A multi-center prospective observational cohort study was conducted between July 2021 and March 2024 at seven adult and seven pediatric centers to assess COVID-19 vaccination immunogenicity and safety in IEIs. The study was conducted in accordance with the Declaration of Helsinki. The Ottawa Hospital Research Institute was the lead site and obtained overarching ethics approval (CTO#1978) for Ontario sites. Each study site also obtained approval from their local ethics board prior to study commencement. All participants provided written informed consent or assent before enrollment. There was no restriction on the number of COVID-19 vaccine doses received prior to study enrolment if they were less than 24 weeks after dose 4 and received any of the following vaccines available in Canada: BNT162b2, mRNA-1273, or ChAdOx1 nCoV-19.

The current article reports data on adult participants aged 18 years or older and up to 24 weeks after dose 3. Eligible participants with IEIs were recruited into three study subgroups based on the type of IEIs (24): A) predominant antibody deficiency (PAD), B) immunodeficiencies affecting cellular and humoral immunity [combined B-cell and T-cell immunodeficiency (CID)], and C) defects of intrinsic and innate immunity (III) or other IEIs not meeting criterion A) or B). Healthy individuals who did not report any known immunocompromising conditions at the time of initial study screening were recruited as controls. Additional healthy controls meeting our eligibility criteria were recruited from the Stop the Spread Ottawa study, a prospective cohort study investigating longitudinal antibody titers since October 2020 (25, 26).

Participants were excluded if they had been diagnosed with an IgG subclass deficiency without evidence of functional antibody deficiency and isolated IgA deficiency, had evidence of human immunodeficiency virus infection, had a positive COVID-19 infection by PCR or rapid antigen test (RAT) <12 weeks prior to vaccination, or were otherwise contraindicated from receiving COVID-19 vaccines.

All participants received vaccines in accordance with their local provincial vaccination guidance, which had varied recommended intervals between vaccine doses. In general, dose 2 was administered 4–12 weeks after dose 1, and dose 3 was administered 6–24 weeks after dose 2.

The study protocol was initially developed in early 2021 when two COVID-19 vaccine doses were recommended as a primary series. Thus, we collected blood at 0–28 days prior to administration of dose 1 (D1) as well as 0–14 days before dose 2 (D2), 4 ± 1 weeks after (D2 + 4wk), and 24 ± 4 weeks after (D2 + 24wk) dose 2. A three-dose primary series was recommended for IEIs as of September 2021 in Canada (27). The study protocol was therefore amended to include data collection before and after dose 3 for IEIs and controls. To minimize the frequency of blood draws and maintain participant retention, we omitted the study visit at D2 + 4wk but added 0–14 days before dose 3 (D3), as well as 4 ± 1 (D3 + 4wk) and 24 ± 4 (D3 + 24wk) weeks after dose 3. Participants who did not want to proceed with dose 3 despite recommendations were followed up until 48 ± 4 weeks after dose 2 (D2 + 48wk) for blood collection.

Plasma and peripheral blood mononuclear cell (PBMC) samples were isolated from participants’ blood and stored at each participating center prior to shipment to the University of Ottawa Serology and Diagnostics High Throughput Facility (plasma) for antibody assays and Dalhousie University (PBMC) for T-cell IFN-γ enzyme-linked immunosorbent spot (ELISpot) assay.

Plasma samples were assessed for levels of IgG, IgA, and IgM antibodies against severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) trimeric spike (S), receptor binding domain of spike (RBD), and nucleoprotein (N) by chemiluminescent ELISA (28). Antibody neutralization was assessed using surrogate neutralization ELISA on the full trimeric spike (wild-type and B.1.1.529-Omicron BA.1 subvariant, NRC Metrology and laboratory of Dr. Yves Durocher) (26). IgG serological titers were reported in WHO international units as binding antibody units (BAU)/mL. IgA and IgM titers were reported as concentrations compared to laboratory-calibrated standard curves (ng/μL). Neutralization data were reported as the inhibitory dilution at 50% inhibition (ID50) calculated from a four-parameter logarithmic regression applied to a 5-point serial dilution of samples.

Human IFN-γ ELISPOT set (BD Biosciences, Mississauga, ON, Canada) was used as per manufacturer instructions to determine the proportion of SARS-CoV-2-specific T cells. PBMCs were assayed in duplicate at 2 × 10⁵/well after 18-hour stimulation with 2.5 µg/mL phytohemagglutinin-L (PHA; Life Technologies, Burlington, ON, Canada) as the positive control, an equivalent concentration of dimethyl sulfoxide (DMSO) (Sigma-Aldrich, St. Louis, MO, USA) as the negative control (0.32%), or 1 µg/mL peptide pools, consisting mainly of 15 mer sequences with 11 amino acids overlapping, derived from the complete protein-coding sequences of S-protein from wild-type, B.1.617.2-Delta (S-Delta) and B.1.1.529-Omicron (S-Omicron) variants, wild-type N, or control peptide pool. Spots were enumerated using a CTL-ImmunoSpot^® S6 Micro Analyzer (CTL, Cleveland, OH, USA), and CMI response was reported as the proportion of SARS-CoV-2 specific T cells (IFN-γ⁺ T cells/10⁶ PBMCs).

Vaccine safety questionnaires were completed by the participants 7 and 28 days after each dose and 24 weeks after their last dose. Details on medically attended adverse events (MAAEs) including onset, symptoms, type of medical attention sought, treatments, severity (29), and evaluation of vaccine relatedness were collected. Additionally, information on historical COVID-19 infections was collected at the baseline visit. Breakthrough COVID-19 infections confirmed by PCR or RAT were noted at each follow-up.

Baseline demographics were summarized using descriptive analyses. Age, body mass index (BMI), and baseline laboratory parameters were summarized using means and standard deviations. Intervals in days between vaccine doses were reported as median with interquartile ranges (IQRs). The remainder of the demographic data were expressed as percentages. Primary analyses were conducted by comparing HI and CMI responses of IEIs with those of controls at each timepoint and 4 weeks after completing the primary series. Secondary analyses included the comparison of CMI and HI responses based on the type of IEIs. HI responses were stratified and analyzed by use of immunoglobulin replacement therapy (IGRT) and previous COVID-19 infection confirmed by PCR or RAT. Hybrid immunity was defined as having positive anti-N IgG and/or previous SARS-CoV-2 infection. Confidence intervals of geometric mean values of serological titers and IFN-γ⁺ T cells/10⁶ PBMCs were calculated. The independent t-test was used to compare geometric means at each timepoint between IEIs and controls. The proportion (Fisher’s exact test) of COVID-19 infections post-vaccination was compared between participant groups and IEI subgroups. Geometric means of HI and CMI among IEI subgroups and controls were compared using one-way ANOVA. The Mann–Whitney U-test was used to compare medians between IEIs and controls. The Kruskal–Wallis test was used to compare medians between IEI subgroups and controls. All tests were two-sided with statistical significance set at α = 0.05. Analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA), and graphs were generated using Prism version 10.2 (GraphPad Software, LLC, Boston, MA, USA).

At D2 + 4wk, anti-RBD and anti-S IgG titers were significantly lower in participants with IEIs compared to controls {geometric mean (95% CI), 265.5 [95.31–739.7] vs. 4,221.4 [3,436.4–5,185.7] BAU/mL, p < 0.0001; 198.7 [66.43–594.3] vs. 3,830.1 [3,010.6–4,872.7] BAU/mL, p < 0.001, respectively}. However, at 6 months after dose 2 (D2 + 24wk), there was no longer a statistically significant difference in anti-RBD and anti-S IgG levels between IEIs and controls. At D3 + 4wk, anti-RBD and anti-S IgG titers were significantly lower in IEIs compared to controls (834.2 [499.9–1,392.1 vs. 6,961.8 [4,936.2–9,818.7] BAU/mL, p < 0.001; 669.8 [393.9–1,138.8] vs. 5,301.7 [3,402.6–8,260.7] BAU/mL, p < 0.0001, respectively). Anti-RBD and anti-S IgG titers at D3 + 24wk were also significantly lower in IEIs than in controls (852 [428–1,696.1] vs. 3,410.7 [2,664.9–4,365.2], p = 0.0004; 763.6 [402.7–1,447.6] vs. 3,273.7 [2,584.9–4,146], p < 0.0001, respectively).

Comparison of serological responses after completion of the primary series showed lower anti-RBD and anti-S IgG titers in IEIs than controls at 4 weeks but not at 24 weeks ( Figures 1A, B , Supplementary Material ). Anti-RBD and anti-S IgA titers at D2, D2 + 4wk, D2 + 24wk, D3 + 4wk, and D3 + 24wk were significantly lower in IEIs compared to controls ( Figures 1C, D ). Anti-RBD and anti-S IgM titers at D2, D2 + 4wk, D3 + 4wk, and D3 + 24wk were also lower in IEIs than in controls ( Figures 1E, F ).

SARS-CoV-2 infections before vaccine doses 2 and 3 and within 6 months of receiving dose 3 were reported in 15/37 (40.5%), 5/37 (13.5%), and 11/37 (29.7%) of controls and 1/149 (0.7%), 9/149 (6.0%), and 32/149 (21.5%) of IEIs. Anti-RBD and anti-S IgG titers were only higher in controls with hybrid immunity compared to controls without hybrid immunity at D2, D2 + 24wk, and D3 + 24wk, while no difference was found in IEIs with or without hybrid immunity at any timepoint ( Figures 2A, B ). Anti-N IgG titers of controls increased at an earlier timepoint, whereas the titers increased at a later timepoint in IEI participants, consistent with the report of SARS-CoV-2 infection ( Figure 2C ). Further, anti-N IgG titers were higher in both healthy and IEI participants with hybrid immunity than those without hybrid immunity ( Figure 2C ). Interestingly, the higher anti-N IgG titers at D3 + 24wk in IEIs were driven by IEI participants on IGRT (12.9 [9.1–18.28] vs. 3.76 [1.79–7.89], p = 0.0008) ( Figure 3 ).

To evaluate if measured serological responses in IEIs were impacted by IGRT, we analyzed serology data based on IGRT ( Figure 3 ). IGRT did not impact anti-RBD and anti-S IgG titers at any timepoint. This was confirmed by the analysis of anti-S IgG levels in IEI participants in the PAD subgroup (i.e., with predominant B-cell deficiency), as they were the most likely to have the poorest HI response to the vaccine ( Supplementary Material ). In fact, PAD participants who were on IGRT tended to have lower anti-S IgG (not statistically significant) and anti-S IgA than those not on IGRT.

At D2 + 4wk, D3 + 4wk, and D3 + 24wk, the median (IQR) ID50 against the ancestral strain was significantly lower in IEIs compared to controls (6.55 [1.00, 22.9] vs. 73.4 [48.6, 155.8], p < 0.0001; 20.34 [3.9, 71] vs. 99.4 [57.5, 150.6], p = 0.0032; 14.7 [1, 68.9] vs. 104.9 [48.6, 426.6], p = 0.0005, respectively) ( Figure 4 ). Similar results were seen against Omicron BA.1 variant, where IEIs had a significantly lower ID50 compared to controls at D2 + 4wk, D3 + 4wk, and D3 + 24wk (1 [0.5, 1.8] vs. 13.6 [6.6, 27.7], p < 0.0001; 2.1 [1, 10.3] vs. 24.3 [12.3, 27.6], p = 0.0021; 1 [1, 6.6] vs. 21.5 [4, 62.7], p = 0.0013, respectively). There were reduced ID50 titers against the Omicron BA.1 variant in both groups as compared to those against the ancestral strain.

The T-cell responses measured by S-specific IFN-γ⁺ T cells/10⁶ PBMCs at D2 + 4wk, D3 + 4wk, and D3 + 24wk were significantly lower in IEIs compared to controls (23.6 [95% CI, 8.8–63.1] vs. 163.0 [77.9–340.9], p = 0.002; 21.65 [13.50–34.70] vs. 109.00 [29.85–398.02], p = 0.03; 13.7 [6.8–27.4] vs. 55.9 [25.0–124.6], p = 0.02) ( Figure 5A ).

There was no significant difference in S-specific T-cell responses against S-Delta and B.1.1.529-S-Omicron responses between IEIs and controls at any timepoint except at D3 + 24wk, where there was a significantly lower response in IEIs compared to controls (31.5 [16.8–57.5] vs. 84.3 [39.3–180.8], p = 0.048) ( Figures 5B, C ).

IEIs had lower T-cell response than controls at 4 weeks but not at 6 months after completing their respective primary series (p < 0.0001 for S-ancestral, p = 0.007 for S-Delta, and p = 0.03 for B.1.1.529-S-Omicron). Similar to the serological response, T-cell response to natural infection was more evident in controls than in IEI at D2 + 4wk and D3 + 4wk ( Figure 5D ), corresponding with their history of prior SARS-CoV-2 infections.

Participants in the PAD subgroup mounted weaker anti-RBD, anti-S, and anti-N responses at D2, D2 + 4wk, D3 + 4wk, and D3 + 24wk compared to controls ( Figure 6 ). Similar differences were also seen in the CID subgroup ( Figure 6 ). Although there was no statistically significant difference in anti-S IgG between the PAD and CID subgroups, the PAD subgroup had lower median anti-S IgG than the CID and III subgroups at D3–4 (533.1 [295.5, 2,512.0] vs. 3,105.7 [52.0, 7,338.8] vs. 5,866.3 [4,377.1, 10,777.2], respectively) ( Figures 6 , 7 ). Conversely, S-specific T-cell response was higher although not statistically significant in the PAD subgroup than in the CID subgroup at D3 + 4wk (46.3 [12.5,85] vs. 5 [2.5, 50], p = 0.07). HI comparisons were performed; however, CMI comparisons with the III subgroup were not performed due to a small sample size ( Figure 7 ).

The degree of correlation of markers of HI and CMI against S-protein differed between healthy and IEI participants ( Supplementary Material , Figure 8 ). In general, there was a strong correlation between anti-S IgG and ID50 (r = 0.83, p < 0.0001 for controls and 0.86, p < 0.0001 for IEI participants), a moderate correlation between serologies and cellular responses (r = 0.564, p < 0.0001 for healthy and 0.373, p < 0.0001 for IEI), but no correlation between ID50 and cellular responses (r = −0.013, p = 0.94 for controls and 0.154, p = 0.11 for IEI participants). At specific timepoints ( Figure 8 ), there was also a strong correlation between anti-S IgG and ID50 at D3 + 4wk (r = 0.750, p = 0.067 for controls and 0.88, p < 0.0001 for IEI participants) and D3 + 24wk (r = 0.867, p = 0.0005 for controls and 0.729, p < 0.0001 for IEI participants). There was weak to no correlation between serologies and cellular responses at D3 + 4wk (r = 0.143, p = 0.78 in controls and 0.149, p = 0.24 in IEI participants) and at D3 + 24wk (r = 0.112, p = 0.73 in controls and −0.023, p = 0.91 in IEI participants). Similarly, there was weak to no correlation between ID50 and cellular responses at D3 + 4wk (r = −0.013, p = 0.94 in controls and 0.153, p = 0.11 in IEI participants) and at D3 + 24wk (r = 0.136, p = 0.69 in controls and 0.007, p = 0.97 in IEI participants).

Among participants enrolled between D1 and D3 + 24wk, there were 21 MAAEs in 17 (11.4%) IEI participants and only one MAAE in controls ( Table 2 ). For IEI participants, over 90% of these events were considered moderate severity. Of all MAAEs, 80.95% were determined to be unrelated to vaccination, while the remaining events were identified as probably (4.8%) or possibly (9.5%) related to vaccination. The average onset of MAAEs in IEI participants was 88.3 days from a vaccination, all of which occurred between D3 and D3 + 24wk. The MAAE in the controls was a spontaneous abortion reported 2 weeks after the first vaccination at approximately 7 weeks of gestation and determined to be possibly related to vaccination.

Overall, including infection history from before and after study enrolment, we found that 24 (64.9%) of controls reported at least one SARS-CoV-2 infection with a total of 31 infections, and 15/31 occurred when the participant was vaccine-naive. However, 40 (26.9%) of IEI participants had at least one SARS-CoV-2 infection. The total number of reported SARS-CoV-2 infections was 42, and only one occurred before any vaccination ( Supplementary Material ). Nine (21.43%) infections were reported between doses 2 and 3 for IEIs, while 5 (16.13%) were reported for healthy controls. Most infections reported by IEI participants occurred after dose 3 (32, 76.19%) and after January 2022 (35, 82.33%).

During the study, 8.7% of IEI participants and 2.7% of controls had medically attended COVID-19 infections (p = 0.31) ( Table 2 ). No severe infections were reported during the study, but there was one severe infection requiring 8 days of hospitalization reported from one IEI patient prior to study enrolment.

Participants with breakthrough infections had lower median anti-S IgG titers, ID50 titers, and S-specific T-cell responses prior to infection than participants who did not have breakthrough infections [301.7 (25.5, 1,445.0) vs. 2,367 (656.7, 6,507.0) BAU/mL, p < 0.0001; 14.0 (1, 56.6) vs. 34.7 (8.5, 98.3), p = 0.0.4; 16.25 (2.5, 56.6) vs. 35 (5.0, 75.0), p = 0.03, respectively] ( Figure 9 ).