Longitudinal study comparing IgG antibodies induced by heterologous prime‐boost COVID‐19 vaccine

Abstract

Vaccines are critical cost‐effective tools to control the COVID‐19 pandemic. The heterologous prime‐boost vaccination has been used by many countries to overcome supply issues, so the effectiveness and safety of this strategy need to be better clarified. This study aims to verify the effect of heterologous prime‐boost COVID‐19 vaccination on healthcare professionals from Dante Pazzanese Hospital in Brazil. It was performed serological assays of vaccinated individuals after 2‐dose of CoronaVac (Sinovac; n = 89) or ChAdOx1 nCoV‐19 (Oxford‐AstraZeneca; n = 166) followed by a BNT162b2 booster (Pfizer‐BioNTech; n = 255). The serum antibodies anti‐S (spike), anti‐N (nucleocapsid), and anti‐RBD (receptor binding domain) were assessed by enzyme‐linked immunosorbent assay. The heterologous booster dose induced a 10‐fold higher anti‐Spike antibody regardless of the 2‐dose of a prime vaccine. It was strikingly observed that BNT162b2 enhanced levels of anti‐spike antibodies, even in those individuals who did not previously respond to the 2‐dose of CoronaVac. In conclusion, the heterologous scheme of vaccination using mRNA as a booster vaccine efficiently enhanced the antibody response against SARS‐CoV‐2, especially benefiting those elderly who were seronegative with a virus‐inactivated vaccine.

1. INTRODUCTION

The first death caused by SARS‐CoV‐2 in Brazil was registered in March 2020, and thousands of deaths have already been noticed worldwide. According to the Brazilian Health Ministry, 34 170 286 individuals were infected by SARS‐CoV‐2, and 681 437 individuals died (lethality 2.0%) up to August 14, 2022, in Brazil. ¹

SARS‐CoV‐2 is an enveloped, plus‐stranded RNA virus with a single‐stranded RNA genome of approximately 30 000 nucleotides. The SARS‐CoV‐2 genome encodes 29 proteins, including 16 nonstructural, 4 structural (S, N, M, and E protein) and 9 accessory proteins. The S protein (spike) is composed of two functional subunits: an N‐terminal half (S1) and a C‐terminal half (S2). The cleavage at S1/S2 of the spike protein is essential for efficient viral entry into its target cells. The SARS‐CoV‐2 S1 subunit contains the receptor‐binding domain (RBD), which is responsible for binding to cellular receptors and uses angiotensin‐converting enzyme 2 (ACE2) to enter target cells. N protein binds to the viral RNA genome to form a ribonucleoprotein complex, which is essential for keeping RNA in an ordered conformation for replication and transcription. N protein acts as a molecular chaperone, it is indispensable for viral RNA synthesis and regulation of cellular processes, such as cell apoptosis and cell cycle progression. Furthermore, the N protein is stable, conserved and highly immunogenic, in addition to being less prone to mutation during infection. ² , ³

In Brazil, the COVID‐19 vaccination started on February 2021, and the priority groups were healthcare workers and individuals with comorbidity, then being progressively extended by age to older than 12‐year populations. The first vaccines approved by the Brazilian National Health Surveillance Agency for emergency use were CoronaVac produced by Sinovac (CV) and the ChAdOx1 nCoV‐19 by Oxford‐AstraZeneca (AZ). Therefore, all workers in our hospital received one of these vaccines according to the 2‐dose initial protocol. Until August 12, 2022, 80% (170 million) of Brazilians have completed the initial COVID‐19 vaccination protocol. ⁴ The omicron (B.1.1.529) SARS‐CoV‐2 variant emerged in Brazil at the end of 2021, and all health workers received BNT162b2 from Pfizer‐BioNTech (PB) as a booster dose according to the recommendation of the Brazilian Health Ministry.

In our previous study, we compared IgG antibody seroconversion among health workers who were immunized with 2‐dose of CV or AZ. ⁵ In this substudy, we described the dynamic changes in anti‐SARS‐CoV‐2 antibodies induced by heterologous vaccines and natural infection, then discussed the importance of booster doses for COVID‐19 prevention.

2. METHODS

2.1. Blood samples collection and storage

The inclusion criteria consisted of having participated in the first phase of the study, being vaccinated with a double dose of CV or AZ, and without prior SARS‐CoV‐2 infection until receiving a PB booster dose.

In our previous study, the serum samples were collected at several time points: before the vaccination, 30 days after first dose, and 30 days after the second dose of CV or AZ. ⁵ For this substudy, participants were invited to be submitted for blood draws, 30 days (December 2021) and 6 months (June 2022) after the PB booster vaccination. In cases of contracting COVID‐19 after June 2022, another sample was collected for analysis.

Blood samples were obtained via antecubital venipuncture by using BD Vacutainer® serum collection tubes. These tubes have been centrifuged at 5000 rpm for 10 min, transferred the serum to another microtube, and stored at –30°C in single‐use aliquots.

2.2. Enzyme‐linked immunosorbent assay (ELISA) for IgG antibody detection

The ELISA protocol used in this study to detect antibodies and the process production of the nucleocapsid (N) are described in our previous publication. ⁶ The recombinant protein antigen spike (S) was kindly provided by Dr. Leda Castilho of the Federal University of Rio de Janeiro, ⁷ and the commercial RBD protein is from Thermo Fisher Scientific (product # RP‐87675). The serum dilution applied for anti‐N was 1/100 and for anti‐S and anti‐RBD was 1/1000. Ten COVID‐19 negative serums and the other 10 positive serums were pooled and used as interassay control. The result higher than the cutoff value of 0.300 OD (optical density) was reported as positive for antibody detection.

2.3. Data collection and statistical analysis

All clinical, laboratory and demographic variables were recorded in REDCap (Research Electronic Data Capture) database, ⁸ supported by the Brazilian REDCap Consortium. Statistical analysis was performed by using R 4.2.0 (R Foundation for Statistical Computing) and Graph Pad Prism® Version 6 for Windows (GraphPad Software). All hypothesis tests were a two‐tailed with 5% significance level.

3. RESULTS

3.1. Characteristics of participants in the study

A total of 550 individuals from our previous study were eligible, however, only 255 voluntarily showed up to participate in this study, being the AZ group (n = 166) and CV group (n = 89). The majority was composed of women in both groups (74.1% and 76.4%), and there were no differences in race or ethnicity (p = 0.251). CV group was older (mean 52.6 vs. 45.2 years; p < 0.001), therefore the number of individuals with dyslipidemia and hypertension was also significantly higher in this group. The number of doctors and nurses or nursing assistants was significantly higher in the CV group (p < 0.001), and the number of administrative workers and persons whose exposure to SARS‐CoV‐2 was higher in the AZ group (p < 0.001) Table 1.

Table 1.

Demographic data of participants according to the prime vaccination

Vaccines	AstraZeneca (n = 166)	CoronaVac (n = 89)	p Valuesa
Gender, n (%)
Female	123 (74.1)	68 (76.4)	0.759
Male	43 (25.9)	21 (23.6)
Age (year)
Mean (standard deviation)	45.2 ± 9.4	52.6 ± 12.9	<0.001
Race/ethnicity, n (%)
White/Caucasian	88 (53.0)	56 (62.9)	0.251
Black/African‐Brazilian	30 (18.1)	7 (7.9)
White and Black mixed	34 (20.5)	19 (21.4)
Asian	6 (3.6)	5 (5.6)
Brazilian Indian	1 (0.6)	0
Not declared	7 (4.2)	2 (2.2)
Staff group, n (%)
Doctor	8 (4.8)	17 (19.1)	<0.001
Nurse or nursing assistant	46 (27.7)	42 (47.2)
Healthcare workerb	39 (23.5)	12 (13.5)
Administrative or executive	54 (32.5)	8 (8.9)
Security and cleaning staff	7 (4.2)	3 (3.4)
Other	12 (7.2)	7 (7.9)
Exposure to SARS‐CoV‐2 at work
Exposure	123 (74.1)	40 (44.9)	<0.001
Non exposure	43 (25.9)	49 (55.1)
Comorbidities, n (%)
Dislipidemia	5 (3.0)	10 (11.5)	0.011
Hypertension	25 (15.2)	26 (29.5)	0.009
Diabetes	9 (5.5)	15 (17.0)	1.000
Asthma	9 (5.5)	4 (4.5)	1.000

Note: Bold values indicate statistically significant at p < 0.05.

^a p Values from χ ² tests, except age compared with t‐test.

^b Dentist, pharmacist, psychologist, social assistant, biologist, and physiotherapist.

The interval between the first and second initial COVID‐19 vaccine was in the 23‐day median for CV, and 88 days for AZ vaccine, and until receiving the booster shot it was at least 6 months in both groups, with a median of 264 days in the CV group and 180 days in AZ group (Table 2).

Table 2.

Interval of immunization according to the type of vaccine

Interval of vaccination	AstraZeneca (n = 166)	CoronaVac (n = 89)	Total (n = 255)
1st to 2nd dose (days)	88 [88−89]	23 [21−25]	87 [24−88]
2nd to 3rd Pfizer booster (days)	180 [169−187]	264 [253−271]	188 [178−257]
1st to 3rd Pfizer booster (days)	218 [206−223]	293 [280−298]	223 [213−282]

Note: Number of days are represented as median [percentile 25; percentile 75].

3.2. Antibody anti‐spike correlates with the anti‐RBD

The reactivity of antibodies against protein‐S (full length with S1 + S2 and RBD) was compared to commercial RBD fragment (aa 330−524) to confirm the effectiveness of the serum samples in recognizing SARS‐CoV‐2 antigen in the serological test. The reactivity (measurement in OD) to protein‐S was higher than to the protein‐RBD (2.04 ± 0.29 vs. 1.72 ± 0.40), paired t‐test (p < 0.0001). However, a good correlation between antibodies anti‐S and anti‐RBD (r = 0.8820, p < 0.0001) was observed (Figure 1).

Figure 1.

Comparison of serum IgG antibodies reactive to SARS‐CoV‐2 spike protein and the receptor binding domain (RBD). The enzyme‐linked immunosorbent assay was applied to verify the effectiveness of serum IgG in recognizing SARS‐CoV‐2 protein. (A) The spike protein was recognized more intensely than the RBD protein. Values were expressed in optical density at 450 nm (2.04 ± 0.29 vs. 1.72 ± 0.40) and compared with paired t‐test (p < 0.0001). (B) There is a good correlation between antibodies anti‐S and anti‐RBD (r = 0.882); Pearson's test (p < 0.0001).

3.3. Level of antibodies followed 2‐prime doses and booster vaccination

The individuals who were immunized with 2‐doses of AZ initially showed undetectable antibody anti‐N until 30 days after vaccination, which is following inclusion criteria. Interestingly, 30 days after the PB booster, 75% of AZ‐vaccinated individuals exhibited a detectable anti‐N IgG (OD > 0.300), and 50% of them reached a level similar to infection (OD > 0.600). Furthermore, a similar level of antibodies anti‐N (p = 0.068), anti‐S (p = 0.290), and anti‐RBD (p = 0.264) between CV and AZ groups were observed (Figure 2).

Figure 2.

Pfizer booster vaccine induced up to 10‐fold anti‐spike antibodies regardless of the initial type of 2‐dose prime vaccine. The enzyme‐linked immunosorbent assay was applied for the detection of antibodies induced by CoronaVac (CV; n = 89), AstraZeneca (AZ; n = 166), and Pfizer booster COVID‐19 vaccine (total n = 255). Anti‐N (nucleocapsid) detection was performed by using prediluted serum (1/100). Anti‐S (spike) was assessed by using serum diluted 1/100 or 1/1000 according to period of vaccination. Statistical analysis performed by Mann−Whitney test.

The PB booster vaccine induced up to 10‐fold of antibody anti‐S and anti‐RBD, regardless of the type of initial 2‐dose vaccine protocol. Of note, it was necessary to prediluted serum until 1/1000 to detect the level of anti‐S and anti‐RBD induced by the PB booster vaccine, while anti‐N was performed with 1/100 of serum dilution (Figure 2).

3.4. Incidence, symptoms, and antibody anti‐spike pattern of COVID‐19

The incidence of COVID‐19 among the population in this study was 12.5% (32/255). Upon considering the period when participants had complete immunization with 2 prime doses and PB booster (up to 30 days), the incidence of COVID‐19 in the CV group was 4.5% (4/89) and in the AZ group was 6% (10/166); p = 0.288. Additionally, 30 days after the PB booster dose, the incidence of COVID‐19 in the CV group was 18.9% (7/37) and in the AZ group was 18.4% (14/76), p = 0.843. Three individuals had COVID‐19 twice, all of them had the first illness with only 1 AZ dose, and the second time after the 6 months of PB booster dose whenever it coincided with the third wave of SARS‐CoV‐2 omicron variant in Brazil. There were no participants in this study who caught COVID‐19 died or required intensive care.

In this study, 26 individuals presented symptoms of COVID‐19, and 6 were asymptomatic. All of them had the diagnosis confirmed by real‐time reverse‐transcriptase‐polymerase chain reaction (RT‐PCR). The asymptomatic individuals who performed the RT‐PCR as colleagues or relatives were positive for COVID‐19. Common symptoms are described in Table 3. A total of 11 individuals had COVID‐19 6 months after getting the PB booster dose, 4 individuals have immunized with 2‐dose CV and 7 with 2‐dose AZ. All of them completed the vaccination scheme with a PB booster dose. These convalescents presented a significant increase of antibody anti‐S (Figure 3), even higher than vaccination (p = 0.0007).

Table 3.

Symptoms described by individuals with COVID‐19

Symptoms of COVID‐19	AstraZeneca n = 21 (%)	CoronaVac n = 11 (%)	Total (n = 32)	p Valuesa
Asymptomatic	4 (19.0)	2 (18.2)	6 (18.8)	1
Sore throat	7 (33.3)	3 (27.3)	10 (31.2)	1
Headache	5 (23.8)	3 (27.3)	8 (25.0)	1
Myalgia	6 (28.6)	2 (18.2)	8 (25.0)	0.830
Cough	6 (28.6)	1 (9.1)	7 (21.9)	0.415
Fever	5 (23.8)	1 (9.1)	6 (18.8)	0.592
Coryza	4 (19.0)	1 (9.1)	5 (15.6)	0.637
Anosmia	3 (14.2)	1 (9.1)	4 (12.5)	1
Fatigue	3 (14.2)	1 (9.1)	4 (12.5)	1
Shortness of breath	2 (9.5)	1 (9.1)	3 (9.4)	1

^a p Values from χ ² tests.

Figure 3.

Antibody anti‐spike protein decreased 6 months after the booster‐dose, but increased sharply with COVID‐19 infection. A total of 11 individuals had COVID‐19 even with the full vaccination protocol (4 received CoronaVac and 7 received AstraZeneca as a 2‐prime vaccine). The level of antibody anti‐S at 30 days was compared to 6 months after the booster dose and after COVID‐19 recovered (mean: 20 days). The anti‐S antibodies were assessed by ELISA by using diluted 1/1000 serum. Statistical analysis was performed by paired t‐test. ELISA, enzyme‐linked immunosorbent assay.

3.5. BNT162b2 booster vaccine induced intense antibody anti‐spike in individuals who were seronegative to CV

Data from 8 individuals in our previous study identified as nonresponsive to CV were analyzed separately to know the effect of the PB booster dose in this group. The median age was 64 years, and percentiles of 25% and 75% were 43 and 65 years, respectively. A significant increase of antibodies anti‐N (p = 0.013) and anti‐S (p < 0.0001) was observed at the time point 30 days after the PB booster dose (Figure 4).

Figure 4.

The BNT162b2 (Pfizer‐BioNTech) booster vaccine induced intense antibody anti‐spike in individuals who did not respond to CoronaVac. Eight individuals from our previous study identified as nonresponsive to CoronaVac (CV) were recruited to assess the effect of the booster in this group. The levels of antibodies anti‐nucleocapsid (A) and anti‐Spike (B) were analyzed by enzyme‐linked immunosorbent assay at the time point: baseline, 21 days after first dose of CV, 30 days after second dose of CV, and 30 days after Pfizer‐BioNTech (PB). Statistical analysis was performed by paired t‐test.

4. DISCUSSION

In our previous study, ⁵ we performed an antiviral serological profile among healthcare workers immunized with CV or AZ vaccines. The results of the 2‐dose vaccination protocol showed a higher humoral response induced by the adenovirus‐vectored AZ vaccine than the virus‐inactivated CV vaccine. The AZ vaccinated individuals also exhibited more side effects compared to CV. Despite global vaccination policies contributing to a reduction in COVID‐19 cases, mutations in the viral genome have led to the emergence of new variants of SARS‐CoV‐2 as the highly transmissible Omicron variant, which caused a new outbreak of COVID‐19 in late 2021. ⁹ , ¹⁰ Although our results showed a higher level of anti‐S antibodies in AZ‐vaccinated individuals, with the outbreak of new SARS‐CoV‐2 variants, the incidence of COVID‐19 was similar between AZ and CV‐vaccinated groups. Due to the known severity of COVID‐19, many health authorities recommended the administration of a booster vaccination to protect high‐risk populations and health workers. Once the Brazilian Omicron outbreak in December 2021, the staff of our hospital was given a heterologous booster dose of the mRNA BNT162b2 PB vaccine.

The serum samples were harvested, and we initially performed ELISA assays to profile antiviral antibodies in all samples. Plates were sensitized both with the full‐length S‐protein or just with the RBD fragment. The full‐length S‐protein harbors subunits S1, S2, and RBD. The RBD antigen contained a fragment between amino acids 330 and 524. Since the RBD is the interaction domain of S‐protein to the cell receptor ACE2, we investigated if there is any advantage to using the RBD fragment instead of the full‐length S‐protein which had been used in our previous work. ⁵ , ⁶ Both, full‐length S‐protein and RBD antigens were used to sensitize ELISA plates, and it was observed a slight increase in IgG antibody signaling when using full‐length S protein. A significant positive correlation was also observed, suggesting that both antigens (S‐protein or RBD) are suitable for serological testing use.

The CV was the first vaccine administered in Brazil at the beginning of the COVID‐19 pandemic to healthcare workers. The limited amount of vaccine demanded to set priorities for those older than 60 years, and workers dedicated to COVID‐19 treatment. Administrative staff and younger healthcare workers were immunized with the AZ vaccine about 3 months later. This could explain the older age in the CV group and the higher number of administrative staff immunized with AZ observed in the demographic data of this study.

The interval between the first and second doses of vaccination followed the recommendation from the manufacturer, 28 days for CV and 90 days for AZ vaccine. The interval between the first and third booster dose was similar in both groups (218 days for AZ and 293 for CV). It allows us to compare the incidence of COVID‐19 after the complete immunization scheme.

It is known that the AZ vaccine uses a genetically engineered adenovirus to produce coronavirus spike proteins, thus generating an immune response. In this study, all individuals of the AZ group had an undetectable anti‐N IgG until 30 days after the 2‐doses of AZ vaccines, indicating that these individuals had not been infected by SARS‐CoV‐2 until that time point. Strikingly, 30 days after the PB booster dose, most individuals exhibited a detectable anti‐N IgG, reaching the same level as the CV group. It is worth mentioning that booster shots coincided with the emergence of the Omicron variant in Brazil. ¹¹ The Omicron variant spreads more easily than earlier variants of the virus that causes COVID‐19. ¹² Theoretically, persons who received adenovirus‐vectored or mRNA vaccines would have anti‐S only, ¹³ if they exhibit anti‐N antibodies, which suggests an infection or asymptomatic contamination by a coronavirus.

From the current study, we observed that a booster dose of the PB vaccine significantly increased the level of serum anti‐S in comparison to the initial 2‐prime homologous vaccination. Our data corroborate with the literature which showed the heterologous vaccine induced a more robust immune response than homologous boosting in human ¹⁴ and mice. ¹⁵ Cheng et al. ¹⁶ in your elegant systematic review and meta‐analysis comparing homologous and heterologous prime‐boost immunization with the COVID‐19 vaccine concluded that an administration of the booster dose effectively recalled specific immune responses to SARS‐CoV‐2 and increased antibody levels, especially in heterologous immunization.

We also observed a marked decrease in the anti‐S 6 months after PB booster. However, in those individuals infected with COVID‐19, a significant increase in anti‐S was observed, even reaching a higher level than vaccination. This study did not consider the ability of neutralization antibodies against wild‐type or variant SARSCoV‐2 or spike‐specific T‐cell immune response, but our results suggest that the vaccine‐generated B memory cells would impact the duration of protective immunity. In a study conducted by Zhang et al., ¹⁷ they observed a persisting decline in the level of serum neutralizing antibodies in COVID‐19 recovered patients, and 1 booster shot effectively enhanced the preexisting neutralizing activity against wild‐type and Delta variant of SARS‐CoV‐2. All these results together show that antibodies induced by vaccination or natural infection decrease over time; therefore, interventions should be conducted to boost the antibodies level to maintain the duration of protective immunity.

Some studies, ¹⁸ , ¹⁹ including our previous publication ⁵ point out a lower level of antibody in elderly people immunized by CV. Furthermore, it was also reported that CV induced a lower neutralizing activity against SARS‐CoV‐2 variants than natural infection. ²⁰ In this study, our results showed a significant increase in the antibodies level, followed by a PB booster dose in elders who did not respond to a 2‐dose of CV. Although titers of antibodies are not an exclusive immune correlate of protection, they are highly predictive of immune protection from symptomatic SARS‐CoV‐2 infection. These data together suggest that elderly people vaccinated with an initial 2‐dose of CV are likely to benefit from an additional heterologous booster vaccine. Considering the long‐term immunogenicity and vaccine equity, Cheng et al. ¹⁶  suggest that only individuals primed with inactivated vaccines, like CV, require a booster dose.

In this study, 65.6% (21/32) of individuals got COVID‐19 after a full scheme of vaccination at the beginning of 2022, which coincided with the emergence of the Omicron variant in Brazil. ¹¹ The Omicron variant harbor a high number of mutations, including 15 mutations in the RBD of spike. ²¹ Probably the vaccines based on the original sequence of spike protein may be less effective to prevent infection for new SARS‐CoV‐2 variants. However, it is worth mentioning that anyone who got COVID‐19 died or needed intensive care in this study, suggesting that the full vaccination followed by a booster dose still protects individuals from severe COVID‐19 caused by the Omicron variant. However, the reformulated COVID‐19 vaccine boosters against new variants may be interesting in the future.

5. CONCLUSION

The heterologous scheme of vaccination using mRNA as a booster vaccine efficiently enhanced the antibody response against SARS‐CoV‐2, especially benefiting those elderly who were seronegative with a virus‐inactivated vaccine.

AUTHOR CONTRIBUTIONS

Study design: Hui Tzu Lin‐Wang and Carlos Gun. Experiments: Hui Tzu Lin‐Wang. Statistical analysis: Lucas Petri Damiani and Eduardo da Silva Farias. Manuscript preparation: Hui Tzu Lin‐Wang and Marcio Chaim Bajgelman.

CONFLICT OF INTEREST

The authors declare no conflict of interest.

ETHICS STATEMENT

This study was submitted to the Ministry of Health Scientific Research (Brazil Platform, CAAE number: 42309321.1.0000.5462) and it was approved by the Dante Pazzanese Institute of Cardiology institutional ethics committee (number: 5.169.559, December 16, 2021). All participants provided their written consent, and the study was conducted following Resolution 466/12 of the Brazilian Health Council/National Health Surveillance Agency and ICH‐GCP for Good Clinical Practices.

Lin‐Wang HT, Damiani LP, Farias EdS, Bajgelman MC, Gun C. Longitudinal study comparing IgG antibodies induced by heterologous prime‐boost COVID‐19 vaccine. J Med Virol. 2022;95:e28379. 10.1002/jmv.28379

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions. The data supporting the findings of this study are available with the corresponding author upon reasonable request.