A prospective longitudinal cohort study was conducted in Medellín, Colombia (June 2022 to July 2023). It involved fully vaccinated children aged 3-11 years who received two doses (0.5mL ~600SU each) of CoronaVac^® vaccine (Sinovac Life Sciences), with an interdose interval of 28-100 days. The study was semi-blinded, with analysts unaware of exposure status until sample processing was completed.

The children were classified into two groups: “Exposed” and “Non-Exposed.” The “Exposed” group included children who met at least one of the following criteria before vaccination: (1) documented SARS-CoV-2 infection confirmed by antigen or RT-PCR testing, or (2) a reported history of close contact with a household member who tested positive for COVID-19 within 14 days of the household member’s diagnosis. The “Non-Exposed” group included children who did not meet either criterion.

Saliva was collected with DNA/RNA shield (Zymo, USA) every 15 days for SARS-CoV-2 testing via RT−qPCR, and blood was drawn at 2- and 6-months post-vaccination for immune response evaluation ( Figure 1 ). Data related to the safety of the vaccine were also recorded. The reagents used are detailed in Supplementary Table S1 . The study adhered to Helsinki Declaration and Resolution 8430 of 1993 of the Minister of Health of Colombia (mainly Chapter III “Research on minors”). It was approved by the Ethics Committee of the Universidad de Antioquia (Act 027/2022).

The RNA was extracted from saliva samples using a SaMag viral nucleic acid extraction kit and a SaMag-12 nucleic acid extractor (Sacace Biotechnologies, IT). RT−qPCR was performed using a BGI 2019-nCoV RT−PCR kit on a CFX-96 thermal cycler (Bio-Rad, USA).

Plasma was obtained by centrifuging blood at 800 × g for 10 minutes and stored at -80°C. Peripheral blood mononuclear cells (PBMCs) were isolated using Ficoll-Histopaque centrifugation gradient.

Anti-S IgG and anti-S/anti-N IgA antibodies against SARS-CoV-2 (Wuhan P0DTC2 and YP_009724397) were detected in plasma by ELISA (MyBioSource. MBS7612298/MBS7612290) following the manufacturers’ recommendations.

Colombian SARS-CoV-2 isolates were used for neutralization assays: An ancestral WT lineage (B.1, EPI_ISL_536399), Delta (B.1.617.2, EPI_ISL_5103929), Mu (B.1.621, EPI_ISL_4005445), and Omicron (BA.1.1, EPI_ISL_8374770). Virus manipulation was conducted in a biosafety level-3 laboratory.

Neutralizing antibody titers from plasma were assessed via a plaque reduction neutralization test (PRNT) as previously described (5), using serial dilutions (1:20 to 1:5120) of heat-inactivated plasma mixed with 200 PFU/0.1 mL of each virus, and incubated with Vero E6 cells (ATCC CRL-1586). For Omicron, 15 µg/mL of trypsin-EDTA was added before incubation. After 4-5 days, plaques were counted ( Supplementary Figure S1 ), and titers were measured as the dilution required to neutralize 50% of the viral plaque formation (PRNT50). A titer of ≥20 was considered positive for SARS-CoV-2 neutralizing antibodies (6).

Pools of lyophilized SARS-CoV-2 peptides (>70% purity by HPLC and MS, from GenScript, China) derived from WT and Omicron were designed based on literature and peptide affinity to human leukocyte antigens (HLAs) prevalent in Colombia (7) using TepiTool (http://tools.iedb.org/tepitool/). Each pool (WT and Omicron) consisted of 117 peptides (8–22 aa in length), 52 lineage-specific, and 65 conserved, from the spike (S), envelope (E) membrane (M) and nucleocapsid (N) proteins ( Supplementary Table S2 ). The peptides were reconstituted at 2000 µg/mL in DMSO.

Thawed PBMCs from 34 children (17 per group) were cultured in 96-well V-bottom plates at 8x10⁵ cells/mL in RPMI-1640 medium with 10% FBS, stimulated with 1 µg/mL anti-CD28 and anti-CD49d antibodies along with 5 µg/mL peptides from WT or Omicron for 18 hours at 37°C. Unstimulated PBMCs with the same antibodies and DMSO concentrations as the peptides served as negative controls (background). PBMCs stimulated with 10 µg/mL phytohemagglutinin or 40 ng/mL PMA and 500 ng/mL ionomycin were used as positive controls. All conditions included anti-human CD107a.

The PBMCs were stimulated as described above, labeled with the antibodies for AIM at 4°C for 30 minutes, acquired on an LSR Fortessa flow cytometer (Becton Dickinson, BD) and analyzed using FlowJo Software V10.9.0 (Tree Star, Inc., Ashland, USA). AIM⁺ CD4⁺ T-cells (CD137⁺OX40⁺ and OX40⁺CD25⁺) and AIM⁺ CD8⁺ T-cells (CD69⁺CD137⁺ and CD25⁺CD107⁺) were detected. The data were reported after background subtraction. Responder’s percentage was set at 0.02% for CD4⁺ and CD8⁺ T-cells. The differentiation state of AIM⁺ T-cells was evaluated by examining central, effector memory, and terminally differentiated (TEMRA) cell subpopulations.

After PBMCs stimulation as before, this time adding 10 µg/mL brefeldin A/monensin, cells were labeled with Viability Dye, anti-CD3, anti-CD4, and anti-CD8 at 4°C for 30 minutes. Then, cells were fixed and permeabilized using FoxP3/transcription factor staining buffer and labeled with antibodies against IFN-γ, TNF-α, IL-2, IL-10, perforin, and granzyme B at 4°C for 30 minutes. At least 100,000 events were acquired on an LSR Fortessa cytometer (BD) and analyzed using FlowJo V10.9.0 (Tree Star). Compensation for fluorochrome spillover was achieved using unstained and single-stained cells with each antibody.

Data was reported after background subtraction. Polyfunctional T-cells were evaluated using Boolean analysis and SPICE software v6.1 (Vaccine Research Center, NIAID/NIH, USA). Responders were defined as individuals with a value >0.001 after background subtraction. The polyfunctional index was obtained using the Funky Cells Toolbox, version 0.1.0 beta (http://www.funkycells.com/main).

The CBA Enhanced Sensitivity Flex Set system (BD) was used to detect IL-6, IL-10, IL-2, IL-4, and IL-17a levels in supernatants of PBMCs stimulated with SARS-CoV-2 peptides after background subtraction, according to the manufacturer’s instructions. Fluorescence was determined using flow cytometry (CitoFLEX, Beckman Coulter), and cytokine levels were analyzed using FlowJo V10.9.0 (Tree Star).

Since the data were not normally distributed according to the Shapiro-Wilk test, nonparametric tests were used for analysis, except for total specific anti-SARS-CoV-2 IgG and IgA, which were analyzed using paired T-tests. The Wilcoxon test estimated differences over time for each variant. The Mann-Whitney test compared Non-Exposed and Exposed at the same time point. The Wilcoxon or Friedman/Dunn correction tests compared variants/lineages for each group. For neutralizing antibodies, the geometric mean titer (GMT) and fold-change of the variants compared to the ancestral lineage were described. Statistical analysis and graphics were generated using GraphPad Prism 10.0.

Out of the 92 screened volunteers, 89 initiated the four-month follow-up. There were no differences in sex or age between Exposed and Non-Exposed children. Among the 7 children who tested positive for SARS-CoV-2 during the follow-up, 3 had a history or epidemiological link to SARS-CoV-19 infection, and 2 reported mild symptoms, with no differences between the Exposed and Non-Exposed groups ( Table 1 ). A total of 84 children (94.4%) completed the follow-up to evaluate the immunogenicity of CoronaVac^®; 38 were allocated to the “Exposed” and 46 to the “Non-Exposed” group. Reasons for withdrawal are shown in Figure 1 .

The predominant adverse event post-vaccination was pain at the injection site (30%), and the most frequent systemic adverse event was fever following both doses. There were no significant differences in the evaluated effects between doses or exposure status, but “other” effects showed a significant association (related to respiratory symptoms) with exposure status ( Table 2 ).

Circulating SARS-CoV-2-specific IgG (anti-S) and IgA (anti-S/anti-N) antibodies were measured. All children exhibited substantial levels of specific IgG and IgA, 2 months post-vaccination, which significantly decreased at 6 months post-vaccination, by at least 45.5% for IgG and by 42.2% for IgA, regardless of SARS-CoV-2 exposure (p<0.0001, Figures 2A, B ). Furthermore, the IgG and IgA titers were positively correlated (r=0.50, p<0.0001). Figure 2C ).

Neutralizing activity (PRNT50 titers ≥20) against B.1 and three variants was observed in >96% of subjects two months post-vaccination and remained at over 91.7% at six months post-vaccination ( Figures 2D, E ). The PRNT50 GMT against Delta and Mu (after two and six months), and against Omicron (at two months) was greater than against B.1 in both Exposed and Non-Exposed subjects, with most differences being statistically significant ( Figure 2E ). Although the median PRNT50 titers against the ancestral strain remained stable over time in both groups, significantly lower PRNT50 titers against Delta (p=0.004), Mu (p=0.0003), and Omicron (p<0.0001) were detected in the Exposed group at six months post-vaccination compared to two months. In Non-Exposed children, the reduction in neutralizing titers over time was observed only against Omicron ( Figure 2D , p=0.0034).

Cellular responses to the SARS-CoV-2 peptide pool through an AIM assay ( Supplementary Figure S2 ) revealed that most vaccinated children had detectable virus-specific AIM⁺CD4⁺ (coexpressing either CD137/OX40 or OX40/CD25), irrespective of the SARS-CoV-2 variant (WT or Omicron), SARS-CoV-2-exposure, or time post-vaccination ( Figures 3A, B ).

In contrast, the frequency of AIM⁺CD8⁺ T-cells (CD69⁺CD137⁺) in response to Omicron was lower than to WT in both Exposed (p=0.043) and Non-Exposed (p=0.011) at 6 months post-vaccination ( Figure 3C ). The CD25/CD107a combination also detected these differences at 2 months post-vaccination (p<0.05, Figure 3D ). More children responded through AIM⁺CD4⁺ than AIM⁺CD8⁺ T-cells (p<0.05, Fisher’s test), regardless of the variant or time post-vaccination. Neither variants, exposure status, nor time post-vaccination affected the memory profile of virus-specific T-cells, with predominant central memory CD4⁺ T-cells and effector memory CD8⁺ T-cells ( Figures 3E, F ).

Cytokine and cytotoxic molecule production were subsequently analyzed in CD4⁺ and CD8⁺ T-cells after peptide stimulation ( Supplementary Figure S3 ). Lower frequencies of IFN-γ⁺CD4⁺ (p=0.0041), TNF-α⁺CD4⁺ (p=0.0380) and IL-2⁺CD4⁺ T-cells (p=0.0214) were observed in response to Omicron than WT in the Exposed children, at 2 months post-vaccination; however, this response decreased over time, only in response to WT (IFN-γ⁺CD4⁺; p=0.0017, TNF-α⁺CD4⁺; p=0.0239 and IL-2⁺CD4⁺; p=0.0289. Figures 4A–C ).

Similarly, IFN-γ⁺CD4⁺ T-cell frequency was also lower in response to Omicron compared to WT in Non-Exposed subjects at 2 months post-vaccination (p=0.0066. Figure 4A ). The response to WT in Non-Exposed was slightly lower than in Exposed at 2 months post-vaccination (p=0.0447. Figure 4A ). The frequencies of IFN-γ-, TNF-α-, and IL-2-producing CD4⁺ T-cells were positively correlated with the frequency of AIM⁺CD4⁺ T-cells (CD137⁺OX40⁺) in both groups, with correlation coefficients ranging from 0.49 to 0.61 (p<0.05, Supplementary Figure S4 ). No significant differences were found in IL-10⁺, CD107a⁺, nor CD107a⁺Perforin⁺ CD4⁺ T cell frequencies among any of the conditions or groups evaluated ( Supplementary Figures S5A–C ). However, CD4⁺ T-cells coexpressing CD107a⁺Granzyme B⁺ were higher in response to Omicron peptides compared to WT peptides in the Exposed group at 6 months post-vaccination (p=0.0391, Figure 4D ).

Similar to those of CD4+ T-cells, the frequencies of IFN-γ⁺CD8⁺ (p=0.0010) and TNF-α⁺CD8⁺ (p=0.0273) T-cells were significantly lower in response to Omicron than in response to WT in Exposed children at 2 months ( Figures 4E, F ), and also at 6 months post-vaccination for IFN-γ⁺CD8⁺ T-cells (p=0.0093). The IFN-γ⁺CD8⁺ response to WT in Exposed subjects decreased at 6 months compared to that at 2 months (p= 0.0245. Figure 4E ).

In addition, a greater percentage of responders and frequency of IL-2⁺CD8⁺ T-cells were observed in the Non-Exposed compared to the Exposed group in response to WT at 2 months (p=0.0458) and to Omicron at 6 months (p=0.0232) ( Figure 4G ). No significant differences were found in IL-10⁺CD8⁺ T-cell frequencies ( Supplementary Figure S5D ).

A significant decrease in the frequency of total CD107⁺CD8⁺ T-cells was observed at 6 months compared to 2 months post-vaccination in response to WT peptides in both Non-Exposed (p=0.0137) and Exposed (p=0.0356) groups, and in response to Omicron, only in the Exposed group (p=0.0059) ( Figure 4H ). The frequency of CD107a⁺perforin⁺-expressing CD8⁺ T-cells was lower in response to Omicron compared to WT in Non-Exposed children at 2 months post-vaccination (p=0.0391) ( Figure 4I ).

Coronavac^® induced polyfunctional CD4⁺ T-cell responses, predominantly coexpressing TNF-α, IFN-γ, and IL-2, against both viral variants. However, the frequency of trifunctional (p=0.0012) and bifunctional (p=0.0215) cells was lower in response to Omicron compared to WT peptides in Exposed children at 2 months post-vaccination. The response to WT in Exposed children also decreased over time (3 functions: p=0.0203, 2 functions: p=0.0074). Similarly, the bifunctional response to WT was also lower in Non-Exposed compared to Exposed children at 2 months (p=0.0169) ( Figure 5A ).

Bifunctional IL-2⁺TNF⁺ cells in the Exposed children were lower in response to Omicron than WT at 2 months (p=0.0020) and this response to WT also decreased over time (p=0.0021) ( Figure 5B ). In addition, the IFNγ⁺IL-2⁺TNF-α⁺perforin⁺ profile was less frequent in response to Omicron than in response to WT in Exposed children at 2 months (p=0.0078) ( Figure 5C ). The polyfunctionality index in CD4⁺ T-cells was lower in response to Omicron than WT in Exposed children at 2 months. It was higher in Exposed than Non-Exposed children in response to the WT, decreasing over time ( Figure 5D ).

In contrast, CD8⁺ T-cells were less polyfunctional overall, with no significant differences across subjects ( Figure 6A ). Bifunctional responses against both variants were predominantly granzyme B⁺perforin⁺, followed by IFN-γ⁺ TNF-α⁺ profile for WT and IL-2⁺IL-10⁺ profile for Omicron ( Figure 6A ). A significant reduction in the frequencies of CD8⁺ T-cells exclusively expressing IFNγ⁺ (p=0.0032) ( Figure 6B ), IFNγ⁺TNF-α⁺ (p=0.0020) ( Figure 6C ) and IFN-γ⁺TNF-α⁺granzyme-B⁺perforin⁺ (p = 0.0156) ( Figure 6D ) was observed in response to Omicron compared to WT in Exposed children at 2 months. Additionally, the frequency of IFNγ⁺TNF-α⁺ profile was lower for Omicron than WT in Non-Exposed children at 6 months post-vaccination (p=0.0020) ( Figure 6C ). A higher polyfunctional CD8⁺ T-cell index was observed for WT compared to Omicron in Exposed children at 2 months post-vaccination (p=0.0097) ( Figure 6E ).

When cytokine production was measured in the supernatants of cultures stimulated with WT or Omicron peptides, IL-2 production was detected in all subjects ( Figure 7A ); IL-17A levels were low in all subjects ( Figure 7B ), and IL-4 was not detected. IL-6 levels in response to Omicron in Non-Exposed subjects increased at 6 months compared to 2 months post-vaccination (p=0.0027). Still, they were lower in Non-Exposed compared to Exposed at 2 months post-vaccination (p=0.019) ( Figure 7C ). IL-10 levels in response to Omicron in the Exposed group increased at 6 months compared to 2 months post-vaccination (p=0.0107) ( Figure 7D ). The IL-2 levels in response to WT were positively correlated with the frequency of AIM⁺CD4⁺ (CD137⁺OX40⁺; r=0.4597, p<0.0001) T-cells in Exposed and Non-Exposed subjects analyzed together ( Figure 7E ). However, these correlations were not observed for IL-2 produced in response to Omicron.