Persistence of immunogenicity after seven COVID-19 vaccines given as third dose boosters following two doses of ChAdOx1 nCov-19 or BNT162b2 in the UK: Three month analyses of the COV-BOOST trial.

Objectives To evaluate the persistence of immunogenicity three months after third dose boosters. Methods COV-BOOST is a multicentre, randomised, controlled, phase 2 trial of seven COVID-19 vaccines used as a third booster dose. The analysis was conducted using all randomised participants who were SARS-CoV-2 naïve during the study. Results Amongst the 2883 participants randomised, there were 2422 SARS-CoV-2 naïve participants until D84 visit included in the analysis with median age of 70 (IQR: 30–94) years. In the participants who had two initial doses of ChAdOx1 nCov-19 (Oxford-AstraZeneca; hereafter referred to as ChAd), schedules using mRNA vaccines as third dose have the highest anti-spike IgG at D84 (e.g. geometric mean concentration of 8674 ELU/ml (95% CI: 7461–10,085) following ChAd/ChAd/BNT162b2 (Pfizer-BioNtech, hearafter referred to as BNT)). However, in people who had two initial doses of BNT there was no significant difference at D84 in people given ChAd versus BNT (geometric mean ratio (GMR) of 0.95 (95%CI: 0.78, 1.15). Also, people given Ad26.COV2.S (Janssen; hereafter referred to as Ad26) as a third dose had significantly higher anti-spike IgG at D84 than BNT (GMR of 1.20, 95%CI: 1.01,1.43). Responses at D84 between people who received BNT (15 μg) or BNT (30 μg) after ChAd/ChAd or BNT/BNT were similar, with anti-spike IgG GMRs of half-BNT (15 μg) versus BNT (30 μg) ranging between 0.74–0.86. The decay rate of cellular responses were similar between all the vaccine schedules and doses. Conclusions 84 days after a third dose of COVID-19 vaccine the decay rates of humoral response were different between vaccines. Adenoviral vector vaccine anti-spike IgG concentrations at D84 following BNT/BNT initial doses were similar to or even higher than for a three dose (BNT/BNT/BNT) schedule. Half dose BNT immune responses were similar to full dose responses. While high antibody tires are desirable in situations of high transmission of new variants of concern, the maintenance of immune responses that confer long-lasting protection against severe disease or death is also of critical importance. Policymakers may also consider adenoviral vector, fractional dose of mRNA, or other non-mRNA vaccines as third doses.


Introduction
Many countries elected to deploy 3rd dose booster vaccines against COVID-19 towards the end of 2021 as a result of waning immunity and emergence of variants with varying degrees of immune escape, 1 . Results previously published from the COV-BOOST study demonstrated that most COVID-19 vaccines delivered as a 3rd dose booster provided a significant boost to both humoral and cellular immunity at 28 days following immunisation 2 . Due to their very high IgG anti-spike titres by day 7 after immunisation, mRNA vaccines were deployed by most high-income countries as the third dose booster. There is emerging real world observational evidence of significantly increased protection following a 3rd dose booster of mRNA vaccine after two initial doses of both mRNA (BNT162b2 (Pfizer-BioNtech, hereafter referred to as BNT); and mRNA1273 (Moderna, hereafter referred to as m1273)) and two doses of adenoviral vector (ChAdOx1 nCov-19 (Oxford-AstraZeneca, hereafter referred to as ChAd)) vaccines 3 . It is currently unclear how rapidly the protection from a 3rd dose booster wanes over time.
In November 2021 reports emerged of a new variant of SARS-CoV-2 (omicron) with a large number of mutations, in particular to the receptor binding domain of the spike antigen against which most currently approved vaccines are targeted. Omicron has a significant transmission advantage over previous variants due to intrinsically enhanced transmissibility and immune evasion 4 . Studies have demonstrated extremely limited neutralisation of omicron from sera following two doses of vaccine or in convalescent individuals [5][6][7] . A third dose of vaccine (or two doses plus infection) augments neutralisation against omicron in laboratory studies 8 , 9 . T cell responses appears to be preserved [10][11][12] (similar to other Variants of Concern (VOC) 2 ) which may help protect against severe disease. Observational studies also suggest a third dose significantly improves protection from symptomatic infection compared to two doses 13 , 14 .
Although a substantial number of people worldwide have already been given third dose boosters, many low and middleincome countries are still working towards administering first doses. It is, therefore, important to characterise differences in the longitudinal immune response following different vaccines given as third doses to inform possible flexible mixed vaccine third dose programmes.
There are limited data on immunogenicity beyond one month following third doses 15 , 16 , and none from randomised controlled trials. To provide further data supporting global policymaking, we conducted this day (D) 84 post-boost analysis to compare immune responses of study vaccines to the corresponding ChAd/ChAd/BNT or BNT/BNT/BNT schedule as BNT is currently the most commonly used booster in clinical practice in high income countries. Due to the emergence of omicron, commonly deployed clinical schedules tested in the trial were also analysed by viral neutralisation assays and are reported here.

Trial design & oversight, treatments
The COV-BOOST trial (ISRCTN: 73,765,130, protocol available at https://www.covboost.org.uk/protocol) has been previously reported 2 . In brief, the trial is a multicentre, randomised, controlled, phase 2 trial of third dose booster vaccination against COVID-19. The 18 study sites were split into three site groups (A, B, and C). Within each site group, the participants were randomised to three or four experimental vaccines, or a control vaccine (MenACWY), with equal probability. Trial recruitment was stratified by the first 2 dose vaccination schedule (ChAd/ChAd and BNT/BNT) and age (30-69 years old and ≥70 years old). The experimental vaccines in group A were ChAd, NVX-CoV2373 (Novavax; hereafter referred to as NVX) or a half dose of NVX; BNT, VLA2001 (Valneva; hereafter referred to as VLA), a half dose of VLA, Ad26.COV2.S (Janssen; hereafter referred to as Ad26) in group B; and m1273, CVnCov (Cure-Vac; hereafter referred to as CVn), a half dose of BNT in Group C ( Fig. 1 ). Immunogenicity bloods were taken at day 0 (pre-boost), D28 and D84 post-boost for all the participants. All the participants, laboratory staff and investigator staff were blinded to treatment allocation until the D84 visit.
Due to the general population being recommended third doses, participants in the control arms were then randomised to receive half-BNT, BNT, or half-m1273 around 6 months after their first two doses of ChAd/ChAd or BNT/BNT. Additional immunogenicity bloods were taken in this group at D0, D28 and D84 post the boost vaccine.

Laboratory methods
Sera were analysed at Nexelis (Laval, QC, Canada) to determine SARS-CoV-2 anti-spike IgG concentrations by ELISA (reported as ELISA laboratory units [ELU]/mL), and for SARS-CoV-2 pseudotype virus neutralisation (PNA) assay, using a vesicular stomatitis virus backbone adapted to exhibit the SARS-CoV-2 spike protein, reported as 50% neutralising antibody titres (NT 50 ). The conversion factors to international standard units can be found in the appendix. Sera from D0 and D84 were analysed at Porton Down, Public Health England, by ECLIA (Cobas platform, Roche Diagnostics) to determine anti-SARS-CoV-2 nucleocapsid IgG status (reported as negative if below a cut-off index (COI) of 1.0). The sera at D28 and D84 from a subset of participants with anti-SARS-CoV-2 nucleocapsid COI < 1.0 at baseline ( n = ∼25) were also tested at Porton Down, UK Health Security Agency to measure the normalised 80% neutralising antibody titre (NT 80 ) for live SARS-CoV-2 virus (wild type) by microneutralisation assays (MNA). The sera from those with the UK deployed vaccine schedules (ChAd prime and ChAd/BNT/half-m1273 third dose boost, BNT prime and BNT/ half-m1273 third dose boost) were also analysed by microneutralisation assays to determine 50% focus reduction neutralisation titres (FRNT 50 ) for live SARS-CoV-2 virus lineages (Victoria/01/2020, Delta variant B.1.617.1, and Omicron variant B.1.1.529) at the University of Oxford, Oxford, UK. The reduction in the number of infected foci was compared with a negative control well without an antibody. All assays were conducted in duplicate at minimum.
The cellular immunology samples were collected from nine sites based on logistical reasons (i.e. proximity to external laboratory) (2). IFN-γ secreting T cells specific to whole spike protein epitopes designed based on the Wuhan-Hu-1 sequence (YP_009724390.1) were detected by modified TSPOT-Discovery test within 32 h (h) of venepuncture, using the addition of T-Cell Xtend reagent to extend peripheral blood mononuclear cell (PBMC) survival, at Oxford Immunotec (Abingdon, UK). T-cell frequencies were reported as spot forming cells (SFC) per 250,0 0 0 PBMCs with a lower limit of detection of one in 250,0 0 0 PBMCs, and these results were multiplied by four to express frequencies per million PBMCs. For the rest of the study sites, sample were not taken as the sample integrity can be affected due to the long distance to the processing laboratory.

Statistical analysis
We conducted analyses on the immunogenicity outcomes at 28 and 84 days after third dose booster vaccines for available laboratory data. The sample size calculation was described previously 2 . The COV-BOOST trial was originally designed to investigate the immune responses by different third dose boost vaccines in ChAd and BNT primed participants. With the rollout of third doses world- wide based on the data generated by the COV-BOOST trial and others, the comparison to control arm has become less relevant to policymaking. BNT has become the most widely used third dose boost vaccine in the UK and most high-income countries. The analysis in this report aims to address the most relevant clinical question of the persistence of immune responses induced by other vaccines as a third dose compared with a third dose of BNT in populations who received ChAd/ChAd and BNT/BNT as their initial two dose vaccine schedules. Since BNT was only used in group B, we joined the three groups in one analysis, and the three control arms were combined into one arm.
The analysis population was all randomised participants with no evidence of SARS-CoV-2 infection up until 84 days post third dose. This was defined as self-reported SARS-CoV-2 infection or anti-nucleocapsid COI ≥1 by the Roche Elecsys anti-Sars-CoV-2 assay at baseline or D84 visit. All the analyses were conducted according to the randomised arms and stratified by first doses (ChAd/ChAd and BNT/BNT). To compare changes overtime, we presented the geometric mean ratios (GMR) and 95% confidence interval (CI) of the absolute immune responses at D28 and D84, respectively, for the study vaccines compared with BNT as the reference. If the GMRs of a vaccine to BNT increased between D28 and D84, it means the decay rate of this vaccine is slower between D28 and D84 than for BNT. We also calculated the fold-change of immunogenicity between D28 and D84 (D84 to D28 ratio) for each participant and presented the geometric mean of D84 to D28 ratio for each vaccine arm with a higher ratio indicating a slower decay. The GMRs of the D84 to D28 ratio (i.e. a ratio of ratios) were also presented with 95% CIs using BNT as the reference. The GMRs and 95% CIs were estimated using a mixed-effect linear regression model. The log10 transformed immunogenicity data (absolute titre or D84 to D28 ratio) was the dependant variable and the 'sites' variable was included as a random effect in the model with age group ( < 70 years, ≥70 years), baseline immunogenicity, the duration between 1st and 2nd vaccine, and the duration between 2nd and boost vaccine as fixed effects. The GMR was calcu-lated as the antilogarithm of the adjusted difference between arms in the model. Subgroup analyses by age ( < 70 years, ≥70 years) were carried out using the above model after removing the fixed effect of age group. Sensitivity analyses were also conducted to check the validity of the pooled analysis by comparing the GMR of each vaccine to the control arm estimated by the simple analysis within each group with the GMR estimated after pooling group A-C and combining all three control arms. Statistical analyses were conducted using R version 4.1.1.

Role of the funding source
The funders of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report.

Results
The baseline characteristics of the trial was reported previously 2 . In summary, between 1st June and 30th June 2021, the study screened 3498 participants, of whom 2883 were randomised and 2878 received a third dose boost vaccine between 10 and 26 weeks following the second dose. Recruitment was stratified by age group (30-69 years and ≥ 70 years). The median age of the younger cohort was 53 and 51 years in the ChAd/ChAd and BNT/BNT primed participants, and, respectively, 76 and 78 years in the older cohort. Amongst the 2878 participants receiving the study vaccines, there were 228 participants primed with ChAd/ChAd and 228 participants with BNT/BNT excluded, leaving 2422 participants in this analysis (CONSORT Fig. 1 ). This report fo-cuses on the results for the trial vaccines with current UK and European Union use authorization, but presents results for all vaccines for transparency.
Overall, a significant drop between D28 and D84 was seen in all study arms for anti-spike IgG, live virus neutralising antibody and cellular responses ( Figs. 2 , 3 ).
In the population who had ChAd/ChAd as first doses, full dose (100 μg) m1273 as the third dose had highest titres but, due to reactogenicity, half dose (50 μg) has been deployed worldwide. BNT standard dose (30 μg) as third dose induced higher anti-spike IgG at D28 and D84 than other vaccines deployed clinically ( Fig. 2 A)  For participants primed with BNT/BNT, the pattern of anti-spike IgG between vaccine arms at day 28 was similar to people primed with ChAd/ChAd, with BNT 30 μg as third dose inducing the highest concentration besides 100 μg m1273 ( Fig. 3 A)  For cellular responses, m1273 (100 μg) had the highest cellular responses at D84 (75, 95%CI: 51-110), though not statistically significant compared with BNT. The cellular responses at D28 and D84, as well as the D84 to D28 ratio in the ChAd and Ad26 arms were similar to the BNT arm ( Fig. 3 D).
In the subgroup analysis, similar patterns of immunogenicity were observed in the two age groups in both ChAd/ChAd and BNT/BNT first doses populations ( Fig. 4 , Fig. 5 , Fig. 6 ).
BNT-half induced similar humoral and cellular responses compared with BNT at D28 and D84 ( Fig. 2 & 3 ). This was seen in populations primed with both ChAd/ChAd and BNT/BNT, and in both age groups ( Fig. 4 , 5 &6).
The live neutralising antibody data were available in five UK deployed schedules, including ChAd/ChAd/ChAd, ChAd/ChAd/BNT and BNT/BNT/BNT with a 3-month interval between second and third doses, and ChAd/ChAd/half-m1273(50 μg) and BNT/BNT/half-m1273 (50 μg) with a 6-month interval between second and third doses. Significant reductions in neutralising titres against delta and omicron variants were observed when compared with the wild type strain at 28 days post boost dose ( Fig. 7 , supplementary Table 3). The drops in neutralisation against delta and omicron were consistent across the schedules (Supplementary Table 3). In ChAd/ChAd/ChAd arm, only 2 out of 24 participants showed detectable neutralisation against omicron, whilst neutralisation against omicron was detected in most participants of the other four schedules.

Discussion
We report D84 immunogenicity data after seven different boost vaccines in participants following ChAd/ChAd and BNT/BNT as first doses. In the ChAd/ChAd primed population, the anti-spike IgG remained highest in the mRNA vaccine arms at D84, although people given Ad26 had antibody levels that declined at a slower rate than that following mRNA vaccines between D28 and D84. In people who received BNT/BNT first doses, the anti-spike IgG at D28 had been significantly lower for ChAd and Ad26 compared with BNT as a third dose, but by D84 there was no significant difference between ChAd and BNT, and the concentration of anti-spike IgG was significantly higher for Ad26 than BNT. As reported for D28 2 , live viral neutralisation against wild type correlates with anti-spike IgG levels at day 84 and the overall pattern between arms was similar to that of anti-spike IgG for D28, D84 and the D84 to D28 ratio. T cell responses remain broad to wild type, delta and beta variants tested at D84 (supplementary Tables 1 and 2).
The anti-spike IgG and live neutralising antibody were significantly lower for NVX at both D28 and D84 in people given ChAD/ChAd and BNT/BNT as first doses compared with BNT, but a slower decay was observed between D28 and D84 than for BNT.
In both ChAd/ChAd and BNT/BNT primed populations, the GMR of anti-spike IgG induced by half-BNT (15 μg) was over 0.7 (ranging between 0.74 to 0.88) compared with BNT standard dose (30 μg), indicating that the anti-spike IgG levels were similar by three months following a third dose of standard or half dose BNT. Although recent data suggest using two first doses of 10 mcg To our best knowledge, this is the first study reporting persistence of immunogenicity for homologous and heterologous boost schedules from a randomised controlled trial. In December 2021, the European Medicines Agency (EMA) published its regulatory considerations on heterologous primary and booster COVID-19 vaccination 19 , based on evidence generated from short-term immunogenicity studies and a vaccine effectiveness study 2 , 20 , 21 . The EMA concluded the immunogenicity of heterologous boost schedule is as good as, or better than, homologous schedules. Our data at D84 post third dose further support the EMA's statement. The mRNA vaccine arms still have the highest anti-spike IgG in the ChAd/ChAd first doses population, although the heterologous boost schedule anti-spike IgG with Ad26 after ChAd/ChAd appears to decay slower than ChAd/ChAd followed by mRNA. Based on limited available data, the EMA also suggested 19 that heterologous schedules with adenoviral vector vaccine prime doses and mRNA vac- cine third dose is more immunogenic than the reverse. However, based on our data adenoviral vector vaccines may be as immunogenic by D84 following third dose as mRNA vaccine. The anti-spike IgG in adenoviral vector vaccine arms (ChAd and Ad26) after the BNT/BNT prime are the most persistent schedules up to D84. The immunogenicity at D84 post boost for ChAd and Ad26 was similar to, or higher than, the three dose BNT schedule (BNT/BNT/BNT), especially in older people. Although the WHO does not yet recommend third doses for healthy adults due to the inequity of vaccine distribution worldwide 22 , the data from our study also supports WHO recommendations to consider using adenoviral vector vaccines for third doses in countries implementing mRNA vaccine as initial doses. The use of fractional mRNA dosing may be another solution to accelerate the worldwide vaccine coverage rate. The anti-spike IgG level in the half BNT (15 μg) arm was > 70% compared with that in the full dose BNT (30 μg) arm, whilst the difference was even smaller in the BNT/BNT prime population.
In the UK, mRNA vaccines were initially chosen for third doses to achieve the highest possible peak antibody levels given a likely resurgent wave in autumn/winter 2021. As maximum antibody levels following third mRNA doses are achieved by day seven after the third dose 2 , our previous data also supported acceleration of the UK third dose programme to try to control omicron transmission. A third dose mRNA vaccine given as a boost to people who received BNT/BNT 23 and ChAd/ChAd 21 has also shown increased effectiveness compared to two doses to prevent symptomatic, severe and hospitalised COVID-19 infection. These data highlight the particular need to use third doses in vulnerable populations to reduce the mortality and burden to healthcare systems. Although a third dose of viral vector vaccine was not a widespread deployed schedule, recent UK data have shown good long term protection against hospitalisation and death for Omicron even in the population who received ChAd/ChAd/ChAd for logistical reasons 14 . Given the high correlation observed between humoral responses and vaccine efficacy following two doses 24 , it is likely that the two doses of mRNA vaccine with an adenoviral vector 3rd dose will achieve similar protection as three doses of mRNA vaccine. Importantly, our antibody decay rate data suggest that two doses of mRNA followed by an adenoviral vector vaccine is likely to achieve more sustainable protection. Our 8-month follow up visit will further investigate the longer-term immunogenicity persistence. Using adenoviral vector vaccines as third doses following two doses of mRNA vaccine will not only make more mRNA vaccine available for people who have not yet received their first two doses, but could also delay any potential need for a fourth dose. In countries that have not yet implemented third doses, and where omicron has already passed through the population, policymakers will need to assess the risk/benefit of a potentially longer lasting third dose schedule balanced against the possibility of the extremely rare side effect of thrombosis with thrombocytopenia syndrome (TTS) which has not been observed after second doses and is not detected across all ethnicities and geographies.
Preliminary data on cross reactive neutralisation against omicron suggest that, amongst the combinations so far evaluated, a lower neutralising response with ChAd/ChAd/ChAd and highest responses where half (50 μg) dose m1273 has been used as third dose, irrespective of primary schedule. The neutralising antibody levels against omicron at D28 following a third dose of mRNA vaccine in people who had received ChAd/ChAd and BNT/BNT were between 125 and 756 (FRNT 50 ). This lies between levels of antibodies against wild type after priming with two-doses of ChAd or BNT (FRNT 50 : 109 and 1501 respectively) 25 . Both assays were run by the same laboratory following the same procedure. Our data suggest that protection against omicron infection after a third dose of mRNA vaccine is likely be the same as that against wild type after two doses of ChAd/ChAd and BNT/BNT. The vaccine effectiveness after three doses of mRNA vaccine were 82% and 90% for severe and hospitalised Omicron predominated COVID-19 infections, which is consistent with our immunogenicity findings 23 .
There are some limitations of this analysis. The original trial included seven candidate third dose vaccines with three also tested at half dose, and the trial was designed by splitting the study sites into three groups to randomise participants into control vaccine and three or four study vaccine arms. This means that the study vaccines were not all randomised within the same study popula- tions, making the comparison of vaccines between groups more complicated than our original report, which compared to the control arm within each group. Little difference was observed in a sensitivity analysis on the GMR to the control arm that was conducted by comparing the results from simple analysis within group (used in the primary endpoint paper) with the results from the combined approach in this analysis (Supplementary Figure 1). Secondly, m1273 was used at full dose (100 μg) as a boost in this study, as the decision (and international regulatory approval) to use 50 μg was made after the start of the study; therefore, the data presented for m1273 third dose cannot be directly used to inform policy making. However, participants in the control arms within the original trial were subsequently randomised to third doses with half-BNT (15 μg), BNT (30 μg), and half-m1273 (50 μg) at a 6month interval, following UK policymaker advice. These data will provide evidence on the optimal interval of mRNA boost and the immunogenicity of 50 μg mRNA1273 as a boost. Finally, this analysis was done in a seronegative population to inform the policy making in September 2021, when the majority of worldwide population were SARS-CoV-2 naïve. This population no longer representative of most global populations, where a substantial proportion of people will have had at least one SARS-CoV2 infection. Subsequent analysis will include the impact of prior infection on post third dose responses over the length of the study.
In conclusion, substantial differences in the decay rates of humoral responses between study vaccines used as third doses were observed. The heterologous schedule with mRNA vaccine first two doses followed by adenoviral vector vaccine third dose showed more persistent humoral responses as well as comparable or higher antibody responses at D84 post third dose. 15 μg BNT also showed comparable immune response compared with standard 30 μg dose BNT when used as a third dose. Using vaccines in heterologous manner ("mix and match") is relatively novel, as are the vaccines being used in the mixed platforms investigated in this study and using different dosing schedules. This analysis has demonstrated that there is much to be learnt about these and other heterologous vaccine combinations for SARS-CoV2, and vaccines against other infectious pathogens.

Contributors
SNF, MDS, XL and JSN-V-T conceived the trial and SNF is the chief investigator. SNF, AM, MDS and XL contributed to the proto-col and design of the study. AM, GB and SS led the implementation of the study. XL, SF, LJ and VC designed and conducted the statistical analysis and have verified the underlying data. XL, AM, SF and SNF drafted the report. All other authors contributed to the implementation and data collection. All authors reviewed and approved the final report.

Declaration of Competing Interest
KC acts on behalf of University Hospital Southampton as an investigator on studies funded or sponsored by vaccine manufactur- Fig. 6. Continued ers including AstraZeneca, GlaxoSmithKline, Janssen, Medimmune, Merck, Pfizer, Sanofi and Valneva. She receives no personal financial payment for this work. SNF acts on behalf of University Hospital Southampton NHS Foundation Trust as an Investigator and/or providing consultative advice on clinical trials and studies of COVID-19 and other vaccines funded or sponsored by vaccine manufacturers including Janssen, Pfizer, AstraZeneca, Glaxo-SmithKline, Novavax, Seqirus, Sanofi, Medimmune, Merck and Valneva vaccines and antimicrobials. He receives no personal financial payment for this work. ALG is named as an inventor on a patent covering use of a particular promoter construct that is often used in ChAdOx1-vectored vaccines and is incorporated in the ChAdOx1 nCoV-19 vaccine. ALG may benefit from royalty income paid to the University of Oxford from sales of this vaccine by AstraZeneca and its sublicensees under the University's revenue sharing policy. JH has received payments for presentations for AstraZeneca, Boehringer Ingelheim, Chiesi, Ciple & Teva. VL acts on behalf of University College London Hospitals NHS Foundation Trust as an Investigator on clinical trials of COVID-19 vaccines funded or sponsored by vaccine manufacturers including Pfizer, AstraZeneca and Valneva. He receives no personal financial payment for this work. PM acts on behalf of University Hospital Southampton NHS Foundation Trust and The Adam Practice as an investigator on studies funded or sponsored by vaccine manufacturers including As-traZeneca, GlaxoSmithKline, Novavax, Medicago and Sanofi. He received no personal financial payment for this work. JSN-V-T is sec- onded to the Department of Health and Social Care, England until 31st March 2022. MR has provided post marketing surveillance reports on vaccines for Pfizer and GSK for which a cost recover charge is made. MDS acts on behalf of the University of Oxford as an investigator on studies funded or sponsored by vaccine manufacturers including AstraZeneca, GlaxoSmithKline, Pfizer, Novavax. Janssen, Medimmune and MCM vaccines. He received no personal financial payment for this work.

Data sharing
The study protocol is provided in the appendix. Individual participant data will be made available when the study is complete upon reasonable requests made to the corresponding author; data can be shared through secure online platforms after proposals are approved. All the sequence datasets used in the T-cell analysis are available in the public GISAID database (https://www.gisaid.org).