The Ciao Corona study is part of the nationally coordinated research network Corona Immunitas [23] in Switzerland. The study protocol was published prospectively (ClinicalTrials.gov identifier: NCT04448717) [24] and the seroprevalence results of all testing rounds can be found elsewhere [22, 25–28]. This cellular immune response analysis is based on a subset of children and adolescents who participated in the testing rounds of either November/December 2021, June/July 2022, or both testing rounds within the Ciao Corona sample. The study was conducted in the canton of Zurich, which has a population of 1.52 million people. This represents 18% of the Swiss population and includes residents from diverse ethnic and linguistic backgrounds, encompassing inhabitants from both rural and urban areas. The study was approved by the Ethics Committee of the Canton of Zurich, Switzerland (Registration No. 2020-01336). Prior to the study enrolment, children and adolescents provided verbal and parents/proxies provided written informed consent.

In Ciao Corona, we included randomly selected primary schools from the canton of Zurich and matched the nearest geographically located secondary school. We invited schools based on the population size of the 12 districts. We contacted 156 schools to take part in the study, and a total of 55 schools, 54 public and 1 private, decided to participate. Classes were stratified by school level and then randomly selected, encompassing grades 1–2 (children aged 6–8) at the lower school level, grades 4–5 (children aged 9–11) at the middle school level, and grades 7–9 (adolescents aged 12–14) at the upper school level. All children and adolescents from the selected classes were eligible to participate at any point in the study. We conducted five testing rounds between Jun/Jul 2020 and Jun/Jul 2022 and collected venous blood samples for antibody detection (anti-spike IgG and anti-nucleocapsid IgG). The first testing round (T1) took place in June/July 2020, the second (T2) in October/November 2020, the third (T3) in March/April 2021, the fourth (T4) in November/December 2021 and the last fifth (T5) testing round in June/July 2022.

In Nov/Dec 2021 we selected a convenience-based subsample of children and adolescents (N total = 109) for analysis of cellular immune responses, limiting sampling to the first five children and adolescents meeting the inclusion criteria per testing day due to limited resources. We invited children and adolescents that previously participated in at least one, but preferably two consecutive testing rounds (i.e., T1-T3). We included unvaccinated seropositive and seronegative children and adolescents in November/December 2021 and followed them up in June/July 2022, regardless of their vaccination status. For this subsample we collected additional venous blood for Interferon (IFN)-gamma ELISpot (N = 109) and IFN-gamma Release Assay (IGRA) (N = 26, randomly selected)-based analyses.

Children and adolescents from the longitudinal cohort were divided into four different groups based on their vaccination and infection history to analyse the evolution of anti-spike IgG and neutralizing antibody levels over time. Adolescents aged 16 and older were eligible for vaccination starting in May 2021, those aged 12 to 15 became eligible in mid-June 2021, and children aged 5 to 11 were eligible starting in January 2022 [29]. Individuals who never tested positive for anti-spike IgG were labelled as negative and unvaccinated children and adolescents who tested positive for anti-spike IgG at any study point as infected. Individuals who were vaccinated but always tested negative for anti-spike IgG prior to vaccination and afterwards never tested positive for anti-nucleocapsid IgG were categorized as vaccinated. Finally, those who were seropositive before vaccination, or those who were vaccinated and tested positive for anti-nucleocapsid IgG, were classified as hybrid.

In all testing rounds venous blood samples were collected and centrifuged to isolate plasma. Plasma aliquots were stored at −20 °C prior to antibody analyses. To isolate peripheral blood monocytic cells (PBMCs), blood samples were subjected to density-gradient centrifugation using Ficoll–Paque (density 1.077 g/mL). PBMCs were frozen in freezing media (90% fetal bovine serum (FBS, Pan Biotech) mixed with 10% dimethyl sulfoxide (DMSO, Sigma)) at −80 °C and subsequently transferred to liquid nitrogen prior to use in the ELISpot assay.

We employed the Sensitive Anti-SARS-CoV-2 spike Trimer Immunoglobulin Serological (SenASTrIS) test to detect SARS-CoV-2-specific antibodies targeting spike and nucleocapsid proteins. The Bio-Plex 356 Amine Coupling Kit (Bio-Rad, Catalogue 10000148774) was used to covalently couple MagPlex beads to either the SARS-CoV-2 spike protein trimer or nucleocapsid protein following the manufacturer’s instructions. Diluted protein-coupled beads were added to each well of the flat bottom 96-well plates (Bio-Plex Pro, Bio-Rad) and washed with PBS using a magnetic plate washer (MAG2x programme).

Next, 50 μL of each plasma sample was diluted 1:300 in PBS and added to the wells. As negative control, a pool of pre-pandemic healthy human serum was utilized (BioWest human serum AB males; VWR). The plates were incubated for 1 h at room temperature with shaking, followed by PBS washes, and then incubated with 50 μL of a 1:100 dilution of polyclonal Goat F(ab’)2 anti-human IgA-PE (for the anti-IgA assay; Southern Biotech; Catalogue 2052-09) or polyclonal Goat anti-human IgG-PE (for the anti-IgG assay; OneLambda; Catalogue LS-AB2) as a secondary antibody for additional 45 min at room temperature while shaking. Following incubation, the samples were washed with PBS, resuspended in reading buffer, and mean fluorescence intensity (MFI) values were measured for each sample using a Bio-Plex (Luminex) 200 plate reader with the Bio-Plex Manager software (version 6.2; Bio-Rad). The MFI ratio was calculated by dividing the MFI value of each plasma sample by the mean value of the negative control samples. For both anti-spike IgG and anti-nucleocapsid IgG antibodies, samples were considered seropositive if their MFI ratios were equal to or exceeded the cutoff value of 6, corresponding to the assay’s 99.2% specificity and 94.0% sensitivity. The cutoff of 6 was established based on results from negative control samples and from SARS-CoV-2 PCR-positive individuals. As the assays used in this study demonstrated stable performance with only minimal decay over a period of up to 8 months post-infection, we did not account for sensitivity changes due to time-related decay.

T cell-mediated responses of our subsample were evaluated as described previously [30, 31] by IFN-gamma ELISpot (R&D Systems, Catalog EL285). In brief, cryopreserved PBMC samples were thawed and seeded at density of 5e5 cells per well. Cells were stimulated for 20 h at 37 °C and 5% CO2 with 15mer peptide pools covering the entire amino acid sequences of viral M or N proteins, the S1 domain from the spike protein, or a combination of the predicted immunodominant peptides of the spike protein of which most belong to the S2 domain (M, N, S1, and S PepTivator peptide pools, respectively; Miltenyi Biotec). Peptides were used at final concentration of 0.6 nmol (∼1 μg/mL) per peptide. As a negative control, unstimulated cells were incubated in culture medium alone and as a positive control, 2.5e5 cells per well were stimulated with 10 μg/mL anti-CD3 antibody (Clone OKT3, Miltenyi Biotec, Catalog 130–093-387). Spot counts were determined using an AID iSpot Reader System and analyzed with EliSpot 7.0 software (AID). Spot numbers in negative control wells were doubled and subtracted from test wells with negative counts set to zero. Results were expressed as spot-forming units (SFU) per 1e6 PBMCs.

From a subsample of 26 children and adolescents sampled in Jun/Jul 2022, T cell responses were further characterized by IGRA as described previously [31]. Briefly, whole blood was collected in Li-heparin vacutainer tubes (BD) and 250 uL per well plated in 96 well U-bottom plates (Sarstedt). Samples were stimulated with combined pools of M and N (M/N) peptides or S1 and S (S1/S2) peptides at a final concentration of 0.6 nmol (∼1 μg/mL) per peptide and incubated for 20 h at 37 °C and 5% CO2. As a negative control, unstimulated blood, incubated without peptide was included and, as a positive control blood was stimulated with 10 mg/mL anti-CD3 antibody. After incubation, stimulated supernatants were stored at −20 °C prior to analysis of IFN-gamma levels by ELISA assay (Human IFN-gamma DuoSet ELISA kit, R&D Systems, Catalog DY285B, and DuoSet ELISA Ancillary Reagent Kit 2, R&D Systems, Catalog DY008), following the manufacturer’s instructions. Samples were measured using a Multiskan SkyHigh Microplate Spectrophotometer with SkanIt software (Thermo Fisher Scientific) set to a wavelength of 450 nm with a 540 nm correction and IFN-gamma concentrations calculated from absorbance values by applying a 4 PL standard curve. Concentrations from the unstimulated negative controls were subtracted from each test well with negative values adjusted to zero and expressed as IFN-gamma in pg/mL. Results were excluded if background IFN-gamma concentrations in unstimulated control wells exceeded five times the median of the test wells or if values of anti-CD3-stimulated positive controls were negative.

The study focused on the general population of children and adolescents, unlike other studies that primarily included specific subpopulations, such as hospitalized children, making the findings more broadly applicable. In contrast to these studies, we had a larger sample size, which strengthens our findings about T cell-mediated responses in this population. The longitudinal study design allowed us to assess changes in both cellular and humoral immune responses over 6 months, taking into accounting for infections, vaccinations, and their combinations throughout the COVID-19 pandemic.

However, there are several limitations that need to be acknowledged. First, the exact timing of SARS-CoV-2 infections in children and adolescents was not known due to the sero-epidemiological study design. This limits our ability to account for the time to and after immune-modifying events (such as infection and/or vaccination), which may influence the strength and detectability of immune responses and thus contribute to the lack of clear correlations between measures. Second, the study relied on ELISpot and IGRA assays to assess immune responses, which, while widely used, have limitations in terms of sensitivity and comprehensiveness in capturing the full spectrum of immune activity. Third, the analysis was restricted to only four domains of the T cell-mediated response, which might not capture the entire complexity of T cell-mediated immunity, potentially overlooking other important aspects of the immune response. Finally, selection bias may be a concern due to our subsample. However, comparisons with the full sample (see recent publication [22]) showed no differences except in vaccination status, which reflects our study’s inclusion criteria. Fourth, infections in our population-based cohort were predominantly asymptomatic to mild, and as we did not collect detailed data on infection severity, it remains uncertain whether our findings can be generalized to individuals with more severe disease.

In conclusion, this study indicates that circulating, virus-specific T cell responses in the general population of children and adolescents are low, heterogenous and not reflective of prior infections and vaccinations. Adding the assessment of cellular immunity to serological surveillance of the general population of children and adolescents appears to be of limited value but comes with substantial effort and cost. Thus, monitoring humoral immunity is likely a more cost-effective approach and should be the primary focus of immunological surveillance in the general paediatric population.