Anti‐SARS‐CoV‐2 IgG antibody levels among Thai healthcare providers receiving homologous and heterologous COVID‐19 vaccination regimens

Abstract Background We examined SARS‐CoV‐2 anti‐spike 1 IgG antibody levels following COVID‐19 vaccination (AstraZeneca [AZ], Sinovac [SV], Pfizer‐BioNTech [PZ]) among Thai healthcare providers. Methods Blood specimens were tested using enzyme‐linked immunosorbent assay. We analyzed seven vaccination regimens: (1) one dose of AZ or SV, (2) two doses of homologous (2AZ, 2SV) or heterologous (1AZ + 1PZ) vaccines, and (3) three doses of heterologous vaccines (2SV + 1AZ, 2SV + 1PZ). Differences in antibody levels were assessed using Kruskal–Wallis statistic, Mann–Whitney test, or Wilcoxon matched‐pairs signed‐rank test. Antibody kinetics were predicted using fractional polynomial regression. Results The 563 participants had median age of 39 years; 92% were female; 74% reported no underlying medical condition. Antibody levels peaked at 22–23 days in both 1AZ and 2SV vaccinees and dropped below assay's cutoff for positive (35.2 binding antibody units/ml [BAU/ml]) in 55 days among 1AZ vaccinees compared with 117 days among 2SV vaccinees. 1AZ + 1PZ vaccination regimen was highly immunogenic (median 2279 BAU/ml) 1–4 weeks post vaccination. 2SV + 1PZ vaccinees had significantly higher antibody levels than 2SV + 1AZ vaccinees 4 weeks post vaccination (3423 vs. 2105 BAU/ml; p‐value < 0.01), and during weeks 5–8 (3656 vs. 1072 BAU/ml; p‐value < 0.01). Antibodies peaked at 12–15 days in both 2SV + 1PZ and 2SV + 1AZ vaccinees, but those of 2SV + 1AZ declined more rapidly and dropped below assay's cutoff in 228 days while those of 2SV + 1PZ remained detectable. Conclusions 1AZ + 1PZ, 2SV + 1AZ, and 2SV + 1PZ vaccinees had substantial IgG levels, suggesting that these individuals likely mounted sufficient anti‐S1 IgG antibodies for possible protection against SARS‐CoV‐2 infection.

To address the knowledge gap of immunogenicity following vaccination with 2SV with AZ or PZ booster, this report details the kinetics and anti-spike 1 protein (S1) IgG antibody levels elicited by COVID-19 vaccination, as measured by enzyme-linked immunosorbent assay (ELISA). This study is part of an ongoing prospective cohort that follows HCPs for 2 years for SARS-CoV-2 immune response, illness incidence, illness cost, and exposure risk.

| Study setting
In January 2021, we established a cohort of 600 HCPs based on con-

| Verification of COVID-19 vaccination
Participants shared a copy of their government-issued documents with study staff to verify their vaccination information.

| Blood collection and processing
Blood specimens were collected from all participants at enrollment, and every 3 months thereafter according to the study's schedule (regardless of participants' vaccination timelines). As vaccination occurred in phases, this resulted in different time intervals from vaccination to blood collection among participants (i.e., those receiving the vaccines early having longer intervals from vaccination to blood collection than those receiving the vaccines late). Heparinized blood specimens were collected and transported to Mahidol University's laboratory. The separated plasma was stored at À20 C until transported to the Armed Forces Research Institute of Medical Sciences' laboratory.

| Laboratory testing
The thawed plasma was screened for the presence of anti-S1 IgG anti-  By month six visit, two participants did not receive COVID-19

| Data analysis
vaccines. Of 561 who did, 533 (95.0%) transitioned from one vaccination regimen to another (e.g., 1SV at month three visit to 2SV at month six visit). When transition did not occur, only data points from the latest blood collection within the same vaccination regimen were used to create an analytic dataset with independent observations. Days from the last dose were calculated as duration between last vaccination and blood collection. We assessed differences in antibody levels between time periods using the Kruskal-Wallis (>2 groups), or the Mann-Whitney (two groups) test, as appropriate. When comparing the same individuals transitioning from 2SV to 2SV + 1AZ or 2SV + 1PZ regimen, the Wilcoxon matched-pairs signed-rank test was used. We used a level of 264 BAU/ml as a threshold for 80% protection against SARS-CoV-2 symptomatic infection with the Alpha variant. 30 We predicted the median antibody levels from estimation of a fractional polynomial of days from the last dose and plotted the resulting curve along with the 95% confidence interval of the median. 31 Data analysis was performed using Stata version 16 (Stata Corp., USA) and GraphPad version 9.2 (GraphPad Software, Inc., USA); p-values of <0.05 indicated statistical significance.   Table S1).
Among five participants for whom data during the first 4 weeks post vaccination were available, heterologous 1AZ + 1PZ vaccination regimen was highly immunogenic (median 2279 BAU/ml, IQR 1690-2609; Figure 2B). All of these vaccinees had anti-S1 IgG antibody levels exceeding the threshold for 80% protection.
For weeks 9-12 (when direct comparisons between vaccination regimens were possible), the anti-S1 IgG antibody levels among    F I G U R E 2 Anti-S1 IgG antibody response among study participants who enrolled into a healthcare provider cohort, Bangkok, Thailand 3.6 | Anti-S1 IgG antibody kinetics Figure 3A shows the kinetics of anti-S1 IgG antibody levels by day since the last dose among 2SV and 1AZ vaccinees. In both groups, anti-S1 IgG antibody levels peaked at days 22-23 and then declined as time progressed. However, the median anti-S1 IgG antibody level among 1AZ vaccinees appeared to decline more rapidly than that of 2SV vaccinees and was predicted to drop below the assay's cutoff level for positive in 55 days compared with 117 days among 2SV vaccinees ( Figure 3A). The number of 2AZ vaccinees was too small for the kinetics assessment.
The antibody kinetics among 2SV + 1AZ vaccinees differed from those of 2SV + 1PZ vaccinees ( Figure 3B). Although the anti-S1 IgG antibody levels peaked around 12-15 days in both groups, the median antibody level among 2SV + 1AZ vaccinees declined during the observed period and was predicted to drop below level suggested for 80% protection in 145 days and below the assay's cutoff level for positive in 228 days. A much stronger response was observed among 2SV + 1PZ vaccinees with median antibody level remaining above the threshold for positive throughout the observed period. At 228 days when the median antibody level among 2SV + 1AZ dropped below the assay's cutoff for positive, that of the 2SV + 1PZ was predicted to remain at 471 BAU/ml ( Figure 3B).

| DISCUSSION
In this study of 563 individuals, we demonstrated that AZ and SV elicited humoral immune response among a generally healthy cohort of participants. Serial vaccination increased the proportion of seropositive persons. In addition, the homologous 2AZ and heterologous 1AZ + 1PZ vaccination regimens elicited higher antibody levels than that of the 2SV. During the period that this cohort had been followed to the time this report was written, those receiving AZ or PZ as a third dose following the completion of a primary series of SV mounted substantial anti-S1 IgG antibody levels that exceeded the threshold suggested for 80% protection. However, antibody levels of the 2SV protection. 30 We found that during the study period, participants had anti-S1 IgG antibody levels above the threshold for 80% protection after receiving homologous 2AZ (32%) and 2SV (8%), and heterologous 1AZ + 1PZ (100%) vaccination regimens. The effect was more pronounced when the third vaccine dose was given, suggesting that these individuals likely mounted sufficient anti-S1 IgG antibodies for possible protection against SARS-CoV-2 infection. Generally, the anti-S1 IgG antibodies among 2SV + 1AZ vaccinees were in the same magnitude as the levels, measured 14 days post vaccination, among healthy HCPs completing two doses of PZ in other study. 33 Likewise, the anti-S1 IgG antibody levels among 2SV + 1PZ were similar to the antibody levels among healthy adults who were primed with AZ and boosted with PZ. 34 Our data corroborate other studies that reported SV-, AZ, and PZ-elicited antibodies wane over time although the rate of decline is unclear, and the impact on protection against infection and on illness attenuation is undefined. [12][13][14][15][16]35 In this study, a significant reduction of antibody levels was observed among participants who completed the primary series of SV and those receiving 2SV + 1AZ, in a relatively short period (i.e., during the observed time). Nonetheless, from an immunological standpoint, the waning of these antibodies is expected.
Similar to any vaccination, other components of adaptive immunity such as memory B-and T-lymphocytes may also play a role in the elimination of invading pathogens following vaccination. This highlights the importance of studying cellular immune response to better understand the role of these memory cells in longer-term protection against infection and severity of illness. 36 We used the threshold of 264 BAU/ml as a proxy of anti-S1 IgG antibody level conferring 80% protection against symptomatic infection to the SARS-CoV-2 Alpha variant. 30 It is not known if this suggested threshold will change as different variants rising to predominance are further studied. This poses challenges to the control of the pandemic as some variants have increased ability to escape the immune response and/or are more contagious. [37][38][39][40] Viral neutralization tests also showed that the elicited antibodies had different efficiency in neutralizing SARS-CoV-2 lineages. 41 In Thailand, detection of other SARS-CoV-2 variants such as the Delta and the Omicron has been reported. 42 There is a public health need to quickly re-assess the proportions of vaccinated individuals protected when levels for correlate of protection against additional SARS-CoV-2 variants of concern are firmly established. This is of particular importance as we observed no adverse events were reported to study staff, we did not systematically assess reactogenicity following vaccination. In addition, the 2SV + 1AZ vaccinees were significantly older than the 2SV + 1PZ counterparts so observed differences may be due to immunosenescent or comorbidities rather than vaccine products. Further, the assay used to detect binding antibodies in this study was based on the conserved S1 domain of the spike protein of SARS-CoV-2 Wuhan strain, whereas mutations in this particular region were well documented in most SARS-CoV-2 variants of concern identified to date. 45

| CONCLUSIONS
Our study contributed to the limited evidence of the impact of diverse  30 The median anti-Spike 1 IgG antibody level was predicted from an estimation of fractional polynomial regression of days from the last dose; the resulting curve was plotted along with the 95% confidence interval of the median. Equations used to estimate the predicted anti-S1 IgG antibody levels in binding antibody units/ml are as follows:

PEER REVIEW
The peer review history for this article is available at https://publons. com/publon/10.1111/irv.12975.

DATA AVAILABILITY STATEMENT
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.