Immunogenicity and reactogenicity of heterologous prime-boost vaccination with inactivated COVID-19 and ChAdOx1 nCoV-19 (AZD1222) vaccines, a quasi-experimental study

ABSTRACT

The global supply of COVID-19 vaccines has been limited, and concerns have arisen about vaccine supply chain disruptions in developing countries. Heterologous prime-boost vaccination, which involves using different vaccines for the first and second doses, has been proposed to enhance the immune response. We aimed to compare the immunogenicity and safety of a heterologous prime-boost vaccination using an inactivated COVID-19 vaccine and AZD1222 vaccine with that of a homologous vaccination using AZD1222. This pilot involved 164 healthy volunteers without prior SARS-CoV-2 infection aged 18 years or older assigned to receive either the heterologous or homologous vaccination. The results showed that the heterologous approach was safe and well-tolerated, although the reactogenicity of the heterologous approach was higher. At 4 weeks after receiving the booster dose, the heterologous approach elicited a non-inferior immune response compared to the homologous approach in neutralizing antibody and cell-mediated immune response. The percentage of inhibition was 83.88 (79.72–88.03) in the heterologous and 79.88 (75.50–84.25) in the homologous group, a mean difference of 4.60 (−1.67–10.88). The geometric mean of interferon-gamma was 1072.53 mIU/mL (799.29–1439.18) in the heterologous group and 867.67 mIU/mL (671.94–1120.40) in the homologous group, a GMR of 1.24 (0.82–1.85). However, the binding antibody test of the heterologous group was inferior to the homologous group. Our findings suggest that the use of heterologous prime-boost vaccination with different types of COVID-19 vaccines is a viable strategy, especially in settings where vaccine supply is limited or where vaccine distribution is challenging.

KEYWORDS:

• COVID-19
• heterologous prime-boost
• vaccine combination
• ChAdOx1 nCoV-19
• AZD1222
• CoronaVac
• BBIBP-CorV

Introduction

The ongoing COVID-19 pandemic has created an urgent need for effective vaccination strategies to prevent the spread of the virus and reduce the burden on healthcare systems worldwide. Multiple vaccines, including mRNA-based vaccines, inactivated virus vaccines, and viral vector-based vaccines, have been developed and approved for use.¹ Most of these vaccines require a two-dose regimen to achieve maximum protection against COVID-19.^2–6 However, the global supply of vaccines has been limited, and concerns have arisen about vaccine supply chain disruptions, especially in developing countries.⁷

The heterologous prime-boost vaccination strategy, which involves the use of different vaccines for the first and second doses, has been proposed to enhance the immune response against SARS-CoV-2. Although the mechanism underlying this enhanced immunogenicity is not fully understood, it is thought to be due to the activation of different arms of the immune system, resulting in a more robust and diverse immune response. In addition, this approach prevents the vector antibody of the initial vaccine in the vector vaccine.⁸ For example, a heterologous prime-boost regimen using an adenovirus vector vaccine for the first dose and an mRNA vaccine for the second dose has been shown to elicit a strong humoral and cellular immune response against SARS-CoV-2. It is believed that the adenovirus vector vaccine stimulates a strong cellular immune response, while the mRNA vaccine primarily elicits a robust humoral immune response. Combining these different types of vaccines may result in a more comprehensive immune response.⁹ While studies have demonstrated the immunogenicity of heterologous prime-boost vaccination regimens using mRNA and adenovirus vector vaccines, there is currently limited solid evidence regarding the immunogenicity of a heterologous prime-boost strategy between inactivated and adenovirus vector vaccines in terms of both humoral and cell-mediated immune response.^8,10

Therefore, in this pilot study, we aimed to compare the immunogenicity and safety of a heterologous prime-boost vaccination regimen using inactivated and adenoviral vectors with that of a homologous vaccination regimen using adenoviral vector vaccine for both doses. The results of this study could have important implications for the design of COVID-19 vaccination programs, especially in settings where vaccine supply is limited or where vaccine distribution is challenging. By demonstrating the non-inferiority of a heterologous vaccination regimen, this study could help to expand the use of available vaccines and improve vaccination coverage, which is critical to controlling the COVID-19 pandemic.

Materials & methods

Study design and participants

This is a single-center, open-label, non-inferiority, quasi-experimental study designed to demonstrate the immunogenicity, safety, and longitudinal dynamic of a heterologous prime-boost vaccination using inactivated COVID-19 vaccines (CoronaVac, Sinovac Biotech, Beijing, China or BBIBP-CorV, Beijing Bio-Institute of Biological Products, Beijing, China) and an adenoviral vector vaccine (AZD1222, developed by Oxford University and produced by AstraZeneca). The study included two intervention groups: the heterologous and homologous groups. In the heterologous group, the participants received the inactivated COVID-19 vaccine followed by the AZD1222 vaccine at 4-week intervals. In the homologous group, participants received 2 doses of the AZD1222 vaccine at 8-week intervals. The study was conducted at Chulabhorn Hospital, Chulabhorn Royal Academy, Bangkok, Thailand, between August 10 and February 10, 2022. Participants were allowed to choose their vaccine regimen.

The study enrolled healthy participants who were ≥18 years of age and had no history of SARS-CoV-2 vaccination or infection. Patients were excluded if they were pregnant or lactating, had uncontrolled underlying diseases, had respiratory tract infection symptoms or fever 14 days before enrollment, or had received any vaccine within 14 days before enrollment. In addition, participants evaluated their COVID-19 history before receiving the booster vaccine. All participants provided written informed consent before enrollment. The study protocol, consent form, and case records form were reviewed and approved by the Chulabhorn Ethics Committee (reference number: 057/2564), and the trial was registered with thaiclinicaltrials.org (TCTR20210714003).

Intervention and procedures

The BBIBP-CorV vaccine from Beijing Bio-Institute of Biological Products Co Ltd, China contained 4 μg of SARS-CoV-2 virion and 0.45 mg of aluminum, while the CoronaVac vaccine contained 3 μg of SARS-CoV-2 virion and 0.45 mg/mL of aluminum.^5,6,11 The AZD1222 vaccine from SK Bioscience Co Ltd, Republic of Korea, contained 5 × 10¹⁰ viral particles.⁴ All vaccines were administered intramuscularly into the deltoid muscle using the standard dose. These vaccines were approved by Thailand’s regulatory agency and were manufactured and vialled following Good Manufacturing Practices guidelines.

After enrollment, the baseline spike protein anti-receptor binding domain (RBD) antibody levels of all participants were measured, and participants with seropositive status at baseline were excluded from the analysis. Immune responses were measured at two time points, 4 and 4 months after the booster dose. At 4 weeks after the second dose, both the humoral immune and cell-mediated immune responses were assessed. The humoral immune response was measured using the total antibody of the spike protein receptor binding domain (RBD) and surrogate neutralizing antibody (sNT), while the cell-mediated immune response was measured by interferon-gamma releasing assay (IGRA). At 4 months, only the total antibody of the spike protein RBD and sNT were measured.

Sample size calculation

The sample size calculation was based on a non-inferiority trial design, with the hypothesis that immunization with a heterologous prime-boost vaccination is not inferior to a homologous prime-boost vaccination. The geometric mean concentration of antibodies after heterologous prime-boost with CoronaVac and AZD1222 was 736.1 BAU/mL (95%CI, 486.0–1115), while the geometric mean concentration after homologous AZD1222 vaccination was 954.5 BAU/mL (95%CI, 886.5–1028), based on data from healthcare workers in our institute. Non-inferiority was defined as a lower limit of the two-sided 95% confidence interval of geometric mean ratio > 0.67. To achieve a power of 80%, with a significance level of 5% (α = 0.05), and accounting for a 20% participant dropout rate, a total of 154 participants were enrolled, with 77 participants in each of the heterologous and homologous groups.

Statistical analysis

The anti-RBD antibody levels, the percentage of inhibition of sNT, and the interferon-gamma level between the heterologous and homologous groups were compared using a multiple linear regression model. For anti-RBD and interferon-gamma level, before hypothesis testing, the level was log-transformed, and the difference of arithmetic means was then converted to the geometric mean ratio (GMR) by anti-logarithm. The non-inferiority criterion was met if the lower bound of the two-sided 95% confidence interval (CI) of the GMR was greater than 0.67 for anti-RBD and interferon-gamma level.¹² For the percentage of inhibition of sNT, the non-inferiority criterion was met if the lower bound of the two-sided 95% confidence interval (CI) of the mean difference was less than 10%. If non-inferiority was established, superiority was then assessed. Summary statistics were presented as mean and standard deviations, medians and interquartile ranges (IQRs), or geometric mean concentrations and 95% CIs as appropriate. The statistical analyses for this study were conducted using IBM SPSS version 26 and GraphPad Prism version 9. A p-value of less than .05 was considered statistically significant.

Endpoints

The primary endpoint was the GMR of anti-RBD concentration between the heterologous and homologous groups at 4 weeks after the booster dose. The secondary endpoints were reactogenicity on 1- and 7-days after each dose of vaccination, as well as the difference in mean percentage of inhibition of sNT at 4 weeks, the GMR of interferon-gamma at 4 weeks, the GMR of anti-RBD at 4 months, and the difference in mean percentage of inhibition of sNT at 4 months.

Safety

All participants were monitored for immediate adverse events for at least 15 minutes after each vaccination at the vaccine clinic. In addition, participants were asked to complete online questionnaires through a short message service to assess reactogenicity at 1- and 7-days post-vaccination. The severity of reactogenicity was graded as mild, moderate, severe, or potentially life-threatening. Mild reactogenicity was defined as not interfering with daily activity or a local reaction less than 5 cm, while moderate reactogenicity was defined as some interference with daily activity or a local reaction of 5.1 to less than 10 cm. Severe reactogenicity was defined as significant interference with daily activity or a local reaction of 10 cm or more. Potentially life-threatening reactogenicity was defined as requiring emergency care or hospitalization or a local reaction including necrosis or exfoliative dermatitis.

Anti-RBD antibody measurements

The immunogenicity was measured using the Elecsys Anti-SARS-CoV-2 S kit, which is a binding antibody assay designed to detect total antibodies against the spike protein receptor-binding domain (RBD) of wild-type SARS-CoV-2. The manufacturer’s cutoff value for a positive result was >0.8 U/mL. However, the report in this study used the binding antibody units (BAU/mL) based on the WHO international standard for anti-SARS-CoV-2 immunoglobulin concentration, using the manufacturer’s conversion factor (Elecsys-S U = 0.972 × BAU). The linearity range was 0.40–250 U/mL.¹³ The binding antibody assays correlated well with the neutralizing antibody test, which is a standard method.¹⁴

Surrogate virus neutralization test (sNT)

Measuring vaccine-induced protection against viral infections typically requires neutralizing tests, but these traditional methods involve complicated and time-consuming procedures. To overcome these constraints, the study employed a surrogate virus neutralization test (sNT) that has been shown to be highly correlated with traditional neutralizing tests.¹⁵ The sNT used in this study was SARS-CoV-2 NeutraLISA (Euroimmun Medizinische Labordiagnostika AG, Lübeck, Germany). The test measures the ability of neutralizing antibodies to block the interaction between the spike protein RBD and recombinant human ACE2. The test is performed on a solid-phase coated with RBD, and biotinylated ACE2 is added to bind the RBD. If neutralizing antibodies are present in the sample, they will bind to the spike protein RBD and prevent the reaction of the recombinant human ACE2. After incubation with the specimen, unbound ACE2 is washed away, and the resulting bound ACE2 is identified using peroxidase-labeled streptavidin. The color intensity generated is inversely proportional to the level of inhibition. According to the manufacturer’s guidelines, a positive outcome is defined as a percentage of inhibition greater than or equal to 35%.

Interferon-gamma releasing assay (IGRA)

The immune response involving T cells is crucial for controlling SARS-CoV-2 infection, and interferon-gamma is a critical cytokine secreted by helper and cytotoxic T cells. This cytokine is a reliable surrogate marker for cell-mediated immune responses.⁹ To assess the T cell response in this study, an interferon-gamma ELISA was used (Euroimmun Medizinische Labordiagnostika AG, Lübeck, Germany). The assay was used to determine SARS-CoV-2 S1 specific T cell activity using heparinized whole blood as the specimen.

The ELISA kit used in the assay consisted of three reaction tubes, with the first tube serving as the background level of interferon-gamma released and the second tube containing the S1 domain of the SARS-CoV-2 spike protein. The third tube contained mitogen to act as an unspecific antigen control indicating T cell stimulation ability. After incubation for 20–24 hours at 37 ± 1°C, the amount of interferon-gamma released into the plasma was measured by ELISA. The measurement range provided by the manufacturer was 3.88–350 mIU/mL. The interferon-gamma level higher than 350 mIU/mL required further sample dilution.

Results

The study enrolled a total of 164 participants between August and September 2021, with 87 participants in the heterologous group and 77 participants in the homologous group. The baseline demographic and clinical of participants were demonstrated in Table 1. Four participants were excluded due to positive serostatus at baseline, three participants in the heterologous group and one participant in the homologous group. These participants were excluded from the immunogenicity analysis. However, they will be included in the safety analysis. There was a significant lower proportion of females in the heterologous group, and the participants in this group trended to be older, although this was not statistically significant. All participants were followed up 4 weeks after receiving the booster dose. However, one-third of the participants (50 out of 160) were lost to follow-up at 4 months. There was no participant diagnosed with COVID-19 during the prime and boost vaccination period. A summarization of the participants’ flow was demonstrated in Figure 1.

Figure 1. Participants’ flow. Between August and September 2021, a total of 164 participants were assessed for eligibility criteria. The study enrolled 87 participants in the heterologous group and 77 participants in the homologous group. Four participants were excluded due to positive serostatus at baseline, one participant in the homologous group and three participants in the heterologous group. These participants were excluded from the immunogenicity analysis. However, they will be included in the safety analysis. All participants were followed up 1 month after receiving the booster dose. However, one-third of the participants (50 out of 160) 20% were lost to follow-up at 4 months.

Table 1. Demographic and clinical characteristic data at baseline of all participants in each group.

Demographic and clinical characteristic at baseline
Characteristic	Heterologous group (IV+AZ) (N = 87)	Homologous group (AZ+AZ) (N = 77)	Total (N = 164)	p-value
Sex – No. of participants (%)				< .01
Male	51 (58.6%)	30 (39.0%)	81 (49.4%)
Female	36 (41.4%)	47 (61.0%)	83 (50.6%)
Mean age (range) — years	42.1(20-68)	38.6(19-64)	40.5(19-68)	ns
Race — No. of participants (%)
Thai	87 (100%)	77 (100%)	164 (100%)
Baseline Anti-RBD Ab status — No. of participants (%)
Negative	84 (96.6%)	76 (98.7%)	160 (97.6%)
Positive	3 (3.4%)	1 (1.3%)	4 (2.4%)
Underlying disease
Hypertension	7 (8.0%)	8 (10.4%)	15 (9.1%)
Diabetes mellitus	3 (3.4%)	3 (3.9%)	6 (3.7%)
Dyslipidemia	0 (0%)	6 (7.8%)	6 (3.7%)
Allergic rhinitis	0 (0%)	2 (2.6%)	2 (1.2%)
Gout	0 (0%)	1 (1.3%)	1 (0.6%)
Deep vein thrombosis	0 (0%)	1 (1.3%)	1 (0.6%)
None	78 (89.6%)	62 (80.5%)	140 (87.5%)

IV: Inactivated vaccine. AZ: AZD1222 vaccine (adenoviral vector vaccine). IQR: Interquartile range. ns: non significant.

The study’s primary endpoint was the concentration of anti-RBD antibodies, and the results showed that the homologous group had significantly higher concentrations at both 4 weeks and 4 months after receiving the booster dose. The geometric mean concentration of anti-RBD antibodies at 4 weeks was 452.41 BAU/mL (95% CI, 351.23–582.73) in the heterologous group and 1137.68 BAU/mL (957.30–1352.04) in the homologous group, with a GMR of 0.42 (0.30–0.58). The subgroup analysis showed that the immune response was higher in the CoronaVac/AZD1222 group compared to the BBIBP/AZD1222 group (616.69 vs. 358.61 BAU/mL). The result at 4 months was similar, with the geometric mean concentration of anti-RBD antibodies being 316.62 BAU/mL (258.28–388.13) in the heterologous group and 479.77 BAU/mL (385.60–596.95) in the homologous group, with a GMR of 0.67 (0.50–0.91) (Table 2 and Figure 2). These results did not meet the non-inferiority criterion, indicating that the homologous booster was superior to the heterologous booster in terms of the anti-RBD antibody response.

Figure 2. Anti-RBD concentration during study period. AZ/AZ: homologous AZD1222 group. IV/AZ: heterologous group. RBD: receptor binding domain. BAU: binding antibody unit. IV: inactivated vaccine. AZ: AZD1222 vaccine (adenoviral vector vaccine). D0: day0 (for prime dose of vaccine). M1: 1 month after booster dose. M4: 4 months after booster dose. *: p-value< .05. ****: p-value < .0001.

Table 2. The immune response follows both a heterologous and a homologous prime-boost vaccination.

	Heterologous group (IV+AZ) (n)	Homologous group (AZ+AZ) (n)	Geometric mean ratio (95%CI)^a
Anti-RBD concentration 4 weeks and 4 months after a booster dose.
At 4 weeks (GMC, 95% CI)	452.41 (351.23–582.73) (n = 84)	1137.68 (957.30–1352.04) (n = 76)	0.42 (0.30–0.58)
Subgroup analysis depends on the prime vaccine of the heterologous group.
	BBIBP/AZ group 358.61 (260.60–493.50) (n = 48)	1137.68 (957.30–1352.04) (n = 76)	0.33 (0.23–0.48)
	CoronaVac/AZ group 616.69 (411.45–924.31) (n = 36)		0.59 (0.39–0.87)
At 4 months (GMC, 95% CI)	316.62 (258.28–388.13) (n = 62)	479.77 (385.60–596.95) (n = 48)	0.67 (0.50–0.91)
Subgroup analysis depends on the prime vaccine of the heterologous group.
	BBIBP/AZ group 286.62 (215.32–381.53) (n = 29)	479.77 (385.60–596.95) (n = 48)	0.61 (0.42–0.89)
	CoronaVac/AZ group 345.56 (256.09–466.30) (n = 33)		0.73 (0.51–1.05)
	Heterologous group (IV+AZ) (n)	Homologous group (AZ+AZ) (n)	Difference of mean (95%CI)^a
Percentage of inhibition of sNT 4 weeks and 4 months after a booster dose.
At 4 weeks (Mean of %IH, 95%CI)	83.88 (79.72–88.03) (n = 84)	79.88 (75.50–84.25) (n = 76)	4.60* (−1.67–10.88)
Subgroup analysis depends on the prime vaccine of the heterologous group.
	BBIBP/AZ group 81.32 (76.17–86.47) (n = 48)	79.88 (75.50–84.25) (n = 76)	1.98* (−5.13–9.11)
	CoronaVac/AZ group 87.28 (80.32–94.25) (n = 36)		8.42** (0.42–16.42)
At 4 months (Mean of %IH, 95%CI)	58.62 (51.97–65.26) (n = 62)	55.85 (49.01–62.68) (n = 48)	2.40 (−7.38–12.16)
Subgroup analysis depends on the prime vaccine of the heterologous group.
	BBIBP/AZ group 55.82 (45.26–66.38) (n = 29)	55.85 (49.01–62.68) (n = 48)	−0.26 (−12.32–11.79)
	CoronaVac/AZ group 60.90 (51.93–69.86) (n = 33)		4.60 (−7.01–16.21)
	Heterologous group (IV+AZ) (n)	Homologous group (AZ+AZ) (n)	Geometric mean ratio (95%CI)^a
Interferon-gamma level (IGRA test) 4 weeks after a booster dose.
At 4 weeks (GMC, 95% CI)	1072.53 (799.29–1439.18) (n = 84)	867.67 (671.94–1120.40) (n = 76)	1.24* (0.82–1.85)
Subgroup analysis depends on the prime vaccine of the heterologous group.
	BBIBP/AZ group 1133.70 (915.53–1403.86) (n = 48)	867.67 (671.94–1120.40) (n = 76)	1.28* (0.81–2.03)
	CoronaVac/AZ group 996.07 (523.41–1895.544) (n = 36)		1.17* (0.70–1.97)

^a:A comparison using a multivariable linear regression model adjusted with age and sex. *: met the non-inferiority criterion. **: met the superiority criterion. RBD: receptor binding domain. IV: Inactivated vaccine. AZ: AZD1222 vaccine (adenoviral vector vaccine). BBIBP: BBIBP-CorV vaccine. CoronaVac: CoronaVac vaccine. GMC: geometric mean. %IH: Percentage of inhibition of surrogate neutralization test. sNT: surrogate neutralization test. n: Number of participants. CI: confidence interval.

Interestingly, for the surrogate neutralization test, the results showed that the immune response in the heterologous group was not inferior to the homologous group at 4 weeks after receiving the booster dose. The percentage of inhibition was 83.88 (79.72–88.03) in the heterologous group and 79.88 (75.50–84.25) in the homologous group, with a mean difference of 4.60 (−1.67–10.88) (Table 2 and Figure 3). This result met the non-inferiority criterion. Similar to the binding antibody test for anti-RBD antibodies, the subgroup analysis showed that the percentage of inhibition was higher in the CoronaVac/AZD1222 group compared to the BBIBP/AZD1222 group (87.28% IH vs. 81.32% IH). Interestingly, this response was superior to that of the homologous group. The result at 4 months showed a similar trend with the 4 weeks, although it was not met the non-inferiority criteria.

Figure 3. Surrogate neutralization test during study period. AZ/AZ: homologous AZD1222 group. IV/AZ: heterologous group. sNT: surrogate neutralization test. IV: inactivated vaccine. AZ: AZD1222 vaccine (adenoviral vector vaccine). M1: 1 month after booster dose. M4: 4 months after booster dose. ns: not significant.

The rate of decline of anti-RBD antibodies in the heterologous group seems to be slower than in the homologous group. At 4 months after the booster vaccination, the concentration of anti-RBD antibodies was about 70% (316.62/452.41 BAU/mL) compared to 42% (479.77/1137.68 BAU/mL) in the homologous group. However, this effect was not observed in the surrogate neutralization test.

The cell-mediated immune response also showed that the heterologous approach was non-inferior to the homologous approach. The geometric mean of interferon-gamma at 4 weeks after receiving the booster dose was 1072.53 mIU/mL (799.29–1439.18) in the heterologous group and 867.67 mIU/mL (671.94–1120.40) in the homologous group, with a GMR of 1.24 (0.82–1.85). The subgroup analysis also supports this response, although the CoronaVac/AZD1222 group had a slightly higher response compared to the BBIBP/AZD1222 group (1133.70 vs. 996.07 mIU/mL) (Table 2 and Figure 4).

Figure 4. Interferon-gamma level 1 month after booster dose. AZ/AZ: homologous AZD1222 group. IV/AZ: heterologous group. IV: inactivated vaccine. AZ: AZD1222 vaccine (adenoviral vector vaccine). ns: not significant.

The reactogenicity observed after the prime vaccination with the inactivated vaccine was lower than that of the AZD1222 vaccine (4.6% vs. 22.1%), and most participants recovered within 7 days. Both systemic and local reactogenicity were low after the prime vaccination with the inactivated vaccine. On the other hand, the most common systemic reactogenicities on day-1 after the prime vaccination with AZD1222 were fever (20.8%), headache (18.2%), and drowsiness (18.2%), with the local reactogenicity being about 10.4%.

Interestingly, the reactogenicity observed after the booster dose was higher in the heterologous group than in the homologous group (48.3% vs. 22.1%). The most common systemic reactogenicities after the booster dose in the heterologous group were fever (46.0%), headache (46.0%), and myalgia (44.8%), with the local reactogenicity being about 25.3%. Conversely, the most common systemic reactogenicities in the homologous group after the booster dose were myalgia (20.8%), headache (18.2%), and fatigue (18.2%), with the local reactogenicity being 16.0%. No life-threatening reactogenicity was reported in this study (Figures 5 and 6).

Figure 5. Reactogenicity after vaccination in a heterologous prime-boost group, inactivated vaccine followed by AZD1222 vaccine on day-1 and day-7.

Figure 6. Reactogenicity after vaccination in a homologous prime-boost group, AZD1222 vaccine followed by AZD1222 vaccine on day-1 and day-7.

Discussion

Based on the study results, the use of a homologous prime-boost of AZD1222 vaccine results in a higher anti-RBD antibody response compared to a heterologous prime-boost of inactivated and AZD1222 vaccine. However, the surrogate neutralization test showed that the immune response in the heterologous group was not inferior to the homologous group, and the cell-mediated immune response was non-inferior in the heterologous group compared to the homologous group. The humoral immune response in the CoronaVac/AZD1222 group seems to be more robust compared to the BBIBP/AZD1222 group. These findings suggest that a heterologous approach could be considered, particularly in cases where the use of an adenoviral vector vaccine may be limited.

Regarding reactogenicity, the study found that the inactivated vaccine had lower reactogenicity after the prime vaccination than the AZD1222 vaccine. However, the reactogenicity after the booster dose was higher in the heterologous group compared to the homologous group. Nonetheless, no life-threatening reactogenicity was reported in the study.

The authors decided to use an 8-week interval between prime and boost vaccinations in the homologous AZD1222 group instead of a 4-week interval due to a more robust immune response. Voysey M. et al. found that longer prime-boost intervals (>12 weeks) after vaccination with the AZD1222 vaccine result in higher vaccine efficacy compared to shorter prime-boost intervals (< 6 weeks).¹⁶

It is possible that the discordance of the results between the binding antibody test and surrogate neutralization test is due to the different functions of the antibodies produced by different vaccines. Binding antibodies may bind to an RBD, but not all the antibodies block its entry into cells.¹⁷ The binding antibody test measures the total level of antibodies that recognize an RBD, whereas the surrogate neutralization test measures the ability of the antibodies to block viral entry into cells.^14,15 Furthermore, it should be noted that while multifunctional antibodies were observed after the initial dose of the AZD1222 vaccine, their levels were significantly enhanced following the second dose. These multifunctional antibodies comprise anti-spike neutralizing antibodies and Fc-mediated functional antibodies, including antibody-dependent neutrophil/monocyte phagocytosis, complement activation, and natural killer cell activation.¹⁸ Conversely, the inactivated COVID-19 vaccine was found to induce relatively low levels of Fc-mediated functional antibodies compared to other vaccine platforms.¹⁹ The significance of non-neutralizing antibodies cannot be overstated, particularly in the face of the variant of concern pandemic.^20,21 However, due to the study by Lee W.S. et al. showed that Fc-mediated functional antibodies had a longer half-life compared to neutralizing antibodies,²² this explanation cannot account for the result that the rate of decline of anti-RBD antibodies in the homologous group is more rapid than in the heterologous group. Further investigation is necessary to better comprehend the relationship between various antibody types and their efficacy in providing protection against infection in the context of homologous and heterologous vaccination.

Our study results for binding antibody testing differ from those of Wanlapakorn N et al., who found that the level of total RBD antibody was similar between participants receiving heterologous inactivated vaccine and AZD1222 vaccine. They also found that anti-RBD IgG was higher in the heterologous group compared to the homologous group.¹⁰ However, the participants in the homologous group of Wanlapakorn N et al. were older, with a mean age of 47.6 (19.0–85.0) compared to 41.1 (18.0–64.0) in the heterologous group. As we know, the immune response of older individuals is generally lower after vaccination.²³ This factor may explain the differences between our study and that of Wanlapakorn N et al.

In terms of neutralization testing, our study results are consistent with previous research which has shown that heterologous prime-boost was non-inferior to the homologous prime-boost. Specifically, it is generally known that the mRNA vaccine generates a stronger neutralizing antibody response compared to the adenoviral vector vaccine.²⁴ However, the Com-COV study demonstrated that individuals who received an initial dose of the AZD1222 vaccine followed by the BNT162b2 vaccine with a 28-day interval were able to generate neutralizing antibodies comparable to those produced through a homologous prime-boost with the BNT162b2 vaccine. The NT₅₀ of pseudovirus neutralizing antibody was 515 (430–617) in the heterologous group, compared to 574 (475–694) in the homologous group, with a geometric mean ratio of 0.9.⁹ In addition, the neutralizing antibody correlates well with the vaccine efficacy.²⁴

When considering the cell-mediated immune response, studies have suggested that the adenoviral vector vaccine may be more robust compared to other vaccine types.^9,16 The cell-mediated immune response is crucial in preventing COVID-19 patients from progressing to severe disease,²⁵ and therefore, this aspect of the immune response is of utmost importance. The cell-mediated immune response was robustly stimulated after the initial dose of the AZD1222 vaccine, peaking at day 14 and then declining by day 28. After the second dose with a 28-day interval, the cell-mediated immune response slightly increased, although the level was lower than that observed at day 14 after the initial dose. However, the cell-mediated immune response is more robust if the initial vaccination is done with the AZD1222 vaccine and is followed by a boost with the BNT162b2 vaccine.⁹ This effect may be due to the vector antibody of the initial vaccine.²⁶ This may explain the result of our study, which showed non-inferiority of the cell-mediated immune response after initial vaccination with the inactivated vaccine and boost with AZD1222 vaccine compared with the homologous prime-boost with AZD1222 vaccine. Although the inactivated COVID-19 vaccine was found to stimulate the cell-mediated immune response less than other vaccine platforms,¹⁷ a booster with a potent vaccine such as AZD1222 in participants receiving initial vaccination with the inactivated vaccine can stimulate a robust cell-mediated immune response compared to the homologous prime-boost with AZD1222, which may be affected by the initial vector antibody.

A nationwide case-control study in Thailand during the delta variant pandemic found that heterologous prime-boost vaccination with inactivated vaccine and AZD1222 vaccine had comparable effectiveness to homologous prime-boost with AZD1222 vaccine. The study included 1,460,458 individuals from September to December 2021. Between day 15–29 after the booster dose, vaccine effectiveness was 57.8% (56.3–59.2) for the heterologous group, whereas it was 61.4% (59.6–63.2) for the homologous group. In addition, the results were comparable between the two groups from day 30–89 after the booster dose.²⁷

One of the primary strengths of this study is its comprehensive approach to evaluating the immunogenicity of different vaccination regimens through multiple endpoints, including anti-RBD antibody levels, sNT titers, and interferon-gamma releasing assays. This approach provides a more complete understanding of the immune response to different vaccination strategies, enabling a more informed assessment of their effectiveness.

However, the study’s main limitation is that it was not a randomized controlled trial. Participants were allowed to choose their own regimen due to vaccine hesitation issues during the study period, resulting in baseline differences that were adjusted using regression modeling. Although the surrogate virus neutralization test correlates well with live virus-based or pseudovirus-based assays in determining neutralizing activity, it is not the gold standard. Additionally, the study was conducted in a single center in Thailand, which may limit the generalizability of the findings to other populations and settings. Furthermore, the study only evaluated short-term immunogenicity and safety of the vaccine regimens, and long-term follow-up data are needed to determine the durability of the immune response and the long-term safety profile of the vaccine regimens. There was no laboratory evaluation of SARS-CoV-2 infection during the study period. Although we evaluated the participants’ history of COVID-19, some may have had asymptomatic infections which could interfere with the result evaluation. Finally, the study did not assess the effectiveness of the vaccine regimens in preventing COVID-19 infection or disease.

Conclusion

In conclusion, our study provides evidence that a heterologous prime-boost vaccination strategy using inactivated COVID-19 and AZD1222 vaccines is safe and elicits a non-inferior immune response in terms of neutralizing antibody and cell-mediated immune response compared to a homologous vaccination regimen with AZD1222. These findings are important, especially in settings where vaccine supply is limited or where vaccine distribution is challenging. However, further randomized controlled trials with a larger sample size are needed to confirm the long-term efficacy and safety of a heterologous prime-boost vaccination strategy.

Acknowledgments

We thank the clinical research management unit for managing this project. We also thank the central laboratory of Chulabhorn Hospital for the laboratory testing and gratefully acknowledge funding from Chulabhorn Royal Academy and National Vaccine Institute, Thailand.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This study was funded by Chulabhorn Royal Academy and National Vaccine Institute, Thailand, grant number 2564.1/10