Safety and efficacy of COVID-19 prime-boost vaccinations: Homologous BBIBP-CorV versus heterologous BNT162b2 boosters in BBIBP-CorV-primed individuals

Background Booster vaccine doses against SARS-CoV-2 have been advocated to address evidence of waning immunity, breakthrough infection, and the emergence of immune-evasive variants. A heterologous prime-boost vaccine strategy may offer advantages over a homologous approach, but the safety and efficacy of this approach with the mRNA vaccine BNT162b2 (BNT: Pfizer) and inactivated BBIBP-CorV (BBIBT: Sinopharm) vaccines have not been studied. Methods We conducted a non-randomized, non-blinded phase II observational community trial across Bahrain, investigating the reactogenic and immunogenic response of participants who had previously received two doses of BBIBP, followed by a third booster dose of either BBIBP (homologous booster) or BNT (heterologous booster). Immunogenicity through serological status was determined at baseline and on the following 8th week. Reactogenicity data (safety and adverse events) were collected throughout study period, in addition to participant-led electronic journaling. Results 305 participants (152 BBIBP and 153 BNT booster) were enrolled in the study, with 246 (127 BBIBP and 119 BNT booster) included in the final analysis. There was a significant increase in anti-SARS-CoV-2 antibody levels post booster administration in both groups; however, the heterologous BNT arm demonstrated a significantly larger mean increase in the level of spike (S) antigen-specific antibodies (32.7-fold increase versus 2.6, p < 0.0001) and sVNT neutralising antibodies (3.4-fold increase versus 1.8, p < 0.0001), whereas the homologous arm demonstrated a significant increase in the levels of nucleocapsid (N) antigen-specific antibodies (3.8-fold increase versus none). Non-serious adverse events (injection site pain, fever, and fatigue) were more commonly reported in the heterologous arm, but no serious adverse events occurred. Conclusion Heterologous prime-boost vaccination with the mRNA BNT162b2 (Pfizer) vaccine in those who had received two doses of inactivated virus BBIBP-CorV (Sinopharm) vaccine demonstrated a more robust immune response against SARS-CoV-2 than the homologous BBIBP booster and appears safe and well tolerated. Clinical Trial Registry Number (ClinicalTrials.gov): NCT04993560.


Introduction
With over half a billion cases and 6 million deaths in the first two years since the start of the SARS-CoV-2 (COVID-19) pandemic [1], together with the superimposed global economic damage [2], major worldwide disruption has occured. Vaccination roll-out continues to be the primary prevention against the overwhelming of healthcare systems and worsening of the economic impact [3].
Inactivated vaccines have been extensively studied as a primary method of vaccination over many years. In a phase I/II trial, the BBIBP vaccine was shown to be generally safe and effective against SARS-CoV-2 [4]. However, WHO's Strategic Advisory Group of Experts (SAGE)-after summarizing the evidence from clinical trials in Bahrain, United Arab Emirates, Egypt, Jordan, and China-reported that individuals with comorbidities and older adults (60 years) who received 2 doses of BBIBP may have a suboptimal degree of protection against SARS-CoV-2 [5].
Current clinical trials have played a key role in the approval of different SARS-CoV-2 vaccines based on their efficacy data; however, there is still uncertainty regarding the robustness and duration of protection offered [6,7] and the reliability of antibody levels as surrogate markers for immunity [8]. Additionally, recent evidence has shown that the new SARS-CoV-2 variants reduce the efficacy of vaccinations and are predominantly more transmissible [9][10][11]. Rates of breakthrough infections have also been subsequently rising [12,13].
With the concern of waning immunity and a rise in COVID-19 cases, a large number of countries have started introducing a booster dose following the primary-two doses of the COVID-19 immunization schedule, in an effort to enhance and prolong the population's immunity. Although the homologous prime-boost strategy offers a reliable and tested immune enhancement [14], heterologous vaccination has been shown to significantly induce more immunogenicity in a number of recent trials across adenovirus vector and mRNA vaccine platforms [15]. Evidence for the efficacy of heterologous prime-boost strategies may also help facilitate vaccine sharing in an effort to address global distribution inequities [16]. Hence, our study was undertaken, comparing the reactogenic and immunogenic response of a homologous inactivated whole virus COVID-19 prime-boost vaccine schedule (BBIBP-primed and boosted) to a heterologous inactivated vaccine (BBIBP-primed) with an mRNA vaccine (BNT-boosted) schedule.

Study design
This study was a non-randomized, non-blinded phase II observational community trial conducted across the Kingdom of Bahrain, investigating the reactogenic and immunogenic response of heterologous vs homologous COVID-19 vaccine boosting of participants who previously received two doses of BBIBP, followed by a booster dose of either BBIBP (homologous booster) or BNT (heterologous booster). A prior Phase I pilot study was also conducted in order to assess feasibility of the Phase II study and to calculate the required sample size for sufficient power. Patient recruitment occurred through the centralised database of the National Taskforce for the Combatting of the Coronavirus   (COVID-19 [17]. Two prime-boost schedules were compared: an arm that received two doses of BBIBP in the previous 3 to 6 months, followed by a third homologous booster dose of BBIBP, and a second arm that received two doses of BBIBP in previous 3 to 6 months, followed by a third heterologous booster dose of BNT. Immunogenicity via serological status from blood samples was determined at baseline prior to booster administration, and on the follow-up visit 8 weeks post-booster. Reactogenicity data (safety and adverse events) were collected throughout study period through telephone contact, as well as participant-led electronic journaling. The full protocol can be found in the Supplementary Material.

Participants
All adults aged 21 years and older were eligible for recruitment. Inclusion criteria included that the participants had to have completed 3 to 6 months after the second dose of the BBIBP vaccine, were asymptomatic 24 h before administration of the booster dose and had a negative rapid antigen test for SARS-CoV-2 on the day of vaccination. Exclusion criteria omitted any individuals with active or previous RT-PCR laboratory-confirmed COVID-19 diagnoses.

Randomization
As an observational phase II community trial, no randomization or blinding was undertaken. In Bahrain, where several vaccine platforms are available, citizens and residents sign up for the primary and booster vaccine of their choice on an online application. On arrival to the vaccination centre for the booster of the random individual's choosing, participants who were eligible for inclusion were informed about the study and written informed consent was undertaken to participate. The clinical team assessing the immunogenicity and safety outcomes was also not blinded.

Procedures
The BBIBP and BNT vaccines used in this study became available in Bahrain after a conditional marketing authorization was granted by the National Health Regulatory Authority in November and December of 2020, respectively. Following the negative rapid antigen detection, baseline blood samples were obtained for serological testing of SARS-CoV-2 antibody titres. Participant demographics and medical history were also collected. The BBIBP and BNT booster doses were then administered, followed directly by a 15-min on-site observation for safety monitoring.
Participants were phoned on Day 1 and Day 5 post-booster administration for follow-up regarding any experienced adverse events. A follow-up phone call took place weekly thereafter until the 8th week post-booster. Participants were also provided with an electronic diary system to note down any experienced adverse events throughout the study period and each entry was emailed automatically when uploaded if rated ''severe" by the participant. An in-person follow-up was carried out on Week 8, with blood samples obtained at review to determine updated serological status against SARS-CoV-2.

Outcomes
The primary outcome was the immunogenicity of the different prime-boost schedules (homologous BBIBP booster vs heterologous BNT booster). The secondary outcome was the reactogenicity of the two schedules.
Immunogenicity was assessed by the blood samples obtained at baseline prior to booster administration, and at the end of the study period, 8 weeks after booster administration. Antigenspecific humoral immune response was analysed using one commercial immunoassay [Spike (S) and nucleocapsid (N) antigenspecific immune responses)] and one surrogate viral neutralization test (sVNT) assay before the administration of the booster dose and following the 8th week. S and N antigen-specific humoral immune responses were analysed using The Elecsys Ò anti-SARS-CoV-2 S assay (Roche Diagnostics GmbH, Mannheim, Germany), which is an electrochemiluminescence immunoassay that detects IgG antibodies to the SARS-CoV-2 spike protein receptor-binding domain (RBD) and nucleocapsid antigen on the Cobas e411 module. The manufactuer's conversion factors were used to obtain the WHO international standard binding antibody units (BAU)/ml values, in order to improve the reliability and standerdization of our results. Values higher than 0.8 BAU/mL were considered positive. sVNT neutralization was analysed using the cPass TM SARS-CoV-2 Neutralization Antibody Detection Kit (Genscript Biotech Corporation, China); neutralizing antibodies were calculated as 30 % inhibitory dose (neutralizing titre 30, NT30). Our surrogate virus neutralization assay does not require Level 3 Biosafety containment, making it broadly accessible to the wider community for both research and clinical applications; the test has also been shown to be reliably correlated with the conventional live viral neutralisation test (R squared greater than 0.85) [18]. We have also calculated the margin of error (15 %) using control samples, which was in-line with accepted standards.

Statistical analysis
We considered the S, N and sVNT IgG geometric mean concentrations from a pilot study to determine the required sample size. Considering the sVNT neutralizing virus mean inhibition value of 94.63 % and standard deviation of 1.22 after 14 days of BNT162b2 booster dose administration in individuals who received two BBIBP-CorV doses, and sVNT mean inhibition value of 73.7 % and a standard deviation of 1.64 after 14 days of BBIBP-CorV booster dose administration in individuals who received two BBIBP-CorV doses, the minimum required sample size in each arm was determined to be 120 participants to detect a two-tailed difference between the groups at 95 % power. This also conferred more than 95 % power to detect two-tailed differences in S antigen specific IgG concentrations, which required 52 participants. Therefore, a minimum of 150 subjects per group were to be recruited to account for a potential 20 % loss to follow up.
The level of significance was set at 5%. Data trends were visually and statistically evaluated for normality. Baseline differences were assessed using a chi-square test for categorical variables and independent samples t-test for continuous variables. Differences in immune response across the treatment groups were assessed using independent samples t-test. Covariate adjusted mean immune 2.7. Accounting for age differences in the groups: Since mean age was found to be significantly higher in the heterologous group compared to the homologous group, we accounted for age when looking for differences in immune responses between the groups by using two approaches: stratification and general linear modelling.

Results
The flow diagram for patient participation in the clinical trial is shown in Fig. 1. Of 384 eligible subjects, a total of 305 participants were enrolled into the study and included in the final analysis. All 305 received two primary doses of BBIBP between 3 and 6 months prior to study commencement. 152 received a homologous BBIBP booster, while 153 received a heterologous BNT booster. The mean age of the participants was 40.3 ± 9.8 and 36.9 ± 9.3 years for the BBIBP and BNT groups, respectively. This difference was found to be significant and had to be controlled for in the comparative analysis. Males represented 64 % (BBIBP arm) and 61 % (BNT arm) of the total cohort (Table 1). A total of 119 subjects completed the study in the BNT arm and 127 subjects completed the study in the BBIBP arm.
There were no between-group differences for the baseline serology for the three tested antibodies (S antigen-specific immune response, N antigen-specific immune response, and sVNT neutralisation) ( Table 2).

Primary Outcome: Efficacy
In terms of the primary outcome of immunogenicity 8 weeks after booster administration, both groups experienced significant increases in antibody titres. For the between-group analysis, there were significant differences between the antibody titers of the heterologous BNT-boosted arm and the BBIBP arm after 8 weeks of booster administration (Fig. 2). The heterologous BNT arm showed a significantly larger mean increase in S antigen-specific antibodies compared to the homologous BBIBP arm (14849 ± 7127 vs 1178 ± 1411, p < 0.0001). Similarly, the heterologous BNT arm showed a significantly larger mean increase in levels of sVNT neutralization antibodies (97 ± 2 vs 63 ± 25; p < 0.0001). The homologous BBIBP arm showed a significantly larger (p < 0.0001) mean increase in N antigen-specific antibodies levels compared to the heterologous BNT arm (16 ± 44 vs 110 ± 75) ( Table 3).
In the heterologous BNT booster arm, there was a mean increase of 32.7-fold (p < 0.0001) in the baseline levels of S antigen-specific antibodies (453 BAU/mL ± 1971 vs 14,849 BAU/mL ± 7127), no increase in the levels of N antigen-specific antibodies (21 BAU/ mL ± 54 vs 16 BAU/mL ± 44), and a 3.4-fold increase (p < 0.0001) in the levels of sVNT neutralizing antibodies (28 % ±27 vs 97 % ±2) ( Table 5). Note that the sVNT neutralization levels are reported as percentage of inhibition as per the manufacturer and FDA manuals [19].

Secondary Outcome: Reactogenicity
For the secondary outcome of reactogenicity, all reported adverse events were of Grade 1 (mild) category. The most common adverse event across both groups was injection site pain, more commonly reported in the heterologous BNT-boosted group (55 %) as opposed to the homologous BBIBP group (22 %, P < 0.001). For the BNT group, the second most common adverse event was fever (15 %), which in the homologous BBIBP group was reported in 5.2 % (p < 0.05). The third most common adverse event in the BNT group was fatigue (14.4 %), which was of greater frequency than seen in the BBIBP group (6.5 %, p = 0.02). Headache  . Dyspnea was also reported in two participants from the BNT arm, but none from the BIBP arm, whereas a local allergic reaction and vomiting was reported in one patient in the BBIBP arm and none in the BNT arm ( Table 8). Timing of adverse events following immunization can be seen in Table 9; overall, the BNT group was more likely to experience more adverse events, which would mostly take place between Day 1 and Day 5 post booster administration.
3.3. Accounting for age differences in the groups: Stratification: Below the age of 40 years, the mean age was not found to be significantly different between the two treatment groups. Similarly, above the age of 40 years, mean ages were similar across both groups. As such, we looked at the differences in immune responses across the groups within the strata of < 40years of age and 40-years of age, as can be seen in Table 6. The results remained the same after stratification.
Adjusting for Age: We also looked at the age-adjusted mean values of antibodies across both groups at 8-weeks post booster administration, using general linear models, as shown in Table 7.
Here, the main effects for the group were looked at and age was used as a continuous covariate when looking at mean differences across the three antibodies. Again, results were similar to the original unadjusted analysis, with the effect seeming more pronounced.

Discussion
This study showed a significant increase in anti-SARS-CoV-2 antibody levels post booster administration in both homologous and heterologous prime-boost vaccination groups; however, the heterologous BNT arm demonstrated a significantly larger mean increase in the level of S antigen-specific and sVNT neutralising antibodies, while the homologous arm demonstrated a more significant increase in the levels of N antigen-specific antibodies. Both arms were well-tolerated, with no serious adverse events.
Within group differences pre-and post-booster administration, the homologous BBIBP arm showed a significant improvement in all three (S, N, and sVNT neutralising) antibody titres. This is in accord with the immune response for the inactivated whole virus, that has also been shown to elicit a wider antibody response  against viral epitopes (including the S, N and membrane (M) proteins) when compared to the mRNA BNT vaccine [20]. Conversely, the mRNA BNT vaccine only encodes the SARS-CoV-2 spike (S) protein [21]. Therefore, only the S antigen-specific antibodies and the sVNT neutralising antibodies were significantly increased postbooster in the BNT arm. However, in accord with the literature, our study reported a significantly larger increase in the levels of S antigen-specific antibodies in individuals who received the BNT as opposed to the BIBBP booster vaccine [20,22]. There is no standard threshold for the types and levels of anti-SARS-CoV-2 antibodies that are needed to protect from disease. Some studies have attempted to model correlations between antibody titers and infection or disease outcome. In one model, comparison of mean normalized neutralization levels and reported vaccine efficacy levels across different platforms demonstrated a remarkably strong non-linear relationship model [23]. The risk of symptomatic COVID-19 was found to decrease with increasing levels of anti-spike and anti-RBD IgG and increasing live and neutralisation titres. However, no correlation was found with protection against asymptomatic infection [23]. It is also important to consider that besides from antibodies, other cell-dependent immune factors, such as T-cells, may also play a significant role in long-term protection against SARS-CoV-2 [24].
Attempts at correlating immunity with antibody levels have been made. In one study, S-antigen specific antibody titres of 264 BAU/ml (95 %CI: 108 -806) and a neutralisation titre of 26 IU/ml correlated with a vaccine efficacy of 80 % [25]. Other studies have also corroborated these findings, reporting a correlation between antibody titres and efficacy across several different vaccine platforms [26] and reporting the presence of neutralising antibodies, both from prior infection and from vaccination, to be significantly associated with protection from (re-)infection [27,28]. However, the antibody threshold below which protection is compromised has yet to be identified. This is the first study to report on the safety and immunogenicity of homologous BBIBP-CorV prime-boosters as compared to primary BBIBP followed by BNT162b2 boosters. A study that compared the immunogenicity of a group that had received two doses of BNT vaccine to a group that received two doses of BBIBP followed by a BNT booster reported higher humoral immunity induced by the BNT-boosted group. That the BBIBP-primed BNTboosted arm had higher humoral immunity than the BNT-primed arm is an indicator of the efficacy of the booster dose and that a BBIBP-primed BNT-boosted combination was able to produce superior results to BNT vaccination alone. Another study that compares a homologous inactivated whole-virion booster (Corona-Vac) to a heterologous CoronaVac with mRNA vaccine (BNT162b2) booster found the heterologous schedule to be superior, with an anti-spike IgG geometric fold-increase of 13.4. That study also compared the homologous CoronaVac schedule with two heterologous recombinant adenoviral vectored vaccines (Ad26.COV2-S, Janssen and ChAdOx1 nCoV-19, AstraZeneca), which similarly reported significant geometric fold-increases of 6.7 and 7.0, respectively [29].
A study from the United States was able to compare nine different prime-booster combinations of BNT162b2, mRNA-1273, and Ad26.COV2.S. The study concluded that both homologous and heterologous booster regimens were safe and immunogenic in adults who had completed their primary vaccination series 12 weeks earlier, and that antibody titer increases were generally the same or higher after heterologous boosting compared to homologous boosting. With regards to specific differences, at Day 15, the factor increase of binding and neutralising antibodies compard to baseline was highest in Ad26.COV2 primed individuals boosted with mRNA-1273. As for the raw increase in antibody titers, the highest increase was seen in the homologous mRNA-1273 primed and boosted individuals (though that the study used twice the FDA-authorized dose for the mRNA-1273 booster) [30]. This study thus offers compelling evidence for the role of the specific vaccinations used in the prime-boost schedule-regardless of homologous or heterologous status-in determining the affect on immungenicty; whether this may similarly reflect immunity against infection and transmission cannot be asserted.
Other heterologous and homologous COVID-19 vaccine combinations have been reported, particularly between the BNT and ChAdOx1 (ChAd) vaccines. A systematic review of four trials (five articles [31][32][33][34][35]) covering five different prime-boost schedules (single dose ChAd-primed, ChAd-primed + BNT-boosted, BNT-pri med + ChAd-boosted, BNT-Primed and boosted, and ChAdprimed and boosted) included 1862 participants overall [36]. The review found a heterologous ChAd-primed + BNT-boosted combination elicited the most robust T-cell response and highest neutralising antibodies out of all combinations. Interestingly, the primeboost order seemed to matter. The heterologous BNT-primed + C hAd-boosted combination induced a weaker immunogenicity than the homologous BNT-primed and boosted combination, while the ChAd-primed + BNT-boosted combination elicited stronger immunogenicity [36]. This demonstrates that the immune response is complex and it is not possible to predict the homologous versus heterologous response without clinical trial comparisons.
Other prime-boost combinations have also been studied. In a country-wide cohort study, a heterologous ChAd-primed + mRN A-1273 (mRNA) boosted combination had an efficacy of 79 % (95 %CI: 62-88, p < 0.001) against symptomatic COVID-19 disease, as compared to a 50 % efficacy in the homologous ChAd-primed and boosted arm [36]. This was corroborated by another smaller study that found a ChAd-primed + mRNA-boosted combination to be more potent against SARS-CoV-2 variants when compared to a ChAd-primed and boosted arm [37]. In a randomized controlled trial of 300 participants, a heterologous CoronoaVac-prim ed + AD5-nCOV-boosted group was found to have higher mean Table 3 Anti-SARS-CoV-2 antibody levels 8 weeks post booster administration in the heterologous BNT and homologous BBIBP booster arms.  Table 4 Baseline anti-SARS-CoV-2 antibody levels of participants in the homologous BBIBP group compared to levels at 8-weeks post BBIBP booster administration (N = 127). titres of anti-SARS-CoV-2 antibodies when compared to a homologous CoronaVac-primed and boosted group. A further study of a ChAd-primed + BBV152-boosted group also found higher mean titres of N-antigen and whole-virus neutralisation antibodies when compared to a homologous ChAd or BBV152-primed and boosted group [38]; however, the homologous ChAd group had a higher mean titre of anti-S1-RBD antibodies [39]. This study here and data from the literature suggests that heterologous prime-boost combinations from different vaccine platforms are effective. The immunologic mechanism underlying this improved response with heterologous vaccination is not clear, but may be related to a more diverse and robust immune stimulation due to the antigenic variations between the different platforms.

Type of antibodies
With regards to reactogenicity, our study found that the heterologous BNT-boosted group had a higher rate of non-serious adverse events compared to the homologous BBIBP-primed and boosted group. Specifically, the rates of injection site pain, fever, and fatigue were significantly higher. Similar findings have been reported with other heterologous combinations [15,37,38], while others reported no significant difference. Overall, the adverse events were all minor and self-limiting indicating that the heterologous combination as a safe and well tolerated option.

Strengths, limitations, and implications
A strength of this study was that both arms had similar basic demographic characteristics (adjusting for age), and nonsignificant differences between their baseline antibody levels. It is also a community trial, which allows for natural observations directly relevant to the real-world-setting. Limitations of this study were that this was a non-blinded non-randomized, observational trial, prone to unaccounted for bias. Additionally, we do not have Table 5 Baseline anti-SARS-CoV-2 antibody levels of participants in the heterologous BNT group compared to levels at 8-weeks post BNT booster administration (N = 119).

Type of antibodies
Baseline (Mean BAU/ml ± SD)  Table 6 Anti-SARS-CoV-2 antibody levels 8 weeks post booster administration across the booster arms stratified by age of participants*.    data on the real-world efficacy of the vaccine combinations in terms of COVID-19 infection and symptomatology. Data on different variants of concern, such as Omicron, would have also been beneficial considering the varying reports of efficacy [40]. Finally, although all participants were screened for COVID-19 infection prior to enrolment and any infected individual throughout study was excluded, we cannot rule out asymptomatic infection during study period as a confounder. The discussion on prime-boost COVID-19 vaccination schedules is of increasing importance with the appearance of novel variants [41,42] and the reduction in neutralization titres against some new strains that is reported to be up to 9-fold [43]. These emerging findings suggest that booster doses with altered epitope targets may have to be developed. Additionally, vaccine-induced antibody levels have been shown to wane after several months [6], which is a cause for concern. Breakthrough infections have also been reported at higher rates [12,13]. The role of cellular immunity and whether it may be sufficient for protection is still not clear [8]. The real-world evidence for boosters providing superior protection against symptomatic and severe disease and reducing mortality is emerging [44][45][46][47], with ongoing debate [48,49]. Of note, effective and safe mixing of different vaccine platforms may help facilitate collaborative approaches to addressing the lingering global vaccine inequity [16].
In conclusion, heterologous prime-boost vaccination with the mRNA BNT162b2 in those who had received two doses of inactivated virus BBIBP-CorV reflected in a more robust immune response against SARS-CoV-2 than the homologous BBIBP booster and appears safe. Further studies are needed to evaluate the underlying immunologic mechanisms behind the observed improved responses with heterologous prime-boost vaccination schedules, as well as randomized controlled trials to help identify the optimal permutations against new strains.

Contributors
MQ conceived the trial and MQ is the principal investigator. MQ, AA, PW and JJ contributed to the protocol and design of the study. SS, BMA, BA, AA, DQ, EA, AA, AM, AM, AH, SA and AMM led the implementation of the study. AMM, AEB, and NK did the statistical analysis and have verified the underlying data. SIM drafted the report. All other authors (SS, BMA, BA, AA, DQ, EA, PW, AH, AM, SA and AM) contributed to the implementation and data collection. SA and PW worked on the pharmacovigilance aspect of the trial. All authors reviewed and approved the final report.

Data Sharing
The study protocol is provided in the appendix. Individual participant data will be made available upon requests to the corresponding author; after approval of a request, data can be shared through a secure online platform.

Funding
No monetary funding was provided. This work was supported by the National Taskforce for the Combating of the Coronavirus (COVID- 19), the Ministry of Health -Bahrain, and the Royal College of Surgeons in Ireland -Bahrain.

Data availability
Data will be made available on request.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.