Humoral immunity was assessed by ELISA in pediatric SCTR and age-matched healthy controls approximately 5 weeks after the second and third vaccine dose. Spike S1 specific IgG levels were not substantially different between both groups at either time point. However, whereas quantities in healthy individuals only slightly increased after the third dose, SCTR significantly benefited from the booster as mirrored by more than doubled IgG levels ( Figure 1A , left). This finding was also applicable to patients in a pairwise comparison after second vs. third dose where fewer individuals were included due to paired sample availability ( Figure 1A , right). Similar results were obtained for specific IgA levels ( Figure 1B ). After the second dose, significantly fewer patients than controls showed Omicron neutralizing capacity above threshold. Importantly, this feature normalized after the third dose, resulting in similar portions of responders in healthy individuals and patients ( Figure 1C , left) as well as percentage of neutralizing capacity ( Figure 1C , middle). Neutralization capacity also increased after the third dose in a pairwise comparison in SCTR ( Figure 1C , right). Neutralizing capacity levels did neither correlate with stem cell donor age nor time since transplantation ( Supplemental Figure 1A ). Whereas neutralizing capacity levels positively correlated with those of specific IgG except for HC after the 2nd dose ( Supplemental Figure 1B ), such associations were not noted for IgA ( Supplemental Figure 1C ). Datasets from the few individuals acquiring a SARS-CoV2 infection during the study period were included in all figures for gross comparison. However, as also mentioned in the respective figures and legends, they were excluded from statistics.

Overall, absolute counts of CD19⁺ B cells were within the normal range for the majority of patients ( Supplemental Table 2 ). Spike RBD-specific B cells were detected with fluorescently labeled probes as exemplarily illustrated in Supplemental Figure 2A and reported earlier (16). Rates of responders showing Spike RBD-specific B cell were similar in controls and patients at both time points ( Figure 2A , left). Frequencies substantially rose by booster vaccination in controls, but not in patients, remaining significantly diminished even after the third dose after multiple comparisons testing ( Figure 2A , middle). In a pairwise comparison with a limited set of patient samples, RBD⁺ B cell frequencies showed a significant increase between both time points ( Figure 2A , right). The few SCTR with neutralizing capacity above threshold after the second dose showed significantly elevated frequencies of specific B cells ( Figure 2B ). Of note, specific B cell frequencies at both time points were positively correlated with increasing time since transplantation ( Figure 2C , left), but not with stem cell donor age ( Figure 2C , right), specific IgG ( Supplemental Figure 3A ), IgA ( Supplemental Figure 3B ) or neutralizing capacity, with the exception of SCTR after the 2^nd dose ( Supplemental Figure 3C ). We did not identify differences in specific class-switched CD27⁺IgD^- memory B cell frequencies between groups or time points ( Figure 2D , left). However, patients showed a clear trend towards increased portions of non-class-switched CD27⁺IgD⁺ memory B cells after booster vaccination as compared to controls ( Figure 2D , right). Neither frequencies of class- ( Supplemental Figure 3D ) nor of non-class-switched memory B cells ( Supplemental Figure 3E ) correlated with specific IgG levels with the exception of the CD27⁺IgD⁺ subset being negatively associated with IgG after the 2^nd dose in SCTR.

In summary, pediatric SCTR require a third vaccine dose for development of comparable humoral responses as controls. In addition, they show quantitative and qualitative impairments within the vaccine-specific B cell compartment.

Spike-specific CD4⁺ T cells were identified as depicted in Supplemental Figure 4A according to co-expression of CD137 and CD154 after peptide mix stimulation, as reported earlier (17–19). Numbers of individuals showing CD4⁺ T cell responses were not substantially different between groups ( Figure 3A , left). Interestingly, SCTR showed significantly higher frequencies of specific T cells than healthy individuals after booster vaccination ( Figure 3A , right). The magnitude of responses in patients significantly correlated with stem cell transplant donor age after the third dose ( Figure 3B , left), but not with time since transplantation ( Figure 3B , right). Frequencies of specific CD4⁺ T cells did not correlate with IgG levels for any of the groups ( Supplemental Figure 5A ), but with IgA for HC after the second dose ( Supplemental Figure 5B ) and neutralizing capacity levels in SCTR after the second dose ( Supplemental Figure 5C ). Both groups did not exhibit differences with respect to frequencies of specific CD45RO⁺CD62L^- effector/memory- or CD45RO^-CD62L^- effector-type T cells ( Figure 3C ). However, functional analysis revealed diminished portions of CD4⁺IFNγ⁺ T cells after the second and third dose in patients compared to controls ( Figure 3D , left). We further noted the opposite trend for IL-2, as reflected by increased frequencies in SCTR at both time points ( Figure 3D , middle). Neither of these observations did apply to frequencies of Spike-specific IL-4⁺ T cells ( Figure 3D , right). Polyfunctionality analysis did not reveal appreciable differences in frequencies of cells secreting none, one, two ( Supplemental Figure 5D , left), all three ( Supplemental Figure 5D , left and right) or any of the three cytokines ( Supplemental Figure 5E ) between groups.

We only noted low responder rates with respect to Spike-specific CD8⁺ T cells ( Supplemental Figure 5F , left) that were detected based on CD137 and IFNγ co-expression ( Supplemental Figure 4 ), as shown before (20). Therefore, we did not conduct statistical analysis for comparing frequencies in patients and controls ( Supplemental Figure 5E , right). Absolute CD8 counts were consistently within the normal range ( Supplemental Table 2 ), therefore not providing an obvious explanation for the low responder rates observed.

In summary, SCTR are characterized by elevated frequencies of vaccine-specific CD4⁺ T cells showing a functional programming towards increased IL-2- at the expense of anti-viral IFNγ production.

Given the small size of our pediatric cohort, further stratification of patients for assessment of distinct immunological features was only meaningful for patients experiencing GvHD or not, representing the largest subgroups, respectively. It is noteworthy that as opposed to all previous analyses, also patients with breakthrough infections during the study were included in statistical evaluations. Importantly, SCTR with GvHD developed significantly lower IgA and neutralizing antibody levels after the second dose, an effect that disappeared after the third dose ( Supplemental Figure 6A ). We did not observe significant differences with respect to vaccine-specific CD4⁺ and B cell frequencies between groups ( Supplemental Figures 6B, C ).

This observational study was conducted between January and June 2022 with pediatric stem cell transplant recipients (SCTR) recruited from the Charité Department of Pediatric Oncology and Hematology. Age-matched healthy controls were largely siblings of patients and not diagnosed with any acute or chronic medical condition. Basic characteristics of both groups are depicted in Table 1 , whereas further details of enrolled patients are included in Supplemental Table 1 . SARS-CoV-2 specific humoral and/or cellular responses were analyzed after second and third vaccination. Vaccination data, including vaccination type, number of doses, and vaccination dates were obtained from participants at visits. All individuals were vaccinated with the SARS-CoV2 vaccine BNT162b2 (Comirnaty; BioNTech/Pfizer) according to the manufacturer´s guidelines: all persons aged 12 and above were vaccinated with 30 µg. For children aged 5-11, the dosage was adjusted to 10 µg. Infants and children aged from 6 months to 4 years were vaccinated with the recommended dose of 3 µg.

Blood and serum samples were collected approximately 5 weeks after vaccinations with no significant differences between groups. The local ethics committee of the Charité Universitätsmedizin Berlin, Germany (EA2/227/21) approved this study, and all volunteers and/or their legal guardians provided informed consent to participate.

Previous SARS-CoV2 infection during the study was assessed based on medical history, frequent point of care antigen testing and a SARS-CoV2 nucleoprotein specific ELISA (Euroimmun). Levels of SARS-CoV2 S1 domain specific IgG (QuantiVac, Euroimmun) and IgA (Euroimmun) were also determined by ELISA. Results for IgA and nucleoprotein were calculated as the O.D. ratio of the respective sample over the O.D. of the calibrator provided with the ELISA kit. An O.D. ratio of >1.1 was considered positive, of >0.8 to <1.1 borderline, and of <0.8 as negative. Serum samples exceeding O.D. ratios of 8 were re-measured after pre-dilution. Results for Spike S1 domain specific IgG were provided as BAU/ml and considered positive when above the threshold of ≥35.2 BAU/ml. To quantify virus neutralizing antibodies against SARS-CoV2 Omicron variant, a competitive binding ELISA (sVNT kit, GenScript) was used. In this assay, neutralizing serum antibodies compete for binding to recombinant, horseradish peroxidase-labeled SARS-CoV2 RBD (Omicron variant) with human angiotensin-converting enzyme 2 receptor protein. According to the manufacturer´s instructions, a neutralizing capacity above 30% was considered positive.

PBMCs were isolated from EDTA blood using Lympho24+ Spin Medium (pluriSelect, Leipzig) by means of density gradient centrifugation. For analysis of SARS-CoV2 vaccine-specific B cells by flow cytometry, 5×10⁶-1×10⁷ mononuclear cells were stained with the antibodies listed in Supplemental Table 3 . Among B cells, gated as single live CD19⁺CD3^-CD14^-CD56^- lymphocytes ( Supplementary Figure 2A ), vaccine-specific B cells were detected by co-staining with fluorescently labeled recombinant RBD protein and recombinant full spike protein (both alpha-variant, RnD Systems, Minneapolis, MN, United States) as described previously (18, 19). This double labeling strategy was chosen also in agreement with other groups to reduce background staining (36, 37).

Vaccine-specific T cells were identified after stimulation of 3 × 10⁶ PBMCs for 16h at 37°C and 5% CO₂ with overlapping 15-mer peptides covering the complete SARS-CoV2 spike protein (alpha-variant, JPT, Berlin, Germany) at a final concentration of 0.5 µg/ml per peptide as described earlier(17). Equal quantities of DMSO as contained in the peptide mix were added to unstimulated controls. For intracellular molecule retention, Brefeldin A was added at a final concentration of 10 µg/ml. After stimulation, cells were stained with a surface antibody mixture at room temperature for 20 minutes, followed by fixation and permeabilization (FACS Lysing Solution/FACS Perm II Solution, BD Bioscience, Heidelberg, Germany) and staining with antibodies against intracellular molecules ( Supplemental Table 4 ). Identification of antigen-reactive CD4⁺ T cells was based on CD137 and CD154 co-expression ( Supplementary Figure 4 ). When stimulated samples contained a minimum of twenty events and at least twofold higher frequencies of CD137⁺CD154⁺ T cells compared to the respective unstimulated control (stimulation index of 2), a response was considered positive. Cells were measured with a BD FortessaX20 flow cytometer (BD Bioscience, Heidelberg, Germany).

FlowJo 10 (BD) was used for analysis of flow cytometric data. Frequencies of spike-specific CD4⁺ T cells were background (unstimulated control) subtracted. Boolean gating in FlowJo was used for quantification of cytokine co-expression. Statistical analysis and graph preparation was performed in GraphPad Prism 8 (Graphpad, San Diego, CA, USA). Data was tested for normal distribution with the Kolmogorov-Smirnov test. In case of normal distribution an ordinary one-way ANOVA with consecutive Sidak’s multiple comparison test was used. When data sets did not show normal distribution, a Kruskal-Wallis with Dunn’s multiple comparison was applied. For comparison of two groups a Mann-Whitney test was performed. Analysis of contingency tables was conducted by applying Fisher´s exact test ( Supplemental Table 5 ). Statistical significance was considered for p>0.05.