Longitudinal humoral and memory B-cell responses to repeated SARS-CoV-2 vaccination in long-term care facilities residents

Abstract

Background

Long-term care facility (LTCF) residents represent one of the populations most vulnerable to SARS-CoV-2 infection and have experienced repeated vaccination and natural viral exposure since the beginning of the COVID-19 pandemic. The long-term dynamics of humoral and cellular immunity in this population remain incompletely characterized. In this multicenter longitudinal study, SARS-CoV-2-specific antibody responses were monitored in LTCF residents over 12 months. A total of 388 residents from LTCF across five Italian regions were enrolled and stratified according to receipt of the SARS-CoV-2 XBB.1.5 mRNA booster during the 2023–2024 vaccination campaign. In a subgroup of residents, peripheral B-cell phenotypes and Spike-specific memory B cells were characterized by flow cytometry and compared with those of younger healthcare workers vaccinated with the same formulation.

Results

Anti-Spike IgG titers peaked in residents who received the XBB.1.5 booster and subsequently declined over time, consistent with contraction of vaccine-induced humoral responses. In contrast, individuals who did not receive the booster maintained lower but stable antibody levels. Anti-nucleocapsid seroconversion and clinically diagnosed intercurrent infections occurred at broadly comparable observed rates in boosted and non-boosted residents, suggesting a comparable incidence of SARS-CoV-2 infection during follow-up. Regarding B-cell-mediated immunity, LTCF residents exhibited age-associated remodeling of the B-cell compartment, with reduced total B-cell frequencies but preserved antigen-experienced memory populations. Despite declining circulating antibodies, Spike-specific memory B-cell frequencies remained stable. A late increase in anti-Spike titers coincided with rising anti-nucleocapsid seropositivity, suggesting reactivation of immune memory following natural viral exposure.

Conclusions

These findings indicate that repeated vaccination and natural exposure generate durable immunological memory in LTCF residents. In highly exposed populations, immune maintenance may increasingly rely on reactivation of memory responses rather than persistently high antibody titers.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12979-026-00581-9.

Background

Several years after the emergence of the COVID-19 pandemic and the implementation of global vaccination campaigns, the SARS-CoV-2 immune landscape has evolved toward a complex immunological scenario resulting from the interplay between vaccine-induced responses, breakthrough infections, and variant-specific immune imprinting, ultimately generating heterogeneous trajectories of immune protection across populations. While this layered immunity has substantially reduced the burden of severe disease at the population level, it has also introduced considerable variability in the magnitude, breadth, and durability of immune responses [1–5]. Within this evolving immunological landscape, older adults with frailty and multiple chronic conditions represent a distinct and particularly vulnerable subgroup. Aging is associated with remodeling of the immune system, collectively referred to as immunosenescence, which includes alterations in both innate and adaptive immune compartments [6, 7]. These age-related immune alterations may impair the capacity to mount and sustain protective responses to infections, such as SARS-CoV-2, and vaccinations [8, 9]. Long-term care facilities (LTCF) residents represent one of the most vulnerable segments of the older population, often characterized by advanced age, multimorbidity, functional impairment, frailty, and chronic inflammatory conditions. These factors contributed to the disproportionate impact of COVID-19 during the early phases of the pandemic, where older adults living in LTCFs experienced a disproportionate burden of COVID-19–related mortality [10–13]. Consequently, LTCF residents were prioritized during vaccination campaigns and have frequently received repeated booster doses. The long-term dynamics of SARS-CoV-2 immune responses in this population remain incompletely understood. Longitudinal studies have consistently documented a progressive decline in circulating antibody levels, including anti-Spike (S) IgG titers, following vaccination or natural infection, an occurrence that appears more pronounced in older individuals and those with frailty [14–16].

Circulating antibody levels provide only a partial representation of immune protection against SARS-CoV-2. Emerging evidence suggests that repeated antigenic exposures may lead to the establishment of a relatively stable immune set-point in previously exposed individuals, where additional booster doses show limited capacity to further expand the antigen-specific immune pool, a phenomenon that has been interpreted as the establishment of an immunological ceiling [17–19]. These observations highlight the need to investigate the immunological impact of repeated seasonal boosting in populations that have already accumulated multiple vaccine and infection-derived immune stimuli. In vaccinated populations, the detection of antibodies directed against the nucleocapsid (N) protein provides a useful serological marker of natural SARS-CoV-2 exposure. Monitoring both anti-S and anti-N antibody responses may therefore offer complementary insights into vaccine-induced immunity and incident infections during longitudinal follow-up.

Crucially, long-term protection against severe outcomes is thought to rely less on circulating antibody titers than on the persistence of antigen-specific immune memory, particularly within the memory B-cell (MBC) compartment. Unlike short-lived plasmablasts that contribute to transient peaks of circulating antibodies, MBCs can persist over time, remain detectable months to years after infection or vaccination, and rapidly differentiate into antibody-secreting cells upon antigen re-exposure [20–22]. These cells also retain the capacity to undergo further affinity maturation and adapt to viral variants, thus contributing to durable immune protection even when serum antibody levels decline. Nevertheless, in older adults, the impact of immunosenescence on the maintenance and functionality of this cellular memory compartment remains incompletely defined. While some studies suggest that SARS-CoV-2-specific MBCs can persist even in older individuals who have lost detectable neutralizing antibodies, the durability and functional quality of this memory in vulnerable populations remain uncertain [23].

In this context, a comprehensive characterization of long-term humoral immunity in LTCF residents is essential to better understand the persistence of immune protection in this population. Building on the experience of the GeroCovid Vax network, aimed at monitoring the response to SARS-CoV-2 vaccination in the Italian geriatric population [24], we designed the COVAC-3 study to longitudinally monitor the evolution of the immune landscape in LTCF residents across five Italian regions over 12 months during the late phase of the pandemic, when most individuals had already experienced multiple vaccine doses and/or natural infections. By integrating repeated measurements of anti-Spike and anti-Nucleocapsid antibodies with clinical, demographic, and vaccination data, we aimed to describe long-term antibody kinetics and to assess the impact of vaccination and natural infection on immune responses in this population. In addition, to further elucidate the magnitude and persistence of SARS-CoV-2-specific immune memory in this sample, longitudinal changes in peripheral B-cell subsets and Spike-specific memory responses were characterized in a subgroup of participants and compared with those of a reference group of younger healthy healthcare workers (HCWs). A complementary data-driven analysis of immune trajectories and phenotypic clustering in the same cohort is presented in a companion manuscript.

Methods

Study design

This study was conducted within the framework of the COVAC-3 project, a prospective multicenter longitudinal study designed to monitor SARS-CoV-2 immunity in long-term care facility residents across Italy. Thirteen LTCFs located across five Italian regions (Calabria, Friuli-Venezia Giulia, Lazio, Puglia, and Veneto) participated in the study. Enrollment occurred between 2023 and 2024, and participants were followed for a period of 12 months with repeated clinical assessments and immunological measurements. Residents aged ≥ 65 years living in the participating facilities were considered eligible. Inclusion criteria comprised permanent residence in the facility, provision of informed consent (directly or via a legal representative), and availability for scheduled follow-up visits. Individuals were excluded if they presented with an acute medical condition at enrollment, had an expected life expectancy of less than six months, had a documented active SARS-CoV-2 infection at baseline, or had contraindications to venipuncture. Enrollment occurred between November 2023 and March 2024. Follow-up visits were conducted at 3 months (T1), 6 months (T2), and 12 months (T3) after baseline. During the observation period, only a proportion of residents received exclusively the updated seasonal COVID-19 booster targeting the SARS-CoV-2 XBB.1.5 variant (Comirnaty XBB.1.5, Pfizer-BioNTech). Vaccination occurred either before enrollment or at the time of the baseline study visit. A schematic overview of the study design, including the full LTCF cohort, the B-cell phenotyping subgroups, with sampling timepoints and assays performed at each visit, is provided in Fig. 1.

Fig. 1.

Schematic flowchart of the study design. Left column: the full long-term care facility (LTCF) cohort, comprising 388 residents from 13 facilities across five Italian regions, stratified according to receipt of the XBB.1.5 mRNA booster: boosted, N = 191; non-boosted, N = 197. Longitudinal serum samples were collected for anti-Spike and anti-Nucleocapsid IgG assessment at enrollment (baseline, T0), 3 months (T1), 6 months (T2), and 12 months (T3) from enrollment. Center column: the B-cell phenotyping subgroup of 25 LTCF residents from a single facility, RSA Salus in Rome. T'0 corresponds to the time of XBB.1.5 booster administration in the subgroup, with no associated blood sampling; longitudinal PBMC and serum sampling began at T'1 (5 months post-booster). Longitudinal peripheral blood mononuclear cells (PBMCs) and serum samples were collected at 5, 8, 11, and 17 months after XBB.1.5 vaccination (T’1 to T’4). Right column: the reference cohort of 33 healthcare workers from Bambino Gesù Children’s Hospital in Rome who received the XBB.1.5 formulation. PBMCs were collected at pre-vaccination baseline (T0), 1 month (T’’1), and 12 months (T’’2) after vaccination. *Antigen-specific B cells refers to the comprehensive longitudinal evaluation of total B-cell phenotypes, which specifically includes the characterization of anti-Spike-specific memory B cells

Clinical and epidemiological data collection

Baseline demographic and clinical information was recorded for all participants, including age, sex, chronic conditions, functional status, previous SARS-CoV-2 infection, and vaccination history. During follow-up, additional data were collected on incident SARS-CoV-2 infections, hospital admissions, mortality, and newly diagnosed medical conditions.

Data from all participating centers were entered into the REDCap (Research Electronic Data Capture) secure electronic data capture system hosted by the Italian Society of Gerontology and Geriatrics (SIGG), with standardized data entry procedures performed by trained study personnel.

Blood sampling and serological assays

Peripheral blood samples for serological analysis were obtained at four predefined time points: baseline assessment at study enrollment (T0), 3 months (T1), 6 months (T2), and 12 months (T3) (Fig. 1). Serum was separated by centrifugation and stored at − 80 °C until laboratory analysis. Quantitative IgG antibodies directed against the SARS-CoV-2 Spike (S) and Nucleocapsid (N) proteins were measured at the central laboratory of the Department of Infectious Diseases, Istituto Superiore di Sanità.

Since the seasonal booster targeted the SARS-CoV-2 XBB.1.5 variant, we validated the correlation between an XBB.1.5-specific ELISA (ACRO Biosystems, Newark, DE, USA) and the LIAISON^® TrimericS IgG immunoassay (DiaSorin, Saluggia, Italy). The correlation between the XBB.1.5-specific ELISA and the LIAISON TrimericS assay was assessed in an preliminary validation experiment assessing 84 sera collected at all the T0, T1 and T2 timepoints and spanning a broad range of Anti-trimeric S concentrations (range: 610-22500 BAU/ml). Spearman rank correlation was performed on samples collected across all available timepoints, yielding r = 0.91 (p < 0.001). Based on this validation and the availability of standardized results in binding antibody units (BAU/ml), the Trimeric S assay was selected for the longitudinal analysis of the study cohort. Measurements were performed on the automated LIAISON^® XL analyzer and expressed as binding antibody units per milliliter (BAU/ml). The analytical measurement range extended to 2080 BAU/ml. Samples exceeding the upper limit of detection were automatically diluted (1:20) and re-analyzed to quantify values within the assay range. No manual titer derivation from OD values was performed for this assay.

IgG antibodies against the SARS-CoV-2 nucleocapsid (N) protein were measured using the Anti-SARS-CoV-2 NCP ELISA (Euroimmun, Lübeck, Germany). Anti-N IgG results were calculated as the ratio of the optical density (OD) of the patient sample to the calibrator OD. Specifically, results are interpreted as negative for ratios < 1, and positive for ratios ≥ 1. This methodological choice aims to minimize the impact of borderline results on the overall model variance. The threshold of 1.0 was selected as the midpoint of the manufacturer’s indeterminate zone, providing a conservative balance between sensitivity and specificity. Seroconversion of anti-N IgG between two consecutive study visits was considered indicative of incident SARS-CoV-2 exposure during the follow-up period.

PBMC isolation and B-cell phenotype

To compare between the immune dynamics of aged LTCF residents and the more plastic immune response of a younger population with a similar vaccination history, B-cell analyses were conducted on residents recruited from a single participating LTCF (RSA Salus, Rome, Italy) and a cohort of healthy, younger healthcare workers (HCWs) from the Bambino Gesù Children’s Hospital (Rome, Italy), the latter serving as a reference for B-cell phenotypic comparisons (Fig. 1). Inclusion criteria were receipt of the XBB.1.5 vaccine formulation, availability for PBMC collection, absence of known immunological disorders or immunomodulatory treatment, no active SARS-CoV-2 infection at baseline, and no contraindications to venipuncture.

LTCF cohort was sampled 5 months (T’1 n = 23), 8 months (T’2 n = 23), 11 months (T’3 n = 15) and 17 months (T’4 n = 10) after the sixth dose). Blood sampling in HCWs was performed at three timepoints: immediately before vaccination (T’’0, pre-dose baseline), one month after vaccination (T’’1), and twelve months after vaccination (T’’2) (Fig. 1 and Supplementary Fig. 1).

EDTA peripheral blood mononuclear cells (PBMCs) were isolated by density gradient centrifugation using Ficoll Paque™ Plus 206 (Amersham PharmaciaBiotech). Cells were immediately frozen and stored in liquid nitrogen until use. The freezing medium consisted of 90% Foetal Bovine Serum (FBS) and 10% DMSO. Flow cytometry analysis was performed on cryopreserved PBMCs. Cells were stained with the appropriate combination of fluorochrome-conjugated antibodies (Supplementary Table 1) to identify B-cell subsets according to standard techniques and were acquired on a Symphony A5 (BD Biosciences). Plasmablasts were defined as CD19⁺CD24⁻CD38⁺⁺ (Supplementary Fig. 2). After exclusion of plasmablasts, naïve B cells were identified as CD19⁺CD24⁺CD27⁻ (Supplementary Fig. 2), while total memory B cells (MBCs) were defined as CD19⁺CD27⁺. MBCs were further subdivided into resting MBCs (CD19⁺CD24⁺CD27⁺) and activated B cells CD19⁺CD24^−/+CD27⁺ (Supplementary Fig. 2).

Detection of antigen-specific B cells

SARS-CoV-2-specific B cells were identified using biotinylated wild-type spike protein (R&D Systems) and XBB1.5 Spike protein (Acrobiosystems) (Supplementary Fig. 2 and Supplementary Table 1). Proteins were individually multimerized with fluorescently labelled streptavidin as previously described [22, 25]. To increase the specificity of detection, WT Spike-specific MBCs and activated MBCs were identified as double-positive for the WT Spike probe labelled with PE and BUV395. Within the WT Spike-positive B-cell population, we then evaluated the frequency of cells also binding the XBB.1.5 Spike probe, here defined as WT⁺Omicron⁺ cross-reactive B cells. This gating strategy was designed to assess XBB.1.5 cross-reactivity within the WT Spike-specific B-cell compartment. XBB.1.5-specific B cells lacking detectable binding to ancestral WT Spike were below the detection threshold and for this reason were excluded from the analysis. Stained PBMC samples were acquired on a Symphony A5 (BD Biosciences). A minimum of 2 × 10⁶ cells per sample were acquired and analysed using Flow-Jo10.10.1 (BD Bioscience). Antigen-specific B cell phenotypic analysis was performed only in samples with at least 10 events detected within the respective antigen-specific gate. The antigen-probe staining strategy used in this study was based on procedures previously validated by our group where we validated the use of a decoy probe to exclude relevant non-specific probe binding [26].

Statistical analysis

Antibody titers were log-transformed to normalize the distribution. Longitudinal changes were assessed using the Wilcoxon signed-rank test, while differences between groups (sex and vaccination status) were evaluated via Mann-Whitney U or Kruskal-Wallis tests. To explore potential clinical and demographic influences on humoral immune profiles within the full LTCF cohort, formal non-parametric analyses were conducted. Differences in anti-Spike IgG kinetics between male and female residents were assessed using the Mann-Whitney U test. The intra-cohort impact of advanced age was evaluated via Spearman’s rank correlation between continuous age and circulating antibody levels. Lastly, residents were stratified by comorbidity burden into two groups (0–4 vs. >4 chronic conditions based on the cohort distribution in Table 1), and their longitudinal antibody trajectories were compared using the Mann-Whitney U test. For multiple comparisons, p-values were adjusted using the Bonferroni correction. Statistical significance was set at α = 0.05. Statistical analyses were performed using Stata (v18).

Table 1.

Demographic and clinical characteristics of the study population

Variable	Total Population (N = 388)	Vaccinated (N = 191)	Non vaccinated (N = 197)	p-value
Age, years (Mean ± SD)	84.20 ± 8.70	84.60 (7.96)	83.80 (9.36)	0.39
Female, n (%)	274 (70.61%)	134 (70.16%)	140 (71.07%)	0.75
Male, n (%)	114 (29.38%)	57 (29.84%)	57 (28.93%)
Vaccine doses received (Mean ± SD)	4.4 ± 0.9	4.8 ± 0.8	3.9 ± 0.8	< 0.001
Prior SARS-CoV-2 infection, n (%)	260 (67.01%)	123 (64.39%)	137 (69.54%)	0.281
Number of chronic conditions*	Total sample (N = 336)	Vaccinated (N = 184)	Non vaccinated (N = 152)
0–4	224 (66.67%)	106 (57.61%)	118 (77.63%)	< 0.001
>4	112 (33.33%)	78 (42.39%)	34 (22.36%)	< 0.001

Data are presented as mean ± standard deviation (SD) for continuous variables and as absolute numbers (percentages) for categorical variables. The study cohort is stratified by XBB.1.5 vaccination status (Vaccinated vs. Non-vaccinated). P-values indicating statistically significant differences (p < 0.05) are highlighted in boldface.. *The analysis of chronic conditions was conducted on a subgroup of 336 participants for whom complete clinical records were available

Results

Study sample

A total of 388 residents from LTCF across five Italian regions were enrolled in the study (Fig. 1). Demographic and clinical characteristics of the study population are shown in Table 1. The mean age of participants was 84.2 years, with 70.4% being female and 74% over the age of 80. Overall, 191 participants (49.2%) received a booster dose during the 2023/2024 vaccination campaign. More than half of participants (67%) showed serological evidence of prior SARS-CoV-2 infection at baseline, defined by the positivity of anti-N antibodies. The cohort presented a high burden of comorbidities, with an average of 4.0 chronic conditions per individual (range 0–12), with one-third of the cohort (33%) presenting more than 4 chronic conditions. A substantial proportion of enrolled residents displayed a high degree of vulnerability. The most prevalent chronic conditions were arterial hypertension (58.5%), dementia (51.0%), and ischemic heart disease (35.0%). Functional independence was limited, with 29.8% of residents being highly dependent or bedbound (Supplementary Table 2). Follow-up assessments were conducted at 3, 6, and 12 months post-baseline (T1, T2, T3), with high retention rates: 352 participants completed T1, 314 completed T2, and 249 reached T3 (Fig. 1). During the follow-up period, 83 residents (21%) died, while 56 participants (14%) withdrew from the study. Twelve residents (3.1%) experienced new SARS-CoV-2 infections, of whom three (25%) were asymptomatic, and the remainder presented with mild-to-moderate symptoms; oxygen supplementation (low-flow) was necessary for only three participants. Baseline characteristics of Completers and Non-Completers are reported in Supplementary Table 3. No significant differences were observed in baseline anti-Spike IgG titers or anti-N serostatus between groups; number of prior vaccine doses and three individual comorbidities (ischemic heart disease, obesity, malnutrition) differed significantly, though none were associated with baseline immunological parameters.

Longitudinal SARS-CoV-2 anti-S IgG antibody titers

As detailed above, preliminary validation experiments showed a strong correlation between an XBB.1.5-specific ELISA and the Trimeric S CLIA assay (Spearman rank correlation: ρ = 0.91 p < 0.001). Consequently, the latter was used for all longitudinal analyses. Anti-Trimeric S IgG titers exhibited a progressive decline over time across the entire cohort; compared with baseline (T0), IgG levels were significantly different at T1, T2, and T3 (p < 0.01). (Table 2). No significant associations were found between longitudinal anti-Spike IgG titers and sex, age, or comorbidity burden at any study timepoint (Supplementary Table 4), suggesting that the observed antibody kinetics were not substantially driven by the main clinical variables assessed. When the study sample was stratified by XBB.1.5 vaccination status, three subgroups were identified: residents vaccinated within one month before enrollment (≤ 1-month); those vaccinated more than one month before enrollment (> 1-month); and non-vaccinated residents (Fig. 2; Table 2). Differences in Anti-Trimeric S IgG titers were observed across groups at all time points. Further post-hoc pairwise analysis with Bonferroni correction revealed specific group dynamics. At T0, residents vaccinated > 1 month before enrollment had significantly higher titers compared to non-vaccinated residents (p < 0.001), while no significant difference was observed relative to ≤ 1-month residents (p = 0.803). At T1, ≤ 1-month vaccinated individuals reached their study peak, exhibiting significantly higher titers compared with the other two groups (p < 0.001 for both). At this time point, the difference between > 1-month residents and non-vaccinated residents did not reach significance. By T2, both the vaccinated groups exhibited a significant difference compared to non-vaccinated residents (> 1-month p = 0.0042; ≤ 1-month p = 0.0051). Notably, at T2, no significant differences were observed between the two vaccinated subgroups (p = 0.5481), indicating a convergence in antibody levels. At T3, no pairwise comparisons remained significant. However, the group vaccinated within the previous month included only four individuals, limiting statistical power and the interpretability of comparisons at this time point.

Table 2.

Longitudinal trajectories of anti-SARS-CoV-2 S-IgG titers

	Anti-S IgG in BAU/mL GMT (geometric 95% CI)
	T0			T1			T2			T3
	GMT	95% CI	p value	GMT	95% CI	p value	GMT	95% CI	p value	GMT	95% CI	p value
Whole sample	5233.7	4550.7-6019.4	(ref)	5334.9	4708.1-6045.1	< 0.01	3978.0	3494.7–4528.0	< 0.01	3989.5	3496.4-4552.1	< 0.01
Sex
Female	5434.1	4602.8-6415.4	0.47	5678.3	4880.8-6606.2	0.051	3988.4	3403.3-4674.1	0.79	3854.6	3308.1-4491.3	0.49
Male	5016.0	3914.1- 6428.1		4499.1	3623.7-5585.9		3950.3	3167.5-4926.6		4343.7	3336.6-5654.9
Vaccination Status*
Vaccinated ≤ 1 month	6098.1	4048.5–9188.0	0.0001	15456.3	10206.4- 23406.6	0.0001	6964.2	4540.6-10681.5	0.0003	1833.0	574.0- 5854.1	0.0278
Vaccinated > 1 month	7584.5	6187.7-9296.5		5598.4	4624.3- 6777.7		4756.7	3899.5-5802.2		4715.7	3890.3- 5716.3
Non vaccinated 2023/2024	3770.9	3077.4-4620.6		4331.2	3659.8-5125.8		3182.9	2656.5- 3813.7		3558	2962.2- 4273.6

Results are expressed as Geometric Mean Titers (GMT) with 95% Confidence Intervals (95% CI) in Binding Antibody Units (BAU/mL). P-values indicate significant differences across groups at each time point (T0: enrollment; T1: 3 months; T2: 6 months; T3: 12 months). P-values indicating statistically significant differences (p < 0.05) are highlighted in boldface

*The booster vaccine administered was the XBB.1.5-adapted mRNA formulation (Comirnaty XBB.1.5, Pfizer-BioNTech). Vaccination status is defined based on the interval between the last XBB.1.5 dose and study enrollment

Fig. 2.

Longitudinal anti-Spike IgG kinetics in LTCF residents stratified by XBB.1.5 booster vaccination status. Serum anti-Spike IgG titers (log BAU/mL) measured at baseline (T0), 3 months (T1), 6 months (T2), and 12 months (T3) in LTCF residents stratified by vaccination status: recently boosted (Vaccinated ≤1 month, dark blue), previously boosted (Vaccinated >1 month, medium blue), and non-vaccinated 2023/2024 (light blue). Individual data points are shown for each group. Boxplots display the median and interquartile range (IQR). Individual observations are overlaid as jittered points. Pairwise longitudinal comparisons within each vaccination group were performed using the Wilcoxon signed-rank test for paired samples. P-values were adjusted for multiple testing using the Bonferroni correction. Statistical comparisons between groups at each timepoint are indicated (*p<0.05, **p<0.01, ***p<0.001, ****p<0.0001)

SARS-CoV-2 anti-N IgG seroprevalence and incident infections

Analysis of anti-N antibodies showed that a large proportion of participants was already seropositive at baseline (69%), and this prevalence remained relatively stable over time. Anti-N IgG seroconversion was used as a proxy for SARS-CoV-2 exposure. A longitudinal seroconversion analysis was performed, restricted to participants who were anti-N seronegative at baseline (N = 128). The proportion of residents who were Anti-N negative at baseline was similar between vaccinated and non-vaccinated groups. Among these participants, 19.1% of residents who received the most recent booster and 28.3% of those who did not, seroconverted during follow-up, with no statistically significant difference between groups (Table 3). Similarly, no significant differences in intercurrent infection rates were observed between residents who received the seasonal XBB.1.5 booster and those who did not (Table 3).

Table 3.

Anti-N prevalence and intercurrent SARS-CoV-2 infections during the study period

Cohort characteristics	T0	T1	T2	T3	p-value
Overall Anti-N positivity	69% (268/388)	70% (245/352)	68% (214/314)	63% (158/250)	0.428

	Vaccinated (N=191)		Non-vaccinated (N=197)		p-value
Baseline Anti-N negative(N=128)	68 (35.6%)		60 (30.4%)		0.310
Seroconverted	13 (19.1%)		17 (28.3%)		0.296
Intercurrent infections	6 (3.1%)		6 (3.0%)		0.957

Data are presented as absolute numbers and percentages. Anti-N positivity indicates previous or intercurrent SARS-CoV-2 exposure. "Seroconverted" refers to participants who were Anti-N negative at baseline and became positive during follow-up. "Intercurrent infections" denote clinically documented episodes of COVID-19. P-values indicate statistical significance for longitudinal changes across timepoints (top row) and for cross-sectional comparisons between vaccinated and non-vaccinated subgroups (bottom rows, Fisher’s exact test)

Age-associated B-Cell remodeling and durable SARS-CoV-2 memory responses

To elucidate the cellular mechanisms underlying the long-term persistence of SARS-CoV-2 immunity in older adults, we performed a longitudinal B-cell phenotyping analysis in a subgroup of LTCF residents and compared the results with those obtained from a reference group of younger HCWs (Supplementary Table 5). The LTCF group consisted of 25 residents (mean age, 83 years, range 69–91) who had received the XBB.1.5 mRNA vaccine as their sixth dose 5 months before enrollment. For this analysis, sampling intervals in LTCF residents were expressed as months post-vaccination, with samples collected at 5 months post-vaccination (T’1), and subsequently at 8 (T’2), 11 (T’3), and 17 months (T’4). The HCW group consisted of 33 subjects (mean age, 45 years; range, 27–69 years) who received the same vaccine formulation, with sampling performed at baseline (vaccine administration, T’’0), 1 month post-dose (T’’1), and 12 months post-dose (T’’2) (Fig. 1 and Supplementary Fig. 1).

When comparing the distribution of peripheral B-cell subsets between LTCF residents and HCWs, we found a significant age-associated remodeling of the B-cell compartment (Fig. 3). Total peripheral B cell frequencies, as well as naïve B cells, were significantly reduced in LTCF residents (Fig. 3a). In contrast, the overall pool of memory B cells and plasmablasts was slightly increased (Fig. 3a), indicating preservation of the measured memory B-cell compartment. Notably, LTCF residents exhibited significantly higher frequencies of both activated and atypical memory B cells (CD19⁺CD24-CD27⁻CD38⁻CD21⁻) relative to HCWs (Fig. 3b).

Fig. 3.

Age-associated remodeling of the peripheral B cell compartment. (a) Frequencies of total B cells among lymphocytes, naïve B cells among total B cells, total memory B cells (MBCs), and plasmablasts (PBs) in healthcare workers (HCWs) and long-term care facility (LTCF) residents. (b) Distribution of memory B cell subsets, including resting MBCs, activated MBCs, and atypical B cells among total B cells. Each dot, (HCWs, blue) and long-term care facility (LTCF) residents (red), represents an individual donor; horizontal lines indicate median values. Statistical significance was assessed using the Mann-Whitney U test, with ****p < 0.0001

We then tracked B cells specific for the ancestral SARS-CoV-2 Spike (WT) and that cross-react with the XBB.1.5 variant (Fig. 4 and Supplementary Fig. 2). In the HCW reference, inclusion of an early post-vaccination time point (T’’1, 1 month) revealed a dynamic cellular response. Following the booster dose, HCWs exhibited a significant expansion of activated memory B-cells specific to the WT and that cross-react with the XBB.1.5 Spike proteins (Fig. 4b), which declined by one-year post-vaccination (T’’2), consistent with the physiological contraction of the acute response. In contrast, resting memory B cells remained elevated at T’’2 compared to pre-vaccination baseline (T’’0), indicating the establishment of durable immune memory (Fig. 4a). The LTCF resident cohort was sampled from 5 to 17 months post-vaccination. Longitudinal follow-up up to T’4 showed that both resting and activated Spike-specific B-cell frequencies were stable over time (Fig. 4a and b), suggesting maintenance of memory in the LTCF sample. HCWs at T’’2 and LTCF residents at T’4 showed no significant differences in the frequency of antigen-specific B cells, for both WT and XBB.1.5 and across resting and activated compartments (Fig. 4c). When evaluating humoral responses in the LTCF subgroup, anti-S IgG levels showed a progressive decline up to 11 months post-vaccination (T’3), followed by a significant increase at the final time point (T’4) (Fig. 4d). In parallel, anti-N IgG seropositivity increased significantly at T’4 (Fig. 4e), indicating that LTCF residents were incidentally exposed to subclinical SARS-CoV-2 infection during the study period. This late-stage rebound in anti-S IgG levels followed the same temporal trend as the anti-N seroconversion, may represent an indirect evidence consistent with memory reactivation.

Fig. 4.

Longitudinal analysis of Spike-specific memory B cell responses and antibody dynamics in HCWs and LTCF residents. (a) Frequencies of resting memory B cells (MBCs) specific for ancestral SARS-CoV-2 Spike (WT) and XBB.1.5 in healthcare workers (HCW) and long-term care facility (LTCF) residents. HCWs were analyzed at baseline (T’’0), 1 month (T’’1), and 1 year (T’’2) after the sixth vaccine dose. LTCF residents were sampled longitudinally from 5 to 17 months post-vaccination (T’1–T’4). (b) Frequencies of activated Spike-specific MBCs recognizing WT and XBB.1.5 in HCW and LTCF cohorts across the indicated time points. (c) Cross-sectional comparison of WT- and XBB.1.5-specific resting and activated MBCs between HCWs at T’’2 and LTCF residents at late time point (T’3). (d-e) Serum anti-S IgG (d) and anti-N IgG (e) titers in LTCF residents measured longitudinally from T’1 to T’4. Data are shown as individual values with distribution plots. Statistical significance was determined using Mann–Whitney U test or Wilcoxon matched pair T test (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001)

Discussion

In this longitudinal multicenter study of LTCF residents, we observed a progressive decline in SARS-CoV-2 anti-S IgG titers over 12 months following booster vaccination, consistent with the known waning kinetics of mRNA-induced humoral responses in older adults [27–30]. A more detailed analysis revealed that participants vaccinated more than one month prior to enrollment exhibited the highest IgG levels at baseline (T0). In contrast, residents vaccinated within 1 month prior to enrollment reached their peak at T1, followed by a steeper decline, until convergence. These trajectories suggest that antibody waning observed in the total sample was primarily driven by the post-vaccination dynamics of the immunized subgroups. Conversely, non-boosted individuals maintained lower but remarkably stable antibody levels throughout the follow-up. This stability, coupled with broadly comparable anti-N seroconversion rates across groups, is consistent with the hypothesis that frequent environmental exposure to SARS-CoV-2 may act as a natural booster, sustaining baseline humoral immunity even in the absence of recent vaccination. SARS-CoV-2 infection surveillance was symptom-driven throughout the follow-up period, with no systematic routine testing protocol across participating facilities. Clinically diagnosed infections represent confirmed symptomatic cases only, and asymptomatic or paucisymptomatic infections were likely underdetected. Anti-N seroconversion rates should therefore be interpreted as a conservative proxy of viral exposure rather than as definitive evidence of equivalent infection burden between groups. However, despite the vulnerability of this population, the incidence of severe clinical outcomes during the study period was relatively low. The lack of significant differences in symptomatic infection rates between boosted and non-boosted residents suggests that, in a population with a history of multiple vaccinations and a high rate of previous natural infections, an “immunological ceiling” may have been reached [17–19; 31]. In this context, cumulative immunity appears to establish a basal immune responsiveness that may contribute to protection against severe outcomes in this population.

The B-cell analysis provides a mechanistic context for these findings. LTCF residents exhibited an expansion of activated and atypical memory B-cell subsets, consistent with the coexistence of conditions such as multimorbidity and frailty, which are associated with chronic immune stimulation and inflammaging [32–34]. Whether these findings reflect age per se or the broader clinical complexity of this population could not be formally assessed given the small sample size of the B-cell subgroup. However, the B-cell changes observed in our cohort are consistent with well-described age-associated remodeling of the B-cell compartment. In a previous work, we showed that the frequency of total B cells and naïve B cells declines with age, whereas memory B cells, plasmablasts and atypical B cells increase with age [35]. Despite reduced total circulating B-cell counts, Spike-specific memory B cells, both resting and activated, are stably maintained over time in LTCF residents, with no significant decline across longitudinal sampling. The persistence of an antigen-experienced memory pool aligns with previous observations in older populations [23]. Moreover, the frequency of antigen-specific memory B cells in LTCF residents at late time points is comparable to that observed in HCWs at one-year post-vaccination, indicating that they retain a quantitatively similar pool of antigen-specific memory B cells despite the absence of early peak measurements. This parity is particularly notable given the advanced age of the LTCF cohort and likely reflects a saturated immunological plateau reached through a high cumulative vaccine burden (mean 5.5 doses, Supplementary Table 5), alongside with natural exposures. Such continuous antigenic stimulation appears to maintain the memory B-cell compartment at a stable baseline despite the physiological waning of systemic antibodies. Indeed, longitudinal analysis shows that while anti-S IgG levels declined progressively from T’1 through T’3, a marked rebound occurs at T’4, coinciding with increased anti-N IgG levels.These data suggest recent SARS-CoV-2 exposure within the LTCF residents and that pre-existing possess the potential to reactivate and generate antibodies upon antigen re-encounter. In this context, the kinetics of circulating antibodies and MBCs diverge, with vaccination or natural exposure inducing transient anti-S IgG peaks and a stably maintained memory B-cell compartment. Together, these findings suggest that long-term immune protection in older adults is less dependent on sustained high antibody titers and instead relies on the maintenance and recall capacity of a mature, antigen-experienced memory B-cell pool.

Several limitations of this study should be acknowledged. First, although antibody titers are a commonly used as proxy for humoral immunity, they do not fully capture functional neutralization capacity or cellular immune responses, both of which are critical components of adaptive immunity. Regrettably, due to the real-world operational constraints of this multicentre study during an active vaccination campaign, serum sample volumes were prioritized for longitudinal binding assay. The assessment of Spike-specific MBCs was based on antigen-probe binding and phenotypic characterization but did not include functional B-cell assays such as neutralization, antibody-secreting cell differentiation after stimulation, BCR affinity measurements, or somatic hypermutation analysis, therefore, conclusions regarding the functional quality of memory B-cell responses should be interpreted with caution. Second, the observational design and real-world setting precluded standardized timing of booster doses, which may have introduced heterogeneity in immune trajectories independent of intrinsic host factors. Third, the observational non-randomized design precluded full multivariable adjustment for baseline clinical imbalances between boosted and non-boosted residents. Targeted sensitivity analyses for the principal confounders of interest did not reveal clear clinically meaningful associations with longitudinal antibody trajectories; however, residual confounding from unmeasured variables cannot be fully excluded. Fourth, the subgroup analysis of B-cell phenotypes involved a limited number of participants and employed different sampling schedules between LTCF residents and healthcare workers, which may affect the comparability of temporal dynamics. Fifth, anti-N seroconversion was used as a proxy for SARS-CoV-2 exposur. However, it should also be noted that anti-N seroconversion, while a useful and practically accessible marker of natural SARS-CoV-2 exposure, carries important limitations in repeatedly vaccinated older populations. Anti-N responses may be blunted by immunosenescence, may wane rapidly, and may suffer loss of sensitivity when using ancestral nucleocapsid antigens to track infections caused by Omicron sub-lineages [36]. As recently highlighted by longitudinal serological trajectory-based approaches reliance on anti-N positivity alone is likely to underestimate the true burden of viral exposure in LTCF residents [37, 38].

Taken together, and despite these limitations, our findings suggest that, within the evolving context of widespread anti-SARS-CoV-2 immunity, in LTCF residents with repeated prior exposure to infection and vaccines, incremental increases in antibody titers following additional boosters may not translate into improved clinical protection against mild or asymptomatic infections. As such, the incremental benefit of universal seasonal boosters for preventing symptomatic disease may depend on individual host factors. While acknowledging the robust, cumulative protection already present in much of the older population, it is crucial to distinguish between different levels of vulnerability within the LTCF setting. In particular, the potential benefit of boosters in preventing severe disease should be emphasized for the most immunologically vulnerable individuals, including residents with severe frailty, advanced chronic diseases, and those undergoing chemotherapy or receiving immunosuppressive treatments. An evolution toward more personalized immunization strategies that would prioritize boosters for subgroups with suboptimal immune persistence is advisable. A risk-stratified approach to booster administration may therefore represent a more efficient and patient-centered strategy in LTCFs [39, 40].

Conclusions

In summary, our findings highlight the complex interplay between immunosenescence, repeated antigenic stimulation, and hybrid immunity in shaping long-term protection. Future research should aim to identify immunological surrogates that better predict clinical outcomes in frail older adults, moving beyond uniform age-based recommendations toward tailored booster strategies informed individual risk profiles.

Authors’ contributions

Authors’ contributionsConceptualization: AP, GO, GF.Methodology: AP, RC, GF. Validation: AP, CD, EPM, GF. Data curation: AP, CT, AZ.Formal analysis: AP, CD, EPM.Investigation: PL, CS, CA, IS. Resources: AM, RP, FF, TB, AC, PM, SG, RC. Writing – original draft: AP, GF.Writing – review & editing: All authors.Supervision: AP, RAI, RC, GO, GF.Funding acquisition: ATP.

Funding

This research was supported by a grant from the “Rete Italiana per la sorveglianza virologica, il monitoraggio immunologico, la formazione e la ricerca in Preparazione alla gestione delle Emergenze Infettive – R.I.Pr.E.I.”, as part of the project “Monitoraggio immunologico longitudinale della vaccinazione anti SARS-CoV-2 in coorti di operatori sanitari e residenti di RSA – Progetto R.I.Pr.E.I. 2022”.

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

Written informed consent was obtained from all participants or their legally authorized representatives. The study protocol was approved by the Italian National Ethics Committee for clinical trials of public research bodies and other national public institutions (CEN) at the Istituto Superiore di Sanità (ISS) (approval number AOO-ISS 10/10/2023 0045874). Local ethics committees subsequently provided approval or acknowledgment as required. All study procedures were conducted in accordance with applicable ethical standards, and participant data were anonymized prior to analysis.

Consent for publication

Not Applicable.

Competing interests

The authors declare no competing interests.