Strong peak immunogenicity but rapid antibody waning following third vaccine dose in elderly residents of care homes (preprint)

Third dose COVID-19 vaccines are being deployed widely but their efficacy has not been assessed adequately in vulnerable elderly people who exhibit suboptimal responses after primary series vaccination. We studied spike-specific immune responses in 341 staff and residents in long-term care facilities (LTCF) who received an mRNA vaccine following dual primary series vaccination with BNT162b2 or ChAdOx1. Third dose vaccination strongly increased antibody responses with preferential enhancement in older people and was required to elicit neutralisation of Omicron. Cellular immune responses were also enhanced with strong cross-reactive recognition of Omicron. However, antibody titres fell 21-78% within 100 days post vaccine and 27% of participants developed a breakthrough Omicron infection. These findings reveal strong immunogenicity of a 3rd vaccine in one of the most vulnerable population groups and endorse an approach for widespread delivery across this population. Ongoing assessment will be required to determine the stability of immune protection.


Introduction
Age and frailty are major risk factors for severe COVID-19 outcome and elderly residents of long term care facilities (LTCFs) have suffered relatively high rates of mortality during the current pandemic (1). Single or dual COVID-19 vaccination has provided strong clinical protection against severe disease within this group (2,3)but there is concern about the potential impact of immune waning and the need for additional vaccines in those at greatest risk (4) SARS-CoV-2 infection rates have been high in many LTCF and studies have shown that many staff and residents have evidence of prior natural infection (5) Importantly, this acts to strengthen vaccine-induced immunity such that older residents achieve comparable levels of antibody and cellular immunity as younger staff following primary series dual vaccination with mRNA or adenovirus-based vaccines (6,7) However, these responses are markedly attenuated within the older adult population who have remained uninfected. In particular, antibody and cellular responses here are reduced by 62% and 50% respectively compared to younger donors (4). As such, the delivery of a third vaccine for this population has been prioritised.
This importance of boosting and sustaining vaccine-induced immune responses in elderly LTCF residents has been given considerable impetus through the emergence of the Omicron variant. This has a very high infection rate and evades a large component of the vaccine-induced humoral immune response (8)(9)(10). In contrast, spike-speci c cellular responses are more reliably maintained (11) although these have not been assessed in older people. Third dose vaccination appears effective in helping to suppress Omicron infection rates (12) although it is not clear for how long this effect will be maintained due to antibody waning (13) We undertook an analysis of humoral and cellular spike-speci c immune responses in staff and residents of LTCF after the 3rd vaccine dose and compared these to values that had been recorded after dual vaccination. We nd that there is a robust antibody and cellular response to third vaccination in the elderly resident population which is on par with the responses seen in the much younger population within LTCFs.

Sample Collection
The VIVALDI study (ISRCTN14447421) is a prospective cohort study which was set up to investigate SARS-CoV-2 transmission, infection outcomes and immunity in residents and staff in LTCFs in England that provide residential and/or nursing care for adults aged 65 years and over (https://wellcomeopenresearch.org/articles/5-232/v2).
Eligible LTCFs were identi ed by the Care Provider's Senior Management Team, or by the National Institute for Health Research (NIHR) Clinical Research Network. Pseudonymised clinical (vaccination status, PCR test results, hospitalisation, death) and demographic (age, sex, staff member versus resident) data were retrieved for staff and residents from participating LTCFs through national surveillance systems. All participants provided written informed consent for blood sample collection or if residents lacked the capacity to consent, a personal or nominated consultee was identi ed to act on their behalf.
Blood sampling was carried out from 25th May 2021 until 23rd of February 2022. An anti-coagulated EDTA blood sample was sent to the University of Birmingham and a serum tube was also obtained for The Doctors Laboratory where anti-nucleocapsid IgG (N) testing using the Abbott immunoassay was performed as well as a viral neutralisation assay at the Crick Institute. Ethical approval for this study was

Data linkage
Abbott antibody test results were submitted to the COVID-19 datastore (https://data.england.nhs.uk/covid-19/), pseudonymised and linked to routinely held data on age, sex, LTCF, role (staff or resident), and results of PCR or lateral ow device (LFD) SARS-CoV-2 testing performed through the national SARS-CoV-2 testing programme. Asymptomatic screening using PCR tests is performed weekly in staff and monthly in residents with more frequent LFD or PCR testing during outbreaks. Using the common pseudo-identi er based on the individuals' NHS number, linkage was undertaken to vaccination status (date and vaccine type) derived from the National Immunisations Management System (NIMS) and dates and diagnostic codes for hospitalisations recorded in the Hospital Episode Statistics (HES) dataset as well as for any deaths from the O ce for National Statistics (ONS) dataset. Individual-level records were further linked to each LTCF using the unique Care Quality Commission location ID (CQC-ID), allocated by the Care Quality Commission who regulate all providers of health and social care in the UK.

Inclusion criteria
Both staff and residents were eligible for inclusion if samples could be linked to a pseudo-identi er enabling data linkage. We included samples from participants that had received a primary vaccine course with or without a third vaccine dose. Participants sampled in the 7 days following third vaccine administration were excluded from the analysis to ensure peak immune responses were reached before sampling. Due to limited PCR testing in the rst wave of the pandemic, it was not possible to determine when individuals had been infected with SARS-CoV-2 based on PCR alone. Past infection with SARS-CoV-2 was de ned based on results of MSD antibody test and Abbotts test using thresholds and methods outlined below.

Sample Preparation
Samples were processed within 24 hours of reception at the University of Birmingham. Blood was spun at 300 xg for 5 minutes. Plasma was removed and spun at 500 xg for 10 minutes prior to storage at -80°C.
The remaining blood was separated using a SepMate (Stemcell) density centrifugation tube. The resulting PBMC layer was washed twice with RPMI and rested overnight in R10 (RPMI + 10% FBS + Pen/Strep) media at 37°C in 5% CO2.
Serological analysis of SARS-CoV-2-speci c immune response Quantitative IgG antibody titres were measured against trimeric Spike (S) protein and Nucleocapsid (N) protein and variants of concern (VOCs) using the MSD V-PLEX COVID-19 IgG Kit (SARS-CoV-2 Panel 2, 22 and 23) following manufacturer's instructions (Lot number K0081795). Brie y, 96-well plates were blocked. Following washing, plasma samples were diluted at 1:5000 in diluent and added to the wells with the reference standard and internal controls. After incubation, plates were washed and anti-IgG detection antibodies were added. Plates were washed and were immediately read using a MESO TM QuickPlex SQ 120 system. Data was generated by Methodical Mind software and analysed with MSD Discovery Workbench (v4.0) software. Presented data were adjusted for any sample dilutions.

Quanti cation of SARS-CoV-2-speci c cellular responses by ELISpot
Pepmixes pool containing 15-mer peptides overlapping by 10aa from either SARS-CoV-2 Wuhan or Omicron Spike S1 or S2 protein domains were purchased from JPT Peptide Technologies (Germany). T cell responses of post-vaccination and post-booster samples to the above peptide mixes were determined using a Human IFNγ ELISpot PRO kit (Mabtech, Sweden). Isolated PBMC rested overnight in R10 (RPMI + 10% FBS + Pen/Strep) at a concentration of 2-3x10 were stimulated in duplicate with peptide mixes at 2ng/ml per peptide, anti-CD3 and CEFX cell stimulation mix (JPT Cat: PM-CEFX-2) as a positive control, or DMSO as a negative control for 16-18 hours. Supernatants were harvested and stored at -80°C. Following the development of plates per the manufacturer's instructions, the plates were read using the BioSys Bioreader 5000. Mean spot counts in DMSO treated negative control wells were deducted from the means to generate normalised spot counts for all other treated wells. Cut off values were previously determined (16). Intracellular cytokine staining 1.5x10 PBMC were stimulated with either SARS-CoV-2 Spike S1 and S2 peptide pool at a nal concentration of 2ng/ml per peptide for 6 hours. Protein transport inhibitor and CD107a-speci c antibody were added after 1 hour and PBMC washed with MACS (PBS+5% BSA+1% EDTA) prior to addition of Brilliant Stain Buffer (BD) and surface staining at 4oC for 30 minutes (supplementary table 1). Cells were washed and resuspended in Cell Fixation Buffer (eBioscience) at 4oC overnight. Cells were re-washed and human serum and saponin added to samples 5 minutes before the addition of cytokine-speci c antibodies (supplementary table 2) and incubation in the dark at room temperature for 30 minutes. Cells were washed twice with MACS and run on a BD Symphony A3 ow cytometer (BD Biosciences) with analysis carried out using FlowJo v10.7.1 software (supplementary gure 3) (FlowJo). Cells that appeared in both the IFN-γ and IL-2 positive gates were taken as the dual positive IFN-γ+IL-2+ cells.
All viral isolates were propagated in Vero V1 cells. Brie y, 50% con uent monolayers of Vero V1 cells were infected with the given SARS CoV-2 strains at an MOI of approx. 0.001. Cells were washed once with DMEM (Sigma; D6429), then 5 ml virus inoculum made up in DMEM was added to each T175 ask and incubated at room temperature for 30 minutes. DMEM + 1% FCS (Biosera; FB-1001/500) was added to each ask. Cells were incubated at 37° C, 5% CO2 for 4 days until extensive cytopathogenic effect was observed. Supernatant was harvested and clari ed by centrifugation at 2000 rpm for 10 minutes in a benchtop centrifuge. Supernatant was aliquoted and frozen at -80°C. Full protocol for variant virus culture and microneutralization assay available from Wu et al 2022.
High-throughput live virus microneutralisation assay was run as previously described (eLife 2021;10:e69317 DOI: 10.7554/eLife.69317). Speci cally, VERO E6 cells (Institut Pasteur) at 90-100% con uency in 384-well plates (Greiner) were infected with SARS-CoV-2 variants at an MOI of <1 in the presence of patient serum samples in serial dilutions. Cells were xed with 4% nal formaldehyde, blocked and permeabilised with 3% BSA + 0.2% TritonX-100 in PBS (v/v), and infected cells stained using a Biotin-CR3009 antibody (produced in-house), which speci cally detects SARS-CoV-2 N-protein, and Streptavidin-Alexa488 (Invitrogen). Cellular DNA detected using DAPI. An Opera Phenix (Perkin Elmer) automated microscope was used to image whole wells at 5x and the uorescent areas calculated using the Phenix-associated softwards Harmony (Perkin Elmer). The sample IC50 against a variant was estimated by tting a 4-parameter dose response curve using SciPy and reported as the fold-dilution of serum samples required to inhibit 50% of detected infection, with additional annotation if the result lies outside the quantitative range (complete inhibition, no inhibition, weak inhibition).

Statistical analysis
All data were checked for normality using the Kolmogorov-Smirnov test. For comparative analysis of 2 groups a Mann Whitney test was applied. For paired data, 2-tailed paired t-tests were applied. For comparative analysis with 3 or more groups a Kruskal-Wallis test was used, and for multiple comparisons uncorrected Dunn's test was used for non-parametric data. Spearman's rank correlation coe cients were calculated and tested for correlations. P values <0.05 were considered to be statistically signi cant.
The cumulative incidence of breakthrough infection following 3rd vaccination dose based on LFD or PCR testing was estimated and Kaplan-Meier curves were plotted. Participants entered the cohort 7 days following the date of their 3rd vaccination dose and were censored at the date of the nal PCR or LFD test if infection did not occur, or on the date of 4th vaccine administration (n=5). Participants who had not undergone testing with PCR or LFD after vaccination were excluded from this analysis (n=36). The cumulative incidence of infection was also compared between those aged 65 years and over and those under 65 using the log-rank test.
In a subset of participants who had undergone blood sampling between 7 and 50 days following 3rd vaccination dose (taken as time of peak vaccine response) spike-speci c antibody and cellular response values were modelled against breakthrough infection using linear regression against time from vaccine dose. Participants were grouped into those with spike antibody responses above and below the predicted median titre and those with cellular responses above and below the predicted median level for their time point of observation. The cumulative risk of infection was compared separately against spike-speci c antibody titre or the spike cellular immunity using Kaplan-Meier curves. Participants entered this cohort analysis at 50 days following the 3rd vaccination dose and exited at the date of a positive LFD / PCR test or were censored at the last recorded test date. Separate Cox proportional hazards models were constructed to estimate the hazard ratios of breakthrough infection based on the level of humoral and cellular immunity against spike antigen (high or low).

Results
Antibody responses are boosted strongly following third vaccination in staff and residents of care facilities Blood samples were obtained from 341 staff and residents within LTCF following the third vaccine dose ( Table 1). The median age of the staff was 48 years (IQR: 40-58, n=183) whilst that of the residents was 84 years (IQR: 76-92, n=158). 48% of donors had received primary series vaccination with mRNA vaccines (either BNT162b2 or mRNA-1273) whilst 52% had received ChAdOx1. The third vaccine comprised an mRNA formulation in every case with 336/341 recipients receiving BNT162b2 (P zer) and 5/341 receiving mRNA-1273 (Moderna). Residents received their third vaccine dose somewhat earlier than staff and samples were obtained at a median of 92 (IQR: 31-113) days following the third vaccine. Spike-speci c and nucleocapsid-speci c antibody levels were determined using the MSD platform. A positive nucleocapsid-speci c value or prior history of PCR-con rmed COVID-19 infection before sampling were taken as evidence of prior natural infection and subsequent studies were analysed in relation to infection status.
Initial analyses compared antibody levels in donors aged <65 and >65 years which segregated >95% of staff and residents by age. Antibody levels in both staff (<65 years) and residents (>65 years) were seen to increase strongly following the third vaccine. Amongst those with prior infection, titres increased by 2.4-fold (106700 AU/ml (961 BAU/ml) vs 259327 AUml (2336 BAU/ml)) and 2.7-fold (136898 AU/ml (1233 BAU/ml) vs 374158AU/ml (3371 BAU/ml)) (p=<0.0001) in staff and residents respectively compared to dual vaccination (4). This increment was more marked in those who remained infectionnaive where values increased by 3.3-fold (39149 AU/ml (352 BAU/ml) vs 131127 AU/ml (1181 BAU/ml)) and 4.3-fold (16299 (146 BAU/ml) AU/ml vs 70522 AU/ml (635 BAU/ml)) (p=<0.0001) respectively ( Figure 1A). Despite this increment antibody titres in infection-naïve older residents remained 42% lower than younger staff members (131,127 AU/ml (1181 BAU/ml) vs 70,522 AU/ml (635 BAU/ml)) (p=0.05). Of note, prior infection with SARS-CoV-2 consistently boosted antibody responses across the life course even after the third vaccine. Charlson Comorbidity Index did not show any in uence on the spike speci c antibody titres (Supplementary gure 1) The impact of age and natural infection on spike-speci c antibody response was next assessed using age as a continuous variable ( Figure 1B). A trend was seen towards higher level response in older people with prior infection but this was not signi cant (r=0.13, p=0.053).
These data show that third vaccine doses are effective in augmenting antibody levels beyond those seen following the primary series and that this increment is particularly marked in the older adult cohort.
Antibody responses following third vaccine show signi cant waning within the rst 100 days postvaccine We next assessed spike-speci c antibody levels in relation to time after 3rd vaccine delivery in order to assess for potential immune waning. Median values within the rst 50 days after vaccine were taken as the peak response and were compared to those at 100-150 days. Antibody levels were seen to fall markedly during this 100 day period and this was in uenced by both prior infection status and age. In particular, median titres fell by 62% and 21% respectively in <65s and >65s with a prior SARS-CoV-2 infection. These compared to 78% and 75% respectively in those who were infection-naive. Spearman's ranked correlation analyses con rmed the stability of immune response in prior infected older people (r=0.003, p =0.97) although a decline was observed in staff (r=-0.48, p=<0.0001) (Figure 2A). A strong statistical association with antibody waning over time was observed in staff and residents who remained infection-naive (r=-0.63, p=<0.0001 and r=-0.50, p =<0.0001 respectively) ( Figure 2B).
To further assess the importance of antibody waning over time we also utilised access to 47 pairs of serum samples that were taken at different timepoints from individual donors after the 3rd vaccine dose.
As such, 3rd vaccine doses elicit strong peak antibody responses irrespective of age or infection status. However, these values are not sustained and fall substantially within 3-months in donors regardless of infection status.
Antibody binding to the whole spike or RBD region of the Wuhan, Delta and Omicron viral variants was also assessed on MSD plates (Supplementary Figure 2). Antibody binding was strongest against the Wuhan spike protein but reduced by 40% (p=<0.0001) and 52% (p=<0.0001) respectively against Delta in donors with prior infection or who were infection-naive. In contrast, values were 76% (p=<0.0001) and 75% (p=<0.0001) lower respectively against Omicron. Binding to RBD of Wuhan and Delta was equivalent but a marked 81% (p=<0.0001) and 78% (p=<0.0001) fall in binding to Omicron was seen in prior infected or infection-naive cohorts. These ndings show that relative antibody binding to Omicron, and most particularly the RBD domain, is markedly reduced after three vaccine doses.  Figure 4) compared to 300789 AU/ml (IQR: 142710-550512) in those with neutralisation. Selection of an antibody level that predicts neutralisation is important as choosing an inappropriately low threshold could lead vulnerable individuals being denied additional vaccine doses or therapeutic monoclonal antibodies. These data indicate that an MSD value >142710 AU/ml (1285 BAU/ml) would identify all non-responders with >70% speci city.

Neutralisation of the Delta and
Strong, stable and cross-reactive cellular responses are induced following 3rd vaccination Cellular responses are increasingly recognised as key mediators of immune protection with a central role in protection against viral variants and severe infection. As such we next went on to assess spike-cellular responses in LTCF donors through use of IFN-γ ELISpot.
As previously reported (4)  Given that signi cant antibody waning was observed within the rst 100 days post 3rd vaccine we next went on to assess potential waning of the cellular response. Importantly, in both prior infected and infection-naive donors we did not observe signi cant waning in the cellular response ( Figure 5D, PI r=0.04, p=0.66, NPI r=-0.12, p=0.22).
These ndings show that cellular responses are consistently boosted after three vaccines and become equivalent between younger and older donors. Furthermore, the great majority of this response is retained against the Omicron variant, implying strong residual protection during the current pandemic, and no short-term waning of response is observed.

Spike-speci c CD4+ T cells are dominated by IL-2 expression, but prior infection drives T cell differentiation and increased production of IFN-γ
To assess the features of spike-speci c T cells in more depth we then used intracellular cytokine analysis to identify and phenotype virus-speci c populations. CD4+ T cells were seen to dominate the global T cell response and the magnitude of the CD8+ populations was too low to reliably facilitate further analysis.
Virus-speci c CD4+ cells comprised populations that secreted single or dual expression of IL-2 and IFN-γ ( Figure 6A). Indeed, single positive IL-2 cells were the most common, representing 8.  Figure 6C).
Finally, the relationship between memory subset, cytokine production and differentiation status as assessed by CD27 and CD28 co-expression was determined (Supplementary Figure 7). This showed loss of CD27-in 25-50% of cells during transition to the effector pool and accumulation of a CD27-CD28subset within the effector-memory CD45RA-revertant (TEMRA) pool. IFN-γ+ expression was markedly enhanced within the highly differentiated pool.

Spike-speci c immune responses after 3rd vaccine do not correlate with protection against Omicron infection
Initial third vaccine doses within this cohort largely predated the Omicron variant but national booster administration programs were accelerated with the aim of reducing its impact. As such, we next determined the rate of breakthrough infection within the whole cohort and assessed the potential in uence of the magnitude of spike-speci c antibody or cellular responses on the risk of infection.
Donors were followed from 7 days after the date of 3 rd vaccine and 305 were identi ed that had PCR or lateral ow tests undertaken post-vaccine. 81 (27%) were seen to develop infection during follow-up of 170 days (4,822 person-days) (Fig 7A). Only one donor was hospitalised and there were no fatal outcomes. Breakdown by age showed no difference in infection rate in relation to age <65 and >=65 years (p=0.0993) (Fig 7B).
Data on spike-speci c antibody and cellular immune response within the 7-50 day period of peak response post 3rd vaccine were available for 122 participants. Comparison of donors with antibody or cellular responses that were below or above median predicted value showed no relationship to risk of infection over the subsequent 120 days (Fig 7C/D).

Discussion
The COVID-19 pandemic is estimated to have led to the death of over 18 million people and older people who require residential and/or nursing care are at particularly high risk of mortality. Vaccines have transformed the clinical outlook but many questions remain regarding their optimal delivery. Here we undertook a detailed assessment of the e cacy of a third vaccine dose within this age group and identi ed a number of features that can help to guide future vaccine policy.
At the time of the introduction of COVID-19 vaccines, the primary series course consisted of either one or two vaccine doses and there was hope this might provide long-term protection. However, the uncovering of immune waning and breakthrough infection has led to recommendations that additional vaccine doses should be administered to many demographic groups (4,(14)(15)(16). For healthy immunocompetent people these additional doses are typically termed "booster vaccines", whilst for those with immune suppression they have become regarded as a constituent of the primary series (17). The third vaccine is generally regarded as a booster vaccine in elderly residents of care homes despite the fact that this group are themselves typically relatively immunosuppressed due to age and comorbidity (18,19). As such, within this report we refer to the third vaccine dose without reference to its primary or booster status.
Previous studies have revealed a de cit in immune protection within older people in residential care following COVID-19 vaccination (20)(21)(22). A relatively unique feature of early studies of vaccine immunogenicity in care homes was the high rates of prior natural infection within this community (23). As such, this allowed assessment of the impact of previous SARS-CoV-2 infection on vaccine immunogenicity. Indeed, infection was seen to substantially increase both the magnitude and functional quality of the antibody and cellular response (24). However, a de cit in immune response was seen within infection-naive residents following primary series vaccination with lower levels of antibody and cellular response compared to younger staff members (4,16). This provides the basis for the current assessment.
Third vaccine doses were seen to strongly improve vaccine immunogenicity. In particular, antibody levels increased by between 2.4-fold and 4.3-fold when assessed by prior infection status or age. An incremental, albeit smaller, improvement was also observed in the cellular response. An important and encouraging feature was that immune responses were preferentially enhanced in older people without prior infection, the group with the greatest prior de cit.
The Omicron SARS-CoV-2 variant is now globally dominant and it is essential to assess vaccine-induced immunity against this challenge. Overall ndings were encouraging. Antibody binding to Omicron spike and RBD was lower than against the original Wuhan strain. The level of Omicron BA.1 neutralisation was also lower compared to Delta. However, a third vaccine was seen to be essential to induce neutralisation activity. Cellular responses were also extremely well maintained against Omicron with a fall of only 8-12% compared to recognition of peptide pools from Wuhan spike (25,26). These ndings are compatible with epidemiological studies of Omicron infection within care homes which show a markedly reduced rate of disease severity compared to previous waves of infection (12).
A feature of our study that raises some concern for future protection was the rate of decline of spikespeci c antibodies following the third vaccine dose. This fell by 21%-78% within different subgroups and was notably higher in those without prior infection. Furthermore, waning was somewhat higher in staff, a feature previously observed for the nucleocapsid-speci c response (14), and may re ect relative selection for elderly survivors with a more robust immune system. In contrast, cellular responses remained stable. Further assessment of this cohort is required to assess the longer term stability of antibody responses and clinical protection. However, these ndings will contribute to suggestions of the need of a fourth vaccine dose although initial studies within younger donors indicate that this fails to further elevate peak responses but may provide a short-term boost.
The development of immune correlates of protection could transform the introduction of new and optimised vaccines. Antibody magnitude and relative neutralisation activity are two such factors, although neutralisation assays remain challenging to determine. Taking advantage of our large dataset we were able to demonstrate that antibody binding to the ancestral Wuhan spike is strongly correlated with neutralisation of all tested viral variants, including Omicron. As such this provides support for the use of this determinant as a potential measure of personal protection against subsequent infection.
Considerable debate is being given to the potential development of novel vaccine formulations that may act to sustain vaccine-induced immune responses in the longer term. As such, increasing attention is being focussed on detailed features of the adaptive immune response. CD4+ T cell responses were found to dominate the cellular responses and IL-2+ cells were 8-fold higher than those producing IFN-γ, indicating that IFN-γ ELISpot are likely to underestimate virus-speci c cellular memory. Similar features have been observed after natural infection (27) and these high levels of IL-2 production augur well for potential long-term immune memory. Indeed, many spike-speci c CD4+ T cells were seen to remain within the central memory subset that replenishes effector pools. Natural infection increased the proportion of IFN-γ effector cells and the relationship of this observation to improved clinical protection will be an important area for future study. In contrast, a late differentiated CD27-CD28-pool emerged within the TEMRA subset and this phenotype is associated with reduced proliferative potential (28). The extent to which repeated vaccine dosing may lead to 'vaccine exhaustion' within spike-speci c T cell populations has not yet been assessed but these features indicate that this should now be addressed, particularly in elderly populations in whom naive T cell pools are limited.
Finally, we were also interested to assess how the magnitude of the peak immune response after the 3rd vaccine dose might relate to protection against breakthrough infection. 27% of participants developed infection over the 170 days of follow up and this rate is consistent with the high prevalence of Omicron variant over this period. Immune response data of the peak response within the rst 50 days after vaccine was available from 122 participants but was not associated with risk of infection. However, the clinical severity of infection was modest, with only one hospitalisation, and this is likely to re ect the strong impact of vaccine-induced immune protection.
The limitations of this study include the fact that we did not have access to the exact time or severity of primary infection for participants with prior infection. Information on patient ethnicity was also not available. There is also the potential for some waning of nucleocapsid speci c IgG to the extent of prior infected donors being marked as non-prior infected, we estimate the chances of this being low.
Furthermore, this was a retrospective analysis of prospectively collected data.
In conclusion, we show that a third vaccine dose is highly effective at boosting antibody and cellular responses in elderly and vulnerable residents in care homes and is essential to deliver antibody and cellular protection against Omicron. As such we suggest that these should be regarded as a constituent to primary series vaccination in this vulnerable cohort. However, rapid antibody waning was observed in people who were infection-naive and breakthrough infections occurred in all groups. As such further studies are needed to assess the potential value of additional vaccine doses. Figure 1 Spike speci c antibody responses are boosted after 3rd vaccine A) Wuhan spike-speci c antibody titre after 2 or 3 COVID-19 vaccine doses in relation to prior infection status and in staff or resident groups.