Study population

Two cohorts of adult participants were enrolled. The first cohort (general population; n = 838) consisted of community volunteers who completed the whole primary vaccination series and were followed longitudinally, including after booster dose administration. The second cohort (health care workers; n = 462) comprised employees of the National Medical Institute who received a booster dose at the study entry; for this cohort, blood sampling and immunological analyses were performed starting from the time of booster vaccination onward, as no preprimary series samples were available.

Vaccination protocol

Vaccination followed national recommendations. Mostly, 3 COVID‑19 vaccines authorized in Poland were used: BNT162b2 (Pfizer‑BioNTech) and mRNA‑1273 (Moderna), both nucleoside‑modified mRNA vaccines encoding the SARS‑CoV‑2 S protein, administered as 2 intramuscular doses 21 and 28 days apart, respectively; and Ad26.COV2‑S (Johnson & Johnson [J&J]), a nonreplicating adenoviral vector vaccine given as a single intramuscular dose. Booster doses were administered using mRNA vaccines, Ad26.COV2‑S, as well as ChAdO × 1 nCoV‑19 (AstraZeneca), a nonreplicating chimpanzee adenoviral vector vaccine. Irrespective of the initial type of vaccine, most patients received the Pfizer‑BioNTech vaccine as a booster.

Sample collection and immunological assessments

Blood samples were collected at predefined time points. In the general population cohort, sampling occurred at baseline (predose 1; sampling 1), before dose 2 (where applicable; sampling 2), and at 6, 12, and 18 months after completion of the primary schedule (samplings 3–5). In the health care worker cohort, sampling began immediately prior to the booster dose (sampling 1) and continued at 3, 6, and 12 months postbooster (samplings 2–4). Humoral immunity was assessed by measuring the titers of SARS‑CoV‑2 IgG and IgM neutralizing antibodies. SARS‑CoV‑2–specific IgG and IgM antibody concentrations were measured using a quantitative chemiluminescent immunoassay performed on serum samples according to the manufacturer’s protocol. Results were expressed in binding antibody units (BAU/ml). Cellular immunity was evaluated in a predefined subgroup from both cohorts. Peripheral blood mononuclear cells were isolated and analyzed by flow cytometry to quantify major immune cell subsets, including T (CD3+) and B lymphocytes, natural killer (NK) cells, dendritic cells (DCs; classical and plasmacytoid), monocytes, and macrophages. B‑cell and T‑cell levels served as indicators of humoral memory and cellular activation, respectively, while innate subsets were assessed to characterize broader vaccine‑related immune effects.

Study end points

The primary end point was the change in SARS‑CoV‑2 IgG antibody levels following vaccination across the different vaccine types. Secondary end points included the kinetics of IgM antibodies and the evaluation of cellular immune responses at predefined follow‑up time points.

Statistical analysis

All statistical analyses were performed using R software, version 4.2.2 (R Foundation for Statistical Computing, Vienna, Austria). Data distribution was assessed with the Shapiro–Wilk test. Given the typically right‑skewed distribution of immunological variables and the presence of unequal variances between the groups, normality and homoscedasticity assumptions were carefully evaluated prior to comparative analyses. Continuous variables are presented as mean (SD) or median (interquartile range), and categorical variables as counts and percentages. Humoral responses (IgG and IgM) between vaccine types at each time point were compared using the Welch 1‑way analysis of variance with Welch t tests for post hoc pairwise comparisons and Holm–Bonferroni correction. Effect size was expressed using the Hedges g statistic as an unbiased standardized measure of mean differences, appropriate for groups of unequal size. Changes over time within vaccine groups were presented descriptively due to the observational design and unequal sample sizes. Cellular immunity analyses were limited to descriptive statistics, as varying sample sizes and the absence of repeated measures for all participants precluded formal longitudinal testing. A 2‑tailed P value below 0.05 was considered significant.

Study population characteristics

The primary cohort included 838 individuals from the general population (mean age, 38.8 [13.7] years; 51.8% women [n = 434]). Vaccination was performed with the Pfizer‑BioNTech vaccine in 59.7% of the cohort (n = 500), Moderna in 15.5% (n = 130), and J&J in 24.6% (n = 206). The second cohort comprised 462 health care professionals (mean age, 45.8 [12.6] years; 60.9% women [n = 282]). Most of them received the Pfizer‑BioNTech vaccine (87.5%; n = 405), with smaller proportions receiving Moderna (3.2%; n = 15), J&J (1.5%; n = 7), or AstraZeneca (7.8%; n = 36) vaccines.

Antibody responses after primary vaccination

In the general population cohort, 838 participants provided baseline samples for IgG and IgM level assessment. This cohort exhibited a strong humoral response after the primary vaccination series, with antibody levels waning over time and their partial restoration associated with booster administration (Table 1).

When stratified by vaccine type, differences in humoral responses were observed (Table 1). At baseline, IgG levels were low across the mRNA vaccine recipients, while the recipients of the J&J vaccine showed considerably lower values (Welch ANOVA, P <⁠0.001; Moderna vs J&J, P <⁠0.001; g = −0.51; Pfizer‑BioNTech vs J&J, P <⁠0.001; g = 0.22). After completion of the primary series (sampling 2), IgG titers diverged markedly (P <⁠0.001): the Moderna vaccine was associated with the highest levels (mean, 10 033 [9992] BAU/ml), exceeding the Pfizer‑BioNTech (mean, 6872 [8695] BAU/ml; P = 0.004; g = −0.35) and J&J vaccines (mean, 2805 [3124] BAU/ml; P <⁠0.001; g = −0.74). The Pfizer‑BioNTech vaccine also produced higher titers than the J&J vaccine (P = 0.004; g = 0.47). At the third sampling, which partly reflected booster effects, intervaccine differences were no longer significant (P = 0.18) despite similar trends (Pfizer‑BioNTech, 3877 BAU/ml; Moderna, 3625 BAU/ml; J&J, 2108 BAU/ml). At later time points (samplings 4–5), only data for the mRNA vaccines were available for comparison; IgG levels declined with no significant difference between Pfizer‑BioNTech and Moderna (P = 0.12; g = 0.35–0.42). No differences in IgM levels were reported (data not shown).

Booster response

Among the health care workers, the first sampling (n = 462) occurred approximately 9 months after primary vaccination, just before the booster dose. IgG levels rose markedly after boosting, then gradually waned, while IgM levels remained low and showed minimal fluctuations. This indicates residual immunity at 9 months, a strong booster response, and subsequent decline over time (Table 1).

Impact of the interval between primary vaccination and booster dose on the humoral immune response

A preliminary comparison examined whether the interval between the primary series and the booster influenced humoral responses. In the general population, mean IgG levels were 3766 BAU/ml at approximately 6 months, rose to 5313 BAU/ml postbooster, and declined to 3299 BAU/ml thereafter. In the health care workers, the first measurement at approximately 9 months after primary vaccination showed lower IgG values than in the general population (1340 BAU/ml), followed by an increase to 5585 BAU/ml postbooster, a peak of 7891 BAU/ml, and a subsequent decline to 6014 BAU/ml. These data suggest more pronounced waning with a longer interval before boosting in the health care workers, but also a more substantial booster‑induced rise, likely reflecting occupational exposure and a larger immunological gap prior to boosting.

Cellular immune response

Analysis of cellular immunity demonstrated modest and mostly transient changes across immune cell subsets following SARS‑CoV‑2 vaccination. In the general population cohort, the proportion of DCs remained stable after the first dose (4.5% at baseline vs 4.6% predose 2), decreased by approximately 30% at 5–6 months (3.3%), and partially recovered at 12 months (3.9%). Macrophages showed a gradual decline from 2.2% at baseline to 1.7% postdose 2 (reduction by about 30%) and 1% at 12 months (further decrease by about 20%), while the proportion of classical monocytes remained stable (28.3% at baseline vs 27.5% postdose 1 and 29.4% at 12 months). In contrast, T lymphocytes demonstrated a more dynamic pattern, with levels rising from 36.9% at baseline to 45.5% postdose 2 (increase by about 30%) and peaking at 6 months (45.5%, with a subsequent decline to 41% at 12 months). NK cell levels remained unchanged after primary vaccination (11.3% at baseline vs 11.5% postdose 2) but declined markedly at 12 months (4.5%), indicating limited long‑term NK involvement in vaccine‑induced immunity.

Among the health care workers, prior to booster administration, baseline values were comparable for DCs (4.2%), lower for macrophages (1.6%), and slightly higher for T cells (42.9%) relative to the general population (arm I). After the booster dose, the proportion of DCs initially declined (3.3%), then increased at 3–6 months (5.1%), before returning to a near‑baseline value at 12 months (3.8%), suggesting a distinct stimulation pattern in the individuals with higher occupational exposure. Macrophages decreased sharply after boosting (0.6%) and only partially recovered by 12 months (0.9%). T‑cell levels increased following the booster (46%), remained elevated at 6 months (46.2%), and gradually declined at 9–12 months (37.5%). The proportion of NK cells decreased from 9.3% at baseline to 5.2% at 3–9 months, with partial recovery at 12 months (6.5%). Classical monocytes remained stable (25.2% at baseline vs 24.8% postbooster and 27.8% at 12 months). Collectively, these results indicate that vaccine‑related cellular responses were most pronounced in T cells. In contrast, changes in innate cell subsets (DCs, NK cells, monocytes, and macrophages) were mild, transient, or absent over time.

Conclusions

In summary, this real‑world study from Poland shows that mRNA vaccines—particularly mRNA‑1273—induce higher initial humoral responses than Ad26.COV2.S, with marked antibody waning by 6–12 months and apparent restoration after boosting, including with heterologous schedules. Cellular changes were modest, supporting the concept that antibodies predominantly protect against infection, while cellular immunity underpins longer‑term protection from severe disease. These findings reinforce the value of booster vaccination and flexibility in vaccine platform selection.