Serological insights from SARS-CoV-2 heterologous prime and boost responses in Thailand

Abstract

During the COVID-19 pandemic, heterologous vaccination strategies were employed to alleviate the strain on vaccine supplies. The Thailand Ministry of Health adopted these strategies using vector, inactivated, and mRNA vaccines. However, this approach has introduced challenges for SARS-CoV-2 sero-epidemiology studies. Our study analysed 647 samples from healthcare workers who received CoronaVac, ChAdOx1 nCoV-19, and BNT162b2 vaccines. The serological profile encompassed responses to various SARS-CoV-2 variants and vectors, measuring IgG, IgM, and IgA isotypes, alongside IgG avidity assays. The results demonstrated that heterologous CoronaVac/ChAdOx1 nCoV-19 schedules elicited significantly stronger antibody responses compared to homologous schedules (IgG: 1.2-fold, IgM: 10.9-fold, IgA: 3.1-fold increase). Additionally, a heterologous BNT162b2 boost at 4-weeks post-initial vaccination showed greater antibody levels than a ChAdOx1 nCoV-19 boost (IgG: 1.1-fold, IgM: slight decrease, IgA: 1.5-fold increase). Using a combination of three analytes, IgG against wild-type Spike trimer, nucleoprotein and Omicron receptor binding domains, enabled the clustering of responses within a statistical Gaussian mixture model that successfully discriminates between breakthrough infections and vaccination types (F-score = 0.82). The development of statistical models to predict breakthrough infections can improve serological surveillance. Overall, our study underscores the necessity for vaccine (re-)development and the creation of serological tools to monitor vaccine performance.

Keywords: Antibody; COVID-19; SARS-CoV-2; Vaccine.

Fig. 1

Graphical summary of sample collection across groups consisting of Sinovac CoronaVac (S), AstraZeneca ChAdOx1 nCoV-19 (A) or Pfizer BNT162b2 (P) vaccinations. For the AA group (homologous AstraZeneca), blood samples were collected 4 weeks after the second dose (n = 64). In the SA group (heterologous of Sinovac followed by AstraZeneca), samples were obtained at 2 weeks (n = 81), 4 weeks (n = 80), and 12 weeks (n = 80) post-vaccination. The SS group (homologous Sinovac) had samples collected over a range of 4 to 36 weeks (n = 270). After receiving two doses, some of SS, participants received booster doses; the SSA group (boost with AstraZeneca) had samples collected at 4 weeks (n = 163) and 12 weeks (n = 163), while the SSP group (boost with Pfizer) had samples taken at 2 weeks (n = 68) and 4 weeks (n = 68).

Fig. 2

Pan-isotype profile of antibody responses to homologous and heterologous vaccination. IgG, IgM IgA and IgG avidity MFI after second homologous or heterologous vaccination with AstraZeneca ChAdOx1 nCoV-19 and ChAdOx1 nCoV-19 (AA) (n = 64), Sinovac CoronaVac and AstraZeneca ChAdOx1 nCoV-19 (SA) (n = 81), and Sinovac CoronaVac and CoronaVac (SS) (n = 270). Negative controls – unvaccinated individuals, samples collected prior to 2019 (NEG) (n = 64). Samples were collected 3.5 to 8.5 weeks after second primary vaccination. MFI was adjusted to account for batch effects. Wilcoxon signed-rank test was applied as a measure of significance (* p < 0.05; ** p < 0.01; ***p < 0.001; ****p < 0.0001). The seropositivity thresholds (high, low) are based on fitting for each antigen a Gaussian mixture model (GMM) to classify the MFI levels into three groups (negative/low, intermediate, and high).

Fig. 3

IgG significantly increases in homologous CoronaVac recipients after heterologous boost with ChAdOx1 nCoV-19 and BNT162b2. The IgG panel against the Spike trimer, receptor-binding domain (RBD), nucleoprotein and chimpanzee adenovirus are displayed in both SSA (n = 163) (top) and SSP (n = 68) (middle) regimens together with the level of antibodies in both groups at 4 weeks (bottom). *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. The seropositivity thresholds (high, negative) were determined by fitting a Gaussian mixture model (GMM) to the MFI levels for each antigen, classifying them into three groups (negative/low, intermediate, and high).

Fig. 4

Finite Gaussian mixture modelling (GMM) to identify clusters (components) using combinations of antigen and isotype responses (dimensions), with the overlayed colours indicating actual vaccine groups (Negative/Pre-COVID (n = 64); SS (n = 270); AZ = SA (n = 81) or AA (n = 64); Positive = convalescent / vaccine breakthrough (n = 281); Boost = SSA (n = 163) or SSP (n = 68). (Top) 2D model (WT SARS-CoV-2 spike timer, WT nucleoprotein IgG) with 4 components (F-score = 0.88); the ellipses are the estimated covariances, centred on the mean for each component. (Bottom) 3D model (WT SARS-CoV-2 spike timer, WT nucleoprotein IgG, RBD Omicron IgG) with five components (F-score = 0.82).