Neutralizing Antibodies Against Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) Variants Induced by Natural Infection or Vaccination: A Systematic Review and Pooled Analysis

Abstract

Recently emerged severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variants may pose a threat to immunity. A systematic landscape of neutralizing antibodies against emerging variants is needed. We systematically searched for studies that evaluated neutralizing antibody titers induced by previous infection or vaccination against SARS-CoV-2 variants and collected individual data. We identified 106 studies meeting the eligibility criteria. Lineage B.1.351 (beta), P.1 (gamma) and B.1.617.2 (delta) significantly escaped natural infection–mediated neutralization, with an average of 4.1-fold (95% confidence interval [CI]: 3.6–4.7-fold), 1.8-fold (1.4–2.4-fold), and 3.2-fold (2.4–4.1-fold) reduction in live virus neutralization assay, while neutralizing titers against B.1.1.7 (alpha) decreased slightly (1.4-fold [95% CI: 1.2–1.6-fold]). Serum from vaccinees also led to significant reductions in neutralization of B.1.351 across different platforms, with an average of 7.1-fold (95% CI: 5.5–9.0-fold) for nonreplicating vector platform, 4.1-fold (3.7–4.4-fold) for messenger RNA platform, and 2.5-fold (1.7–2.9-fold) for protein subunit platform. Neutralizing antibody levels induced by messenger RNA vaccines against SARS-CoV-2 variants were similar to, or higher, than that derived from naturally infected individuals.

Antibody response established by infection or vaccination might enable effectively neutralize B.1.1.7 (alpha), but with large reductions in neutralizing titers against B.1.351 (beta), B.1.617.2 (delta) and P.1 (gamma). Standardized protocols for neutralization assays and updated prevention and treatment are needed.

Since the first sequence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) was published in January 2020 [1], >2.6 million strains have been documented in Global Initiative on Sharing Avian Influenza Data (GISAID) [2]. Recent reports of several newly emerged lineages have raised significant concerns globally. Of particular concern have been the World Health Organization (WHO)–designated variants of concern (VOCs) alpha (hereafter referred to using the Phylogenetic Assignment of Named Global Outbreak [Pango] lineage designation B.1.1.7), beta (B.1.351), gamma (P.1), and delta (B.1.617.2), harboring several significant mutations in spike glycoproteins, which are key domains of virus-neutralizing antibodies [3, 4]. These mutants rapidly became the dominant circulating virus strains in the regions where they were first isolated, and, especially for B.1.1.7 and B.1.617.2, spread globally [5–7]. Several vital mutations in the receptor-binding domain (RBD), such as L452R and P681R in B.1.617.2, N501Y shared by B.1.1.7, B.1.351, and P.1, and E484K in the latter 2, are associated with increased infectivity and decreased neutralizing potency, with potential to evade humoral immunity from prior infections or vaccination [8–10].

There is concern that SARS-CoV-2 variants can evade immune responses elicited by natural infections and vaccines that are based on the prototype strain. It has been shown that neutralization against variants by convalescent plasma is remarkably reduced by several mutations, including E484K, shared by B.1.351 and P.1 [11]. Serum collected from recipients of licensed vaccines have also a decreased ability to neutralize emerging SARS-CoV-2 mutants [12–14]. In addition, consistent with immunogenicity results, a major loss of efficacy against B.1.351 was seen in NVX-CoV2373 and ChAdOx1 nCoV-19 vaccines [15], although the efficacy was retained against B.1.1.7 for other vaccines [16]. However, existing studies of neutralizing potency against SARS-CoV-2 variants are based on limited numbers of samples and lack comparability between different laboratory methods. Furthermore, there are no studies providing a complete picture of neutralizing antibodies induced by prior infections or vaccination against emerging variants. Here, we systematically summarize the evidence on neutralization ability against various SARS-CoV-2 variants among individuals previously infected with strains from the original SARS-CoV-2 lineage and those who have been vaccinated.

METHODS

Literature Research and Study Selection

According to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline, we conducted a systematic search from 6 databases, including 3 peer-reviewed databases (PubMed, Embase, and Web of Science) and 3 preprint servers (medRxiv, bioRxiv, and Europe PMC), for studies published in English between 1 September 2020 and 8 July 2021 with predefined search terms (Supplementary Table 1). We included studies that reported neutralizing antibodies against SARS-CoV-2 variants, using serum or plasma samples collected from individuals with virologically or serologically confirmed SARS-CoV-2 infections and from vaccine recipients. Detailed inclusion and exclusion criteria were listed in Supplementary Table 2. This study was registered with PROSPERO (no. CRD42021256932).

Data Extraction

We screened all eligible studies to extract the study characteristics, study participants, types of variants, laboratory methods, and antibody titers (Supplementary Table 3). Data were digitized from the figures in articles by pretrained investigators using a digital extraction tool if individual titer values were not available in table format. For studies that did not report individual data or where the reported data are incomplete, we contacted the corresponding author to request data access. We extracted only the titers expressed as reciprocal dilution of serum that neutralizes or inhibits 50% of the virus (eg, 50% neutralization titer and 50% plaque reduction neutralization test). For those individuals with titers lower than limit of detection, we assumed a titer half that limit (eg, a titer of 10 is assumed when “<20” was present). For individuals with multiple specimens, we only included the sample most likely to have neutralization antibodies for each study participant, to avoid repeated inclusion.

Data Synthesis and Analysis

The primary outcome variable was a pooled geometric mean titer (GMT), expressing an average level SARS-CoV-2 neutralizing titers for a group of individual titers. We calculated lineage-specific GMTs with extracted data sets across different study participants and types of neutralization assays, indicating 3 stratified factors (ie, study participants, types of neutralization assays. and lineages).

Study participants were classified into 4 groups (Supplementary Table 4): (1) non–variant-infected individuals, acutely infected or convalescent patients with coronavirus disease 2019 (COVID-19) infected with parental strains or asymptomatic infections from nonvariants; (2) variant-infected individuals, those infected with SARS-CoV-2 variants; (3) uninfected vaccine recipients, healthy vaccinees who were not infected with either parental strains or the variants of SARS-CoV-2 before vaccination; and (4) previously infected vaccine recipients, vaccinees who had been infected with parental strains before vaccination (Supplementary Table 5).

Among different groups of study participants, we further stratified studies by types of neutralization assays. Multiple neutralization assays used in included studies were classified into 3 categories based on the type of virus (authentic or pseudo) used and the types of vectors (lentivirus or vesicular stomatitis virus [VSV]) used in pseudovirus neutralization assays, namely, live virus neutralization assays, lentivirus-vector pseudovirus neutralization assays, and VSV-vector pseudovirus neutralization assays.

Within specific study participants and types of neutralization assays, we calculated lineage-specific GMTs. Specifically, lineage B.1.1.117, B.1.1.26, B.1.1.29, B.1.1.50, and B.1.153 served as the reference strains for comparisons with SARS-CoV-2 variants in some studies [12, 15, 17, 18], and they were classified as lineage B.1 owing to close phylogenetic distance and shared mutation site of 614G [2, 18]. The classification of all lineages involved in eligible studies were shown in Supplementary Table 6. Furthermore, we conducted a matched analysis for samples that had been tested simultaneously against reference strains and variants and calculated the fold changes of GMT. We also explored determinants affecting the GMT, using multivariate linear regression models.

Statistical Analysis

We conducted a random-effect meta-analysis with inverse variance weighting to estimate pooled GMTs across studies and evaluate within-study heterogeneity. Variability between studies was determined by the heterogeneity tests with the Higgins I² statistic. Statistical significance was tested using Kruskal-Wallis rank sum tests with Dunnett post hoc tests and Bonferroni P value adjustment. For paired samples, we used the Wilcoxon matched-pairs signed rank test. Differences were considered statistically significant at P <.05. All statistical analyses were done using R software (version 4.0.1).

RESULTS

Study Selection and Data Extraction

We identified a total of 12 478 studies after systematically searching multiple data sources with 7461 coming from peer-reviewed databases, and 5017 from preprint servers (Figure 1). After screening title, abstract, and full text, we included in our analysis 106 studies containing a total of 5022 individuals and 18 732 neutralization measurements, with previously uninfected vaccine recipients the subjects of more than half of studies (83 studies; 11 891 of 18 732 samples [63.5%]), followed by non–variant-infected individuals (65 studies; 5690 of 18 732 samples [30.4%]), previously infected vaccine recipients (15 studies; 641 of 18 732 samples [3.4%]), and variant-infected individuals (10 studies; 510 of 18 732 samples [2.7%]) (Figure 1). Live virus neutralization assays were most common (48 of 106 studies [45.3%]), followed by lentivirus-vector pseudovirus neutralization assays (39 of 106 [36.8%]) and VSV-vector pseudovirus neutralization assays (24 of 106 [22.6%]). The lineages B.1.1.7 and B.1.351 were the 2 SARS-CoV-2 variants that had been studied the most, comprising more than two-thirds of studies (B.1.1.7, 68 studies [64.2%]; B.1.351, 81 studies [76.4%]) and nearly half of measurements (B.1.1.7, 3840 measurements, [20.5%]; B.1.351, 4333 measurements [23.1%]). Other VOCs, such as P.1 and B.1.617.2, were the subject of 45 studies, including 2407 data points (Supplementary Table 9).

Figure 1.

Selection flowchart of studies, study participants, and variants studied. Abbreviations: mRNA, messenger RNA; VSV, vesicular stomatitis virus.

Level of Neutralizing Antibodies Against SARS-CoV-2 Variants Among Non–Variant-Infected Individuals

Overall, among studies that evaluated neutralizing antibodies in individuals previously infected with nonvariants, the heterogeneity is generally high between studies (I² = 60.8%–91.3%), even after stratification of studies based on type of neutralization assay (Supplementary Table 10). In aggregate, neutralizing titers against B.1.351 and B.1.617.2 were significantly reduced, followed by P.1, while titers against B.1.1.7 were not, when compared with reference lineages titers. With live virus neutralization assays, the pooled GMT was 263.5 (95% confidence interval [CI]: 207.8–334.3) for prototypical lineage B, 193.8 (167.4–224.4) for B.1.1.7, 142.7 (111.1–183.3) for P.1, 64.0 (56.6–72.3) for B.1.351, and 83.4 (63.9–108.9) for B.1.617.2, with an average 4.1-fold (3.6–4.7-fold) reduction in B.1.351, 3.2-fold (2.4–4.1-fold) reduction in B.1.617.2, 1.8-fold (1.4–2.4-fold) reduction in P.1, and 1.4-fold (1.2–1.6-fold) reduction in B.1.1.7, compared with lineage B (Figure 2A). The decrease in neutralization titer was 4.2 (95% CI: 3.8–4.8) for B.1.351 and 3.3 (2.5–4.2) for B.1.617.2, when compared with lineage B.1 with the 614G mutation.

Figure 2.

Neutralizing antibodies against severe acute respiratory syndrome coronavirus 2 variants in non–variant-infected individuals. Neutralizing antibodies against reference strains (gray dots), variants of concern (red dots), variants of interest (blue dots), and alerts for further monitoring (green dots) were determined with live virus neutralization assay (A), lentivirus-vector pseudovirus neutralization assay (B), and vesicular stomatitis virus (VSV)–vector pseudovirus neutralization assay (C). Solid points represent geometric mean titers (GMTs); error bars, 95% confidence intervals; scattering dots, individual titers; numbers in orange rectangle, number of studies and sample size (no. of studies/no. of samples). *P ≤.05; **P ≤.01; ***P ≤.001. Abbreviation: Pango, Phylogenetic Assignment of Named Global Outbreak.

In matched analyses, compared with protolineage B, the reduction in the ability of neutralization decreased from 9.0-fold (95% CI: 6.9–11.9-fold) against B.1.351, to 5.0-fold (4.0–6.2-fold) against P.1, and to 3.8-fold (2.5–6.0-fold) against B.1.1.7 (Supplementary Figure 3). For lentivirus-vector pseudovirus assays, the average reductions in GMT were 4.5–7.3-, 2.6–4.2-, 2.1–3.4-, and 1.2–2.0-fold for lineages B.1.351, B.1.617.2, P.1, and B.1.1.7, respectively, compared with lineage B or B + D614G (Figure 2B). Serum collected from individuals infected with SARS-CoV-2 variants show increased neutralizing antibody levels against corresponding variant strains (Supplementary Figure 6). Multivariate analysis also indicated that B.1.351 and B.1.617.2 had significantly reduced neutralizing activity after controlling for sampling intervals after symptom onset and/or clinical severity, compared with reference lineages (Supplementary Table 11).

Level of Neutralizing Antibodies Against SARS-CoV-2 Variants Among Vaccine Recipients

High heterogeneity was also observed among studies involving previously uninfected vaccine recipients (I² = 87.5%–97.4%) (Supplementary Table 10). Serum collected from uninfected vaccine recipients had diverse neutralizing antibody levels against SARS-CoV-2 variants across different vaccine platforms. In studies using the live virus neutralization assay, the pooled GMT against B.1.351 was generally the lowest across different platforms, with average GMTs of 53.6 (95% CI: 42.0–68.3), 41.9 (29.6–59.3), 86.1 (78.7–94.3), and 42.1 (32.0–55.4) for platforms of nonreplicating vector, inactivated virus, messenger RNA (mRNA), and protein subunit, respectively (Figure 3A). Using lineage B as a reference lineage, we found that the average reduction in neutralizing antibodies against B.1.351 was 7.1-fold (95% CI: 5.5–9.0-fold) for the nonreplicating vector platform, 4.6-fold (4.2–5.0-fold) for the mRNA platform, and 2.5-fold (1.9–3.3-fold) for the protein subunit platform. Paired-sample analysis also showed the most significant reduction in titers of B.1.351 across all vaccine platforms (Supplementary Figure 7).

Figure 3.

Neutralizing antibodies against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variants in uninfected vaccine recipients after administration of different vaccine platforms. Neutralizing antibodies against reference strains (gray dots), variants of concern (red dots), variants of interest (blue dots), and alerts for further monitoring (green dots) were determined with live virus neutralization assay (A), lentivirus-vector pseudovirus neutralization assay (B), and vesicular stomatitis virus (VSV)–vector pseudovirus neutralization assay (C). Solid points represent geometric mean titers (GMTs); error bars, 95% confidence intervals; scattering dots, individual titers; and numbers in the orange rectangles, number of studies and sample size (no. of studies/no. of samples). *P ≤.05; **P ≤.01; ***P ≤.001. Abbreviations: mRNA, messenger RNA; Pango, Phylogenetic Assignment of Named Global Outbreak.

Compared with lineage B + D614G, significant reductions of antibody levels elicited by all included vaccine platforms against lineage B.1.351 were also found in 2 pseudovirus neutralization assays (Figure 3B and 3C). Regarding to other VOCs, the average reduction against B.1.617.2 was 2.4-fold (95% CI: 1.1–5.2-fold) for inactivated vaccines, 1.8-fold (1.5–2.1-fold) for nonreplicating vector platform vaccines, and 1.6-fold (1.4–1.8-fold) for mRNA vaccines. compared with lineage B.1 in live virus neutralization assay. However, the neutralizing antibodies against B.1.1.7 is largely retained, with only a slight decrease of 0.7–1.4-fold among the above vaccines (Figure 3A). The fold reduction of neutralizing antibodies against B.1.525, B.1.526, and B.1.617.1 (WHO-designated variants of interest) and B.1.427/429 (WHO-designated alerts for further monitoring) is not as significant as that of VOCs for mRNA vaccines, ranging from 0.7- to 1.6-fold, compared with lineage B.1 (Figure 3A). In addition, antibody levels in serum collected from previously infected vaccine recipients receiving mRNA vaccines or nonreplicating vector vaccines were significantly higher than those in uninfected vaccine recipients (Supplementary Figure 8).

Comparison Between Neutralizing Antibodies Induced by Natural Infection and Vaccination

Comparisons of the GMT of neutralizing antibodies induced by natural infections (non–variant-infected individuals) and vaccination (uninfected vaccine recipients) showed that individuals infected with parental strains had significantly higher antibody titers than those given nonreplicating vector vaccines against lineage B.1, B.1.1.7, or B.1.617.2, when using live virus neutralization assays (Figure 4A). In contrast, mRNA vaccines induced neutralizing antibody levels against SARS-CoV-2 variants that were similar to or slightly higher than those in previously infected persons, especially for VOC lineages, such as B.1.1.7, B.1.351, and B.1.617.2 (Figure 4A), although this was not evident in studies using pseudovirus neutralization assays (Figure 4B and 4C).

Figure 4.

Comparison between geometric mean titers (GMTs) of neutralizing antibodies induced by natural infection and vaccination by platforms. Neutralizing antibodies comparison were determined in live virus neutralization assay (A), lentivirus-vector pseudovirus neutralization assay (B), and vesicular stomatitis virus (VSV)–vector pseudovirus neutralization assay (C). Neutralization against the same lineage induced by natural infection and different platforms of vaccination was compared. Solid points represent GMTs; error bars, 95% confidence intervals; scattering dots, individual titers; and numbers in orange rectangles, number of studies and sample size (no. of studies/no. of samples). *P ≤.05; **P ≤.01; ***P ≤.001. Abbreviations: mRNA, messenger RNA; NR, nonreplicating.

Discussion

Overall, we comprehensively estimate vaccine and natural infection-induced neutralizing antibody levels against multiple SARS-CoV-2 lineages, including recently emerged variants. Our analyses found that antibodies in both naturally infected and vaccine-induced serum/plasma are slightly reduced but largely retained neutralizing activity against B.1.1.7. However, the neutralizing potency against B.1.351, B.1.617.2 and P.1 was significantly reduced compared with reference lineages. The antibody response after vaccination with varied platforms’ vaccines against lineage B.1.351 was generally the worst. Neutralizing antibody levels induced by mRNA vaccines against SARS-CoV-2 variants were similar to or higher than those derived from naturally infected individuals. Taken together, our findings suggest that immunity derived from prototypical natural infection or vaccination might be less able to neutralize some recently emerging variants, and antibody-based therapies may need to be updated.

Neutralizing antibody titers induced by both natural infections and vaccination against B.1.351, B.1.617.2, and P.1 were significantly lower than those against other variants or reference strains, mainly owing to mutations in spike protein that were associated with decreased neutralizing potency to evade humoral immunity. However, the decrease in neutralizing potency was more obvious in B.1351 than in P.1, although both variants shared the E484K mutation, which could be partly explained by the distinct set of other mutations and/or deletions in the N-terminal domain region or enhanced neutralization of P.1 by anti-RBD antibodies that bind outside the RBD [19]. For lineage B.1.617.2, with several key mutations of L452R, P681R, and T478K [20, 21], we found that it had substantial fold reductions of neutralizing antibodies elicited by vaccinations or infections compared with reference lineages.

It has been documented that the mutation L452R, also shared by B.1.427 and B.1.429, may promote the interaction between the spike and angiotensin-converting enzyme 2 receptor by inducing structural changes in binding domain [22]. An in vitro experiment showed that the mutation T478K could cause the loss of neutralizing potency for some antibody lineages, indicating this mutation was closely associated with immune evasion [23]. The P681R mutation adjacent to the furin cleavage site had been proved to enhance transmissibility and pathogenicity, and its impact on the loss of sensitivity to antibodies needed more investigation. Consistent with previous studies, we found that convalescent and vaccine-induced immune serum samples had neutralizing potency against B.1.1.7 that are comparable to that seen with reference strains, suggesting that those mutations do not remarkably affect neutralizing activity [24]. However, continued vigilance is warranted, given the potential for further mutations that might affect the immunogenicity of the vaccines or reduce the cross-reactivity of antibodies previously induced by natural infections.

Both previous infection and vaccination have been shown to provide potent protection against similar strains, but it is unclear how neutralizing titers against variants might be different. For lineages B.1.1.7, B.1.351, and B.1.617.2, natural infection-induced serum/plasma had significantly higher neutralizing levels than attained by those vaccinated with nonreplicating vector vaccine in a live virus neutralization assay. This is consistent with published efficacy and effectiveness data, which estimated that the efficacy of ChAdOx1 nCoV-19 vaccine against B.1.351 was 10.4% (95% CI: −76.8% to 54.8%), and its effectiveness against B.1.617.2, 59.8% (28.9%–77.3%), indicating reduced protection against symptomatic COVID-19 due to B.1.351 and B.1.617.2 variant infection [25]. In contrast, results from clinical or real-world studies further supported our findings that antibodies induced by mRNA vaccines maintained high neutralizing antibody titers against all VOCs [26, 27], with an estimated effectiveness of 75% (95% CI: 70.5%–78.9%) against B.1.351 infection, 89.5% (85.9%–92.3%) against B.1.1.7 infection [26], and approximately 90% against symptomatic B.1.617.2 infection [27–29]. This suggests that the neutralizing potency elicited by natural infection of previous prototype strains is relatively robust, whereas the immunity induced by vaccination depends on the vaccine platform.

Although some studies have shown that the neutralization levels from live virus and pseudovirus correlate well [30–32], we stratified the results by each of 3 neutralization assays to enhance comparability. The results from live virus assays could provide a comprehensive way to assess inherent viral fitness and the potential impact of other mutations outside the spike region. However, even within the same live virus neutralization assay, different methods were used for experimental end points (eg, cytopathic effect and fluorescence), and to report their final individual titers (eg, 50% and 80% neutralization titers). In addition, experimental procedures, such as virus titration, serum dilution, and virus-serum neutralization, varied greatly across reports from different laboratories. To date there is no universally accepted standardized operating procedure for the SARS-CoV-2 neutralization assay, making comparison between studies difficult. Such standardization of protocols would help reduce heterogeneity and enhance comparability across studies. International efforts to standardize laboratory methods for SARS-CoV-2 neutralization assays are urgently needed. Previous work by WHO in standardizing procedures for avian influenza neutralization assays provides one useful example [33, 34].

Our study has several limitations. First, we synthesized different neutralization assay results to estimate pooled GMTs. While we stratified analyses by neutralization assays according to virus types and vectors to get the most comparable results, significant variation between assays persists. Second, in many studies, individual-level details were not reported, thus limiting our ability to adjust for potential confounding factors in multivariate regression. We tried to contact study authors, but the response rate was generally low. Third, the exact virus lineage causing the infection for some participants is unknown, which may introduce misclassification bias when grouping study participants. However, we have made reliable assumptions according to the information of the locally circulating viruses and the time when the variants begin to circulate in the region. Finally, the limited number of paired samples for certain variants may affect the uncertainty of the estimates.

In conclusion, our study provides a comprehensive mapping of the neutralizing potency against SARS-CoV-2 variants induced by natural infection or vaccination. Our findings suggest that immune serum/plasma from previous infection retained most of its neutralizing potency against B.1.1.7 but significantly lost neutralizing potency against B.1.351, P.1 and B.1.617.2 variants. The evolution of SARS-CoV-2 lineage is still in process, and it is unknown whether long-term accumulation of mutations can erode the neutralizing effectiveness of natural and vaccine-elicited immunity, especially in the context of waning immunity. Therefore, longitudinal monitoring of emerging variants and antibody-induced immunity is of high importance, and standardized protocols for neutralizing antibody testing against SARS-CoV-2 are urgently needed.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgments. We thank Junbo Chen, Qianli Wang, Jun Cai, and Yuxia Liang from the Fudan University.

Author contributions. H. Y. designed and supervised the study. Xinhua C. and Z. C. did the literature search, set up the database and performed all statistical analyses. Xinhua C., Z. C., A. S. A., and D. T. L. drafted the first version of the article. Xinhua C., Z. C., R. S., W. L., N. Z., J. Z., Q. W., X. D., Z. Z., Xinghui C., and S. G. helped with checking data and produced the figures. A. S. A., J. Y., D. T. L., and H. Y. commented on the data and data interpretation and revised the content critically. H. Y. had full access to all the data in the study and had final responsibility for the decision to submit for publication. All authors contributed to review and revision, approved the final manuscript as submitted, and agree to be accountable for all aspects of the work.

Disclaimer. The views expressed are those of the authors and do not necessarily represent the institutions with which they are affiliated. The funders had no role in study design, data collection, data analysis, data interpretation, or writing of the report.

Financial support. This work was supported by the National Science Fund for Distinguished Young Scholars (grant 81525023), the National Science and Technology Major project of China (2018ZX10201001-010), and the US National Institutes of Health (grant R01 AI135115 to A. S. A. and D. T. L.).

Potential conflicts of interest. A. S. A. and D. T. L. received research funding from the National Institutes of Health. H. Y. has received research funding from Sanofi Pasteur, GlaxoSmithKline, Yichang HEC Changjiang Pharmaceutical Co, Shanghai Roche Pharmaceuticals, Shenzhen Sanofi Pasteur Biological Products, and bioMérieux Diagnostic Product; none of this research funding is related to coronavirus disease 2019. All other authors report no potential conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.