariate analysis with adjustment for aTNF use as a possible confounder. An observed negative correlation of inflammatory signalling in circulating T follicular helper (Tfh) cell with Tfh differentiation and antibody response upon influenza vaccination may partially explain this independent association of JIA with a faster nAb decay [29]. JIA patients on aTNF may hence benefit from shorter booster dose intervals to overcome their potentially faster nAb decay rate. More longitudinal studies are needed to further validate and confirm this finding.

The effects of other immunosuppressive treatments on nAb decay are less evident in this study. Previously, we observed in AYAs with cRDs on MMF an attenuated neutralization response after the first vaccine dose that subsequently corrected after the second dose [9]. In adults with RDs, the suppressive effect of MMF persists even after the second vaccine dose [26, 30–32]. In our current study, cSLE patients on MMF trended towards lower nAb titres after the second vaccine dose but did not reach statistical significance. Unfortunately, the samples at M3 and M6 were too few to study the longitudinal effects of MMF on nAb decay. A similar immunomodulatory effect of MTX after the first and second dose mRNA vaccine has been described in other studies—a persistent attenuation was demonstrated in adult studies, but the suppressive effect of MTX was no longer present after the second vaccine dose in our AYAs with cRDs [9, 31, 33]. A recent randomized, open-label trial showed that a 2-week interruption of MTX treatment enhances antibody responses after the COVID-19 booster dose in patients with immune-mediated inflammatory diseases. However, this effect of withholding MTX was no longer significant in the subgroup of patients <40 years old, suggesting a more robust humoral immune response in younger patients [34]. Our study further supports this as withholding MTX after vaccination did not predict a rapid decay rate in the cRDs vaccination cohort. Therefore, the ACR’s recommendation of withholding MTX following each vaccine dose in adults may not be relevant for AYAs with cRDs since it hinders neither vaccine efficacy nor duration of protection.

Initial data on breakthrough COVID-19 infection in patients with RDs is reassuring with large European registries indicating a breakthrough rate of <1% and a single-centred American study reporting 4.7% [35, 36]. However, breakthrough infection rates may continue to rise with waning immunity and new variants. Ahmed et al. [37] reported breakthrough infection in 7% of adult patients with autoimmune diseases. At the time of writing, 15% of the patients in our cohort had breakthrough infection after completing two-dose primary vaccination with an estimated incidence rate of 18 cases/1000 person-months. We postulate that our breakthrough infection rate was relatively high due to the local Omicron wave during sample collection; the Omicron variant has been shown to be less susceptible to serum antibodies from both adults and children who had completed two-dose vaccination [38]. No clinical predictors for breakthrough infection were identified in our cohort. Among the patients with breakthrough infection, the pre-infection median nAb titre of 1103.1 IU/ml (IQR 588.6–1371.1) was well above the previously identified protective threshold for Alpha and Delta variants. However, the sample size for pre-infection nAb titres was too small to compare between patients with and without breakthrough infection. Further analysis on breakthrough infection rates and nAb titres in our cohort is underway with longitudinal monitoring.

Strengths and limitations

Our study is the first to report the nAb kinetics after vaccination in a relatively large cohort of AYAs with cRDs. With the calibration of cPass results against the WHO International Standard, we are able to report our findings in IU/ml which allows comparability of results with future studies. The limitations of our study include the lack of healthy controls to compare the nAb decay rate and identify characteristics affecting the decay rate specific to our AYAs with cRDs. Furthermore, we could only compare our results with adult studies as current studies on nAb decay rate after vaccination in healthy AYAs are limited. Lastly, it is of clinical importance to correlate antibody decay rates and titres with breakthrough infections, but the sample size for patients with breakthrough infections was too low at the time of manuscript writing for meaningful analysis. We aim to better characterize nAb dynamics (post-vaccination after 6 months, booster/second and breakthrough infection) with continued monitoring.

In conclusion, we have shown a robust humoral neutralization response after two-dose COVID-19 mRNA vaccination in AYAs with cRDs. However, a significant nAb decay occurs at 6 months post-vaccination. We have also identified faster decay rates in patients with JIA and aTNF. The standardization of data using the WHO International Standard enables cross-study comparisons. By building upon nAb decay data from current adult studies and future reports in AYAs, our findings may contribute to recommendations on COVID-19 booster dose intervals in AYAs with cRDs.

Supplementary material

Supplementary material is available at Rheumatology online.

Data availability

All data relevant to the study are included in the article or uploaded as supplementary information. The data used to support the findings of this study are included within the article. The data underlying this article will be shared on reasonable request to the corresponding author.

Funding

This study was conducted with the provision of grant funding from the SingHealth Duke-NUS Academic Medicine COVID-19 Rapid Response Research AM/COV004/2020 (S.A.). This research was also supported by the National Research Foundation Singapore under its National Medical Research Council (NMRC) Centre Grant Programme (MOH-000988) (S.A., T.A.) and is administered by the Ministry of Health, Singapore’s NMRC. Other NMRC grant support, NMRC/TA/0059/2017 (J.G.Y.), CIRG21nov-0031 (J.G.Y.), NMRC/MOHIAFCAT2/005/2015 (S.A.), NMRC/TCR/0015-NCC/2016 (S.A.), NMRC/OFLCG/002/2018 (S.A.), CIRG19may0052 (S.A.), MOH-CIRG21nov-0003, NMRC/STPRG-FY19-001 (L.F.W.), NMRC/COVID19RF-003 (L.F.W.) and NMRC/COVID19RF-0014 (L.F.W.), is gratefully acknowledged.

Disclosure statement: W.N.C. and L.F.W. are co-inventors of the surrogate virus neutralization technology that has been commercialized by GenScript Biotech with the trade name cPass.

Ethics approval: The study was performed in accordance with the principles of the Declaration of Helsinki and approved by SingHealth Centralised Institutional Review Board (CIRB: 2019-2961, 2019-2239). The participants signed informed consent on recruitment into the study.

Acknowledgements

We thank the patients and families who contributed their time and efforts in this urgent but crucial study. We would also like to extend our appreciation to our phlebotomists Ms C.H. Yu, Ms M.F. Chia, Ms J.Y.J. Ang, Ms Norhani Bte Tenan and Mr P. Ganesan, C. Prakash for their tireless assistance in collecting blood samples for the study.

References

Author notes