The study included two distinct cohorts of SOT recipients; (i) a first large, multicenter, prospective observational cohort of SOT and IC patients who underwent three consecutive SARS-CoV-2 booster vaccinations and were monitored for distinct serological and cellular immune responses specific to the virus, and (ii) a second cohort of SOT group who, despite receiving three previous vaccine doses, did not have neutralizing SARS-CoV-2-specific IgG antibodies (NAb). These patients participated in an interventional RCT to assess the achievement of NAb and cellular immune memory after a fourth vaccine dose, comparing a standard- of-care immunosuppressive regimen based on CNI/MMF (MMF/mycophenolate sodium (MPS)) with a regimen switched to a CNI/mTOR -i-based immunosuppressive therapy ( Figure 1 ).

Peripheral blood mononuclear cells (PBMC) and serum samples were obtained at the following specific time points for the descriptive cohort: prevaccination (T0), after the first dose (T1: 28 days; interquartile range [IQR], 28–34), 2 months after second dose (T2: 89 days; IQR, 79.25–95.75), 5 months after second dose (T3: 189 days; IQR 182–196), and 1 month after the third dose (T4: 232 days; IQR, 220–244).

In the randomized controlled trial whole blood and serum samples were obtained at the following prespecified time-points: at the time of randomization (T0a) (being one month prior to the fourth booster vaccination); at the time of the fourth dose (in case of Group B patients, being 1 month after mTOR-i conversion) (T1a) and at 1 month after fourth dose (T2a).

Peripheral blood mononuclear cells (PBMCs) were isolated from the patient’s blood by Ficol density gradient centrifugation and subsequently frozen in liquid nitrogen until their use in functional analyses. Serum samples were isolated by centrifugation and stored at – 20°C.

All statistical analyses were performed using SPSS version 26 software and R package version 1.0.12, and graphs were generated using GraphPad Prism version 8.0 software (GraphPad Software, San Diego, CA, USA). Continuous variables were expressed as median [IQR] and categorical variables as number of total (n) and percentage (%). Comparison s between groups was performed using Pearson’s χ ² test for categorical data. Continuous measurements were compared among groups using the Kruskal–Wallis and Mann–Whitney U tests for nonnormally distributed data, and ANOVA and t-tests were used when data were normally distributed. p-values of < 0.05 were considered statistically significant. To assess longitudinal data, considering the missing values and the changes over time, a two/linear mixed model, considering study time points and transplant status as fixed effects, was used. A correlation analysis of repeated measurements (46) was performed. p-values were adjusted for multiple comparisons using the Benjamini–Hochberg method. Univariate logistic regression models were used to investigate the influence of clinical and immunological covariates by means of odds ratios (OR) with 95% confidence intervals.

The first observational cohort consisted of 148 SOT patients who were compared to 32 IC individuals regarding serological and cellular immune memory responses triggered after three consecutive doses of SARS-CoV-2 mRNA-based vaccine s ( Figure 1 ). Most patients were kidney (37%) and liver (28%) transplant recipients, receiving CNI-based immunosuppressive therapy combined with either MMF/MPS (n = 84) or an mTOR-i drug (n = 20). All patients received at least three booster doses. Among the 148 SOT patients, 110 (74.3%) received a fourth dose and 34 (23.7%) received a fifth booster dose. All patients were followed up for a median of 30 months (median, 892 days; IQR, 884–903) ( Table 1 ). While COVID-19 BTI occurred after the third dose in 27.5% of patients (29.9% of SOT and 16.1% of IC patients) and in 17.6% of those who received a fourth dose (19.1% SOT and 6.7% IC), up to 34.3% of SOT patients who received a fifth dose developed COVID-19. Severe and mild COVID-19 only occurred among SOT in 12 (8.1%) and 33 (22.3%) patients, respectively. Asymptomatic COVID-19 was reported in 42 patients (36 (24.3%) SOT and six (18.8%) IC).

As illustrated in Figure 2A , while IgG titers, as well as mBc and T-cell frequencies specific to SARS-CoV-2, significantly increased with the number of vaccine doses in both groups, SOT patients displayed significantly weaker immune memory responses compared to IC patients after each vaccine dose, except for Th1 cells producing IL-2 and Th2 cells producing IL-21 and IL-5 ( Supplementary Tables S1 , S2 ). After the second dose, both at T2 and T3, 41.9% (62/148) and 38.7% (55/142) of SOT patients developed NAb, whereas up to 96.8% (30/31) and 92.3% (24/26) of IC patients did, respectively. Notably, after the third dose (T4), 100% (15/15) of IC patients showed NAb, while only 67.7% (84/124) of SOT patients were NAb positive. There was generally a strong positive correlation between all distinct SARS-CoV-2-specific immune responses except for mBc and T-cell frequencies between T0 and T4 ( Supplementary Table S3 ).

Notably, within the 3-month time frame between the second and third vaccine doses (T2 and T3), no major changes were observed in all SARS-CoV-2-specific T memory responses, and the antibody titers progressively decayed. However, RBD-specific IgG-producing mBc significantly increased over time in both SOT and IC individuals ( Figure 2B ).

When both B- and T-cell immune compartments were assessed together, IC individuals showed earlier detection of memory responses across all compartments compared to SOT patients ( Figure 2C ). Specifically, after two vaccine doses (T3), most IC patients (27/31; 87.1%) displayed broad immune memory responses, whereas only 45.2% (66/146) of SOT patients did (p < 0.001). Similarly, after the third dose (T4), only 63.6% (75/118) of SOT patients showed complete detectable immune responses, while this protection was seen in all IC patients (p < 0.001). Importantly, at this last time point, up to 15.3% (18/118) of SOT patients did not show any immune response, 11.9% (14/118) showed responses only within the T-cell immune compartment, and 9.3% (11/118) showed responses only within the humoral compartment.

In all patient cohorts, a positive correlation between serum IgG titers and the frequencies of IgG-producing mBc was observed in both SOT and IC individuals at all time points of assessment ( Supplementary Figure S2 ). We next assessed the presence of NAb against two different viral strains, Wuhan (D614g) and Omicron (BA.5), to evaluate the cross-reactivity of both serum IgG and those produced by mBc in SOT and IC patients after four vaccine doses. As shown in Figure 3A , 88% (22/25) of patients with high IgG titers (> 143 BAU/mL) showed NAb in the serum against the Wuhan strain, whereas only 20% (3/15) of patients with low IgG titers (< 143 BAU/mL) did. Similar results were observed for the Omicron BA.5 variant, with 60% (15/25) of patients with high IgG titers showing NAb, compared to only 6.7% (1/15) of patients with low titers. Of note, the assessment of the neutralizing capacity of IgG antibodies produced by polyclonally expanded mBc revealed that 50% (13/26) and 31.3% (5/16) of patients showed mBc with NAb against the Wuhan and the Omicron BA.5 variants, respectively ( Figure 3B ). As shown in Figure 3C , there was a high correlation between serum and mBc NAb for both viral strains.

Distinct maintenance immunosuppressive (IS) regimens were associated with different humoral and cellular immune responses ( Figure 4A ). Patients on CNI monotherapy and those on a mTOR-i-based regimen displayed significantly higher immune responses after each vaccine dose, comparable to those of IC individuals. In contrast, patients on CNI/MMF and those few on MMF monotherapy showed the weakest responses, even after three vaccine doses ( Supplementary Table S4 ). When both B- and T-cell immune memory compartments were analyzed together, patients on either CNI monotherapy or CNI/mTOR-i immunosuppression developed a global immune response significantly faster than those on CNI/MMF and MMF monotherapy ( Figure 4B ; Supplementary Table S5 ).

Unsupervised hierarchical cluster analysis of all immune memory compartments prior to the third booster vaccination in all patients showed that the presence of prebooster cellular immune memory, especially RBD-specific IgG-producing mBc, predicted the successful development of NAb after the booster vaccination ( Figure 5 ; Supplementary Table S6 ). Similarly, when only seronegative patients were analyzed ( Figures 6A, B ), those who developed NAb after a third booster vaccination displayed significantly higher IFN-γ, IL-2, and mBc frequencies prior to vaccination. Among these, RBD-specific mBc frequencies showed the best predictive capacity for the subsequent development of NAb (AUC, 0.758) ( Figure 6C ). Moreover, the combination of SARS-CoV-2-specific IL-2-producing T cells and RBD-specific mBc provided the best prediction for subsequent NAb (AUC, 0.828) development.

Multivariate analysis revealed that the main factors favoring the development of NAb were the presence of RBD-specific mBc (OR, 19.313 [95% CI, 3.721–100.24]; p < 0.001), lower age (OR, 0.920 [95% CI, 0.863–0.980]; p = 0.010), and CNI trough levels (OR, 0.771 [95% CI, 0.586–1.013]; p = 0.062). Similarly, for the development of RBD-specific mBc, the time after transplant (OR, 1.007 [95% CI, 1.002–1.013]; p = 0.012) and the absence of MMF-based therapy—with either the combination of CNI/mTOR-i (OR, 11.092 [95% CI, 2.084–59.029]; p = 0.005) or CNI monotherapy (OR, 4.224 [95% CI, 1.482–12.040]; p = 0.007)—were independent correlates.

When we analyzed the association between memory immune responses and subsequent COVID-19 BTI during the study period, significantly lower cytokine-producing T-cell frequencies and a lack of RBD-specific mBc were observed among patients who developed severe COVID-19 compared to those with mild or asymptomatic infections ( Figure 6D ; Supplementary Table S7 ). Multivariate analysis, considering both clinical and immunological variables predicting severe COVID- 19, showed that the only factor favoring severe BTI was the absence of RBD-specific mBc (OR, 0.097 [95% CI, 0.018–0.533]; p = 0.007) ( Supplementary Table S8 ).

A total of 25 SOT patients who did not develop NAb despite receiving three consecutive vaccine doses were subsequently evaluated in a RCT. This trial compared standard-of-care CNI/MMF immunosuppression (group A) with a low-exposure CNI/mTOR-i-based regimen (group B) to assess the development of anti-SARS-CoV-2 protective immunity at 1 month after the fourth vaccine dose ( Supplementary Figure S1 ). The two groups did not differ in terms of main clinical, demographic, or immunological baseline characteristics ( Table 2 ).

Firstly, although none of the 12 CNI/mTOR-i patients developed NAb before the fourth vaccine booster, we observed that cellular immune responses had already appeared 1 month after conversion to CNI/mTOR-i and prior to vaccination (T1a) in some patients (RBD-specific mBc, two patients; IFN-γ- and IL-2-producing T cells, two and one patients, respectively) ( Supplementary Figure S4 ). Secondly, 1 month after vaccination (T2a), 20% more patients developed NAb in group B compared to group A (8/12 [66.7%] vs. 6/13 [46.1%], p = 0.306, respectively) ( Figure 7A ). Additionally, the proportion of patients with detectable SARS-CoV-2-specific IgG-, IFN- γ- and IL-2-producing B- and T-cell responses, respectively, was also numerically higher in group B than in group A at T2a (9/12 [75%] vs. 7/13 [53.8%], p = 0.271 for RBD-specific mBc; 7/12 [58.3%] vs. 2/13 [15.4%], p = 0.025 for IFN-γ; and 4/12 [33.3%] vs. 5/13 [38.5%], p = 0.790 for IL-2-producing T cells, respectively) ( Figure 7B ). Despite the absence of NAb at baseline (T0a), some patients in each group displayed cellular responses; RBD-specific mBc in 12/25 (48%), IFN-γ-producing T cells in 6/25 (24%), and IL-2-producing T cells in 7/25 (28%). Although no clinical, demographic, or immunological variables predicted NAb seroconversion, prevaccine RBD-specific mBc was a significant predictor (OR, 33 [95% CI, 2.909–374.31]; p = 0.005) ( Figure 7C ). Similarly, in the observational cohort, there was a high correlation between serum IgG titers and IgG-producing mBc both after vaccination and between prevaccine mBc and postvaccine IgG titers ( Supplementary Figure S5 ).

The proportion of patients receiving broader adaptive immune responses achieved after booster vaccination (NAb, RBD-specific mBc, and/or IFN-γ/IL-2-producing T cells) was higher in those switched to CNI/mTOR-i immunosuppression compared to those remaining on CNI/MMF (11/12 [91.7%] vs. 8/13 [61.5%], p = 0.078, respectively) ( Figure 7D ).