SARS-CoV-2-specific immune responses and clinical outcomes after COVID-19 vaccination in patients with immune-suppressive disease

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) immune responses and infection outcomes were evaluated in 2,686 patients with varying immune-suppressive disease states after administration of two Coronavirus Disease 2019 (COVID-19) vaccines. Overall, 255 of 2,204 (12%) patients failed to develop anti-spike antibodies, with an additional 600 of 2,204 (27%) patients generating low levels (<380 AU ml−1). Vaccine failure rates were highest in ANCA-associated vasculitis on rituximab (21/29, 72%), hemodialysis on immunosuppressive therapy (6/30, 20%) and solid organ transplant recipients (20/81, 25% and 141/458, 31%). SARS-CoV-2-specific T cell responses were detected in 513 of 580 (88%) patients, with lower T cell magnitude or proportion in hemodialysis, allogeneic hematopoietic stem cell transplantation and liver transplant recipients (versus healthy controls). Humoral responses against Omicron (BA.1) were reduced, although cross-reactive T cell responses were sustained in all participants for whom these data were available. BNT162b2 was associated with higher antibody but lower cellular responses compared to ChAdOx1 nCoV-19 vaccination. We report 474 SARS-CoV-2 infection episodes, including 48 individuals with hospitalization or death from COVID-19. Decreased magnitude of both the serological and the T cell response was associated with severe COVID-19. Overall, we identified clinical phenotypes that may benefit from targeted COVID-19 therapeutic strategies.


SARS-CoV-2-specific immune responses and clinical outcomes after COVID-19 vaccination in patients with immune-suppressive disease
Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) immune responses and infection outcomes were evaluated in 2,686 patients with varying immune-suppressive disease states after administration of two Coronavirus Disease 2019 (COVID- 19) vaccines. Overall, 255 of 2,204 (12%) patients failed to develop anti-spike antibodies, with an additional 600 of 2,204 (27%) patients generating low levels (<380 AU ml −1 ). Vaccine failure rates were highest in ANCA-associated vasculitis on rituximab (21/29, 72%), hemodialysis o n i mm un os up pr essive therapy (6/30, 2 0 % ) a n d s o l id organ transplant recipients (20/81, 25% and 141/458, 31%). SARS-CoV-2-specific T cell responses were detected in 513 of 580 (88%) patients, with lower T cell magnitude or proportion in hemodialysis, allogeneic hematopoietic stem cell transplantation and liver transplant recipients (versus healthy controls). Humoral responses against Omicron (BA.1) were reduced, although cross-reactive T cell responses were sustained in all participants for whom these data were available. BNT162b2 was associated with higher antibody but lower cellular responses compared to ChAdOx1 nCoV-19 vaccination. We report 474 SARS-CoV-2 infection episodes, including 48 individuals with hospitalization or death from COVID-19. Decreased magnitude of both the serological and the T cell response was associated with severe COVID-19. Overall, we identified clinical phenotypes that may benefit from targeted COVID-19 therapeutic strategies.
The rapid development of vaccines against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has been hugely effective in the management of the Coronavirus Disease 2019 (COVID-19) pandemic 1,2 . National vaccination programs have shown that COVID-19 vaccines prevent wild-type SARS-CoV-2 infection and protect against severe disease from other SARS-CoV-2 variants, including Omicron 3 . However, volunteers in the original vaccine trials were healthy, without known chronic disease and not receiving immune-modifying treatments. In the United Kingdom (UK), in 2019, more than 60% of people aged over 65 years had one or more chronic disease, with more than 12 million people aged 18-65 years living with a chronic condition lasting more than 12 months 4 . UK government estimates suggest that 500,000 people have immune-suppressive diseases. Disease cohort studies 5 and population studies using primary care health records 6 showed that immune-suppressed patients are at increased risk of severe COVID-19 and death after SARS-CoV-2 infection in the pre-COVID-19 vaccine era. Many studies have shown suboptimal COVID-19 vaccine immune responses in cohorts of patients with chronic disease and in those receiving immune-suppressive therapy [7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23] . In general, these studies have focused on specific disease cohorts, and few have robustly evaluated cellular immune responses. Furthermore, vaccine responses against SARS-CoV-2 Omicron have been rarely assessed in specific     24,25 . Population studies including immune-suppressed patients have shown lower rates of SARS-CoV-2 spike antibody positivity after vaccination with only moderate vaccine effectiveness 26 and have identified immune-suppressive disease after vaccination as a risk factor for severe COVID-19 and death 27,28 . Immune-suppressive disease remains a risk factor for severe outcomes with Omicron infection [29][30][31] , even though this variant appears less pathogenic, even when accounting for confounders, including vaccination status 32 .
In this prospective, multi-center study (Observational Cohort trial T cells, Antibodies and Vaccine Efficacy in SARS-CoV-2 (OCTAVE)), we evaluated functional humoral and T cell responses after COVID-19 vaccination, using centralized immune assays in patients receiving immune-suppressive therapy (for solid cancer, hematological malignancy, ANCA-associated vasculitis on rituximab, inflammatory arthritis, liver and kidney transplantation, autoimmune liver disease on immunosuppression, inflammatory bowel disease, ulcerative colitis and undefined inflammatory bowel disease); patients receiving autologous and allogeneic hematopoietic stem cell transplant (auto-HSCT and allo-HSCT); patients treated with chimeric antigen receptor (CAR) T cells; or patients with disease states known to modulate immune responses intrinsically (patients with end-stage kidney disease receiving hemodialysis with or without immune suppression and patients with advanced liver disease). Patients were vaccinated using mRNA (BNT162b2 or mRNA-1273) or ChAdOx1 nCoV-19 encoding ancestral SARS-CoV-2 spike antigens according to UK government-recommended vaccine schedules, and vaccine responses were evaluated before and after homologous first dose (V1) and second dose (V2) vaccination.
Patients were recruited for evaluation of SARS-CoV-2 serological responses 28 d after V2 with the magnitude of the T cell response assessed in a large subset of patients longitudinally (primary study endpoints). These responses were compared to a healthy control cohort matched by age, sex, prior SARS-CoV-2 infection and vaccine type, and the safety profile of vaccines in patient populations was assessed. Cellular and humoral responses were associated with SARS-CoV-2 infection events and COVID-19 disease severity. Exploratory endpoints included characterization of functional T cell and humoral responses and immune analysis in blood and saliva against variants of concern (VOCs). Using pairwise and regression analysis, we determined the contribution of disease phenotype, drug therapy and vaccine type to COVID-19 humoral and cellular vaccine responses, identifying patient subgroups that failed to seroconvert. Using uniform sampling timepoints and centralized immune assays, we directly compared COVID-19 vaccine immune responsiveness and infection outcomes among multiple disease phenotypes in immune-suppressive disease.

SARS-CoV-2 spike and common cold coronavirus immunoglobulins
Serological and cellular immune responses were evaluated before V1, immediately before V2 and 28 d after V2 in 674 patients and in healthy controls matched by age, sex, prior SARS-CoV-2 and vaccine type (Supplementary Table 6). Median anti-RBD Ig titers were lower before V2 in ANCA-associated vasculitis, hemodialysis on immunosuppression, liver transplant, allo-HSCT and CAR-T groups compared to healthy controls ( Fig. 2a and Supplementary Table 7) but increased after a second COVID-19 vaccine in all disease groups other than ANCA-associated vasculitis. Spike, RBD and N-terminal domain (NTD) IgG and spike IgA responses significantly increased after both one vaccine dose (P < 0.0001) and two vaccine doses (IgG P < 0.0001 and IgA P = 0.0003). Spike IgM and NTD IgA increased significantly only after the first dose (Fig. 2b). IgM responses to RBD and NTD did not increase after either dose (Extended Data Fig. 2a,b). We correlated IgG/IgA/IgM to each common cold coronavirus (CCC) spike protein at baseline with SARS-CoV-2 spike IgG after one and two vaccines in seronegative anti-nucleocapsid IgG patients: only IgG to HCoV-OC43 showed a weak positive correlation after the first vaccine ( Fig. 2c and Extended Data Fig. 3).

Cellular immune responses after vaccination
T cell responses to spike and nucleocapsid were evaluated before V1, before V2 and 28 d after V2 using the Oxford Immunotec T-SPOT Discovery IFNγ ELISpot assay in 656 patients and 210 matched healthy controls. After V2, the hemodialysis (P < 0.003) and allo-HSCT (P < 0.003) groups had a significantly higher proportion of T cell a * * * * * *  T   151  39  18  29  116  582  25 183 62  48  89  131 92  3  28  49  4  74  58  8  0  82  108  6  271 19  20  31  25  15  2  5  41  4 1,497 695 non-response compared to the healthy control group ( Fig. 3a and Supplementary Table 8). IFNγ-secreting T cell magnitude to spike antigens was lower in liver transplant (z = 3.821, P = 0.004) and allo-HSCT (z = 3.339, P = 0.03) groups compared to the healthy control group (Fig. 3b  and Supplementary Table 9). To complement the serological characterization of Omicron BA.1 responses, an in-house IFNγ ELISpot assay was used to investigate T cell responses to ancestral and Omicron BA.1 spike in the 59-patient subset after V2. Regardless of prior SARS-CoV-2 status, the T cell response to full spike was maintained against Omicron peptides relative to ancestral (Fig. 3c,d), although stimulation with a reduced peptide set containing only peptides with variant amino acids relative to ancestral showed a significant decrease in Omicron reactivity (Fig. 3c). Serological and T cell responses showed weak correlations at pre-V2 only in healthy controls (r = 0.24, P = 0.02) but at pre-V2 and post-V2 timepoints in the overall patient cohort (pre-V1: r = 0.34, P < 0.0001; post-V2: r = 0.22, P < 0.0001) (Fig. 3e,f and Extended Data Fig. 6). In the ANCA-associated vasculitis group (with all patients taking B-cell-depleting therapies), there was no relationship between anti-RBD Ig and SARS-CoV-2 spike T cell responses; here, antibody responses were low after both vaccines, but many generated robust T cell responses after one vaccine (Fig. 3g,h). Positive correlations were also seen in disease subgroups at pre-V2 and/or post-V2 timepoints (Extended Data Fig. 6).

Predictors of vaccine humoral and cellular responses
The contribution of demographics, disease group, vaccine type, prior SARS-CoV-2, therapeutic regimen and time between vaccines to vaccine immunogenicity was assessed using multivariable logistic regression in OCTAVE patients compared to matched healthy controls. Patients aged 65-74 years had significantly lower odds of having a robust serological response (Roche anti-RBD Ig >380 AU ml −1 ) compared to patients in the 15-44-year age group ( Fig. 4a and Supplementary In evaluating T cell responses, we used a responder threshold of ≥4 spot-forming cells (SFCs) per 10 6 peripheral blood mononuclear cell (PBMCs) (Fig. 4b and Supplementary Table 12). The only disease group with reduced cellular responses was allo-HSCT (OR: 0.09, 95% CI 0.02-0.41). In contrast to the serological results, vaccination with BNT162b2 was associated with significantly decreased odds of generating a cellular response (OR: 0.23, 95% CI 0.11-0.51). Previous SARS-CoV-2 infection significantly increased the odds of generating a cellular response (OR: 4.05, 95% CI 1.5-10.9). No other variables were associated with T cell response.

Discussion
People with immune-suppressive disease remain vulnerable to [29][30][31], and identifying patient populations most at risk remains a UK government imperative. We show that, after two vaccines, in comparison to healthy volunteers, a substantial minority of immune-suppressed patients generated low-magnitude SARS-CoV-2 antibodies (in particular, ANCA-associated vasculitis on rituximab, hemodialysis on immunosuppressive therapy and solid organ transplant recipients), and that, although T cell responses were generally maintained, these were also reduced in some patient groups (hemodialysis, allo-HSCT and liver transplant recipients). Lower serological or T cell responses were associated with hospitalization or death from COVID-19.
Although vaccine correlates of immune protection against SARS-CoV-2 are not precisely defined, there is consensus that higher antibody titers are advantageous [33][34][35][36][37][38] . Waning immunity enhances disease susceptibility, especially in patients with additional comorbidities 39 , whereas higher levels of antibodies generated by booster vaccines are protective 40 . SARS-CoV-2-specific T cells protect against SARS-CoV-2 infection 41 and appear less susceptible to viral escape as VOCs have emerged 42 . Previous studies highlighted the role of serological responses in protecting immune-suppressed patients, including (1) inflammatory arthritis with breakthrough infections increased in those who fail to seroconvert after vaccination 43 ; (2) primary immune deficiency with increased COVID-19 mortality compared to the general population after vaccination 44 ; and (3) renal disease with both breakthrough infection and COVID-19 severity/mortality correlating with serological responses [44][45][46] .
SARS-CoV-2 infection rates varied among disease subtypes and were higher in patients with no detectable antibody or T cells. However, infection rates cannot be confidently ascribed to disease phenotype, as social shielding behavior and SARS-CoV-2 exposure are likely to have differed among groups. However, disease severity in those infected can be definitively correlated with vaccine responsiveness. Although most (93.6%) patients had asymptomatic or mild infection, a substantial number had severe COVID-19 (33/440), and, additionally, 15 patients died. Patients with severe COVID-19 included ANCA-associated vasculitis, inflammatory arthritis, hemodialysis, hemodialysis on immunosuppression, kidney transplant, liver transplant, cirrhosis, Crohn's disease, allo/auto-HSCT and CAR-T. Failure to seroconvert and the magnitude of the serological and cellular response were each associated with severe disease. However, one quarter of patients with severe disease seroconverted and had antibody levels similar to healthy controls, highlighting the fact that other factors contribute to disease susceptibility-for example, disease phenotype and/or comorbidities. Although most infections occcured during the Omicron time epoch, there were proportionally many more severe infections in the Delta epoch. The Omicron epoch coincided with the rollout of new therapeutic strategies, additional vaccines and a dominant VOC that is less pathogenic 32 , and it is not possible to disentangle the relative contribution of each of these to clinical outcomes in our study.
Vaccination with BNT162b2 generated higher antibody responses, whereas cellular responses were higher in patients who received ChAdOx1 nCoV-19, as previously shown in healthy populations [47][48][49][50][51][52] . Two studies in hemodialysis and solid organ transplant recipients assessed cellular responses in different vaccine types and showed no difference, but responses were low magnitude and the studies were underpowered to detect a difference 16,53 . The reasons why ChAd vaccines generate higher T cell responses than mRNA vaccines may relate to more persistent antigen expression in lymph nodes and the stimulation of distinct immune pathways, with ChAdOx dependent on robust follicular helper T (Tfh) cell responses 54 , whereas mRNA vaccines are Tfh independent 55 . Heterologous vaccination most effectively boosts T cell and antibody titers in healthy people and in solid organ transplant recipients 50,56,57 and should be further evaluated in immune-suppressed patient groups. In this study, all immune-suppressant drug classes evaluated (other  than biological therapies) were associated with a suboptimal serological response to vaccination, and, therefore, all patients receiving these therapies should be considered at risk for severe COVID-19.
In evaluating the effect of CCC cross-reactive antibodies on vaccines responsiveness, only HCoV-OC43 antibody titers (the CCC with the highest homology to SARS-CoV-2 spike 58-60 ) were predictive of serological response to the first vaccine only. This shows that cross-reactive memory immune responses do not contribute to vaccine responsiveness once a significant SARS-CoV-2 memory pool has been established in immune-suppressed patients. Although T cell cross-reactivity using whole spike antigens against Omicron was maintained, antibody responses against all VOCs were decreased in both blood and saliva, and, at Roche anti-RBD Ig titers below 4,000 AU ml −1 , most patients failed to neutralize Omicron BA.1. We suggest that this antibody titer threshold be explored in subsequent studies as a possible correlate of protection in immune-suppressed patients. New COVID-19 vaccine boosting regimens that account for the loss of recognition against emerging SARS-CoV-2 variants are likely to be particularly required for clinically vulnerable patients. The strengths of our study include recruitment of a large number of patients and healthy controls, a wide range of disease phenotypes, national geographical spread, robust standardized procedures and standardization of timepoints with infection outcomes. Limitations include missing baseline data (due to the very rapid delivery of the vaccination program in vulnerable groups), no randomization for vaccine type, controls recruited in a separate study and heterogeneity of disease duration, severity, therapeutic regimen and comorbidities within groups. Additionally, OCTAVE assessed responses after two COVID-19 vaccines, whereas vulnerable patients have now typically received three or more vaccines. Nevertheless, because all healthy volunteers generate high-magnitude immune responses after two vaccines, the head-to-head comparison to assess relative vaccine effectiveness in disease cohorts compared to healthy controls remains informative.
Overall, our data are reassuring because most patients generated robust T cell responses and moderate serological responses and had mild COVID-19. However, the fact that some patients groups fail to generate high-magnitude immune responses, and the association of these with severe COVID-19, highlights the importance of developing Odds ratio (95% CI) 10.00  third vaccine doses in immune-suppressed patients has fully recruited (ISRCTN15354495). Although our data may be directly used to inform COVID-19-related vaccination strategies in vulnerable patients, the future use of stored biological samples to evaluate biological pathways in secondary immune deficiency, alongside the relative risks of COVID-19 disease severity and immune responsiveness in disease subgroups, may inform clinical strategies in relation to vaccines or infection susceptibility in general.

Online content
Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https:  (h) stratified by COVID-19 infection/severity-includes all non-hospitalized individuals with or without SARS-CoV-2 infection (infected and uninfected, non-hospitalized) and infected individuals who were not hospitalized with COVID-19 (infected, non-hospitalized) and individuals who were hospitalized or died with COVID-19 (infected, hospitalized). Fisher's exact tests (d,f) or two-sided Mann-Whitney rank-sum statistical tests (g,h) were used without correction for multiple comparisons. a,b, Lines represent COVID-19-free progression, and shading represents 95% CI. g,h, Boxes represent median and IQR; whiskers represent ±1.5× IQR. *P < 0.05, ****P < 0.0001. Ab, anti-SARS-CoV-2 RBD-binding total Ig; Hosp-Oxygen, hospitalized with COVID-19 and required oxygen; Hosp-No Oxygen, hospitalized with COVID-19 but did not require oxygen; mo, month.

Trial design and oversight
The OCTAVE trial is a multi-center, multi-disease, prospective cohort trial of the immune response to SARS-CoV-2 vaccination in patients receiving COVID-19 vaccination as part of routine, publicly funded National Health Service (NHS) care. The trial is a collaboration between the universities of Birmingham, Glasgow, Imperial All patients who fulfilled the patient characteristics inclusion criteria could be enrolled into either study group. Investigators generally recruited patients into group 1 first where possible, before patients had received two vaccines as part of the rapid UK COVID-19 vaccine program. Some patients opted for group 2 because fewer study visits were involved (patient choice).
OCTAVE recruited 2,686 adult patients: 2,012 patients into the serology group and 674 patients into the deep immune phenotyping group (see Supplementary Table 21 for recruitment per site). The full protocol (Supplementary Appendix 1) is available in the supplementary material.
The deep immunophenotyping cohort was assessed pre-vaccine (baseline), pre-second vaccine dose (pre-V2), 28 d post-second vaccine dose (post-V2, within ±3 d), 6 months post-second vaccine dose and 12 months after the first vaccine dose (as close to timepoints as possible). The serology cohort was assessed 28 d post-second vaccine dose (−7/+56 d) and followed up 6 months post-second vaccine dose and 12 months after the first vaccine dose (as close to timepoints as possible).
Anti-RBD total Ig immunogenicity data were available in 2,204 patients and 225 healthy controls. T cell data were available in 656 patients in the deep immunophenotyping cohort and in 210 controls. Infection data were available in 1,648 OCTAVE patients at the <6-month post-V2 timepoint and 1,613 OCTAVE patients at the 6-month post-V2 to 12-month post-V1 timepoint. Severity data were available in 93% of those infected. AEs were reported in 2,662 patients post-V1 and 2,669 OCTAVE patients post-V2. Detailed information regarding the number of samples included in each immunogenicity assays is available in Extended Data Fig. 1.
After trial entry, 66 adult participants were found to have been recruited at the wrong timepoint in accordance with the eligibility criteria: 'Have not received their second dose of the vaccine for the 'deep immunophenotyping group' or have not passed the day 28 post-second vaccine dose timepoint (21-84 d after second vaccination) for the 'serology group''; and, for 24 participants, trial consent was obtained after the collection of post-booster samples (although these patients were recruited into another ethically approved study with full consent) before trial consent in OCTAVE. This was reported to the MHRA as a serious breach, and patients were still included in the analysis.

Outcome measures
The primary outcomes for this trial are the magnitude of the anti-SARS-CoV-2 IgG antibodies and the magnitude of the T cell responses to SARS-CoV-2 peptides after vaccination. The secondary outcome is the proportion of first-symptomatic, PCR-proven COVID-19 infection 14 d after V1 in participants without evidence of prior COVID-19 infection. The exploratory outcomes are described in detail in the protocol (Supplementary Appendix 1).
This manuscript represents the definitive analysis of the primary outcome for the adult cohort.

Vaccine administration
Vaccine (BNT162b2 (Pfizer/BioNTech) or ChAdOx1 nCoV-19 vaccine) was administered in line with its temporary authorization under Regulation 174 of the Human Medicines Regulations 2012, the national recommendations and guidance of the Joint Committee on Vaccination and Immunisation ( JCVI) and current standard NHS practice. The trial has no influence on the type of vaccine given or the timing of the booster vaccine delivery. Vaccines were administered both through NHS pathways and by OCTAVE study investigators. The interval between vaccines was in accordance with national recommendations and the guidance of the JCVI. As vaccines were being given to new patient populations, this study was registered with the MHRA (UK MHRA clinical trial authorization number: 21761/0365/001).

AEs
AEs were captured up to 28 d after the second vaccine and were graded 1-5 using the Common Terminology Criteria for Adverse Events (CTCAE), version 4.03.

Sample collection
Serum samples were collected 4 weeks post-second dose (−7/+14 d) for all participants, alongside whole blood for the Oxford Immunotec assay, PBMCs and plasma, when feasible. Where available, baseline (pre-vaccine samples, including samples that may have been collected before recruitment to OCTAVE) or pre-second dose samples taken any time after V1 but before the second dose were included. All samples were collected in accordance with national regulations and requirements, including standard operating procedures for logistics and infrastructure. Samples were taken in appropriately licensed premises Article https://doi.org/10.1038/s41591-023-02414-4 and stored and transported in accordance with Human Tissue Authority guidelines and NHS Trust policies.

Anti-SARS-CoV-2 Ig analysis
The magnitude of anti-SARS-CoV-2 antibodies was measured using the Roche Elecsys AntiSARS-CoV-2 S and Roche Elecsys AntiSARS-CoV-2 N assays by the UKHSA Laboratories at Porton Down. The Roche assay measures the presence and the amount of serum antibodies to the spike RBD antigen of SARS-CoV-2. Seroconversion is defined as a response equal to or greater than 0.8 AU ml −1 , and no response is defined as less than 0.8 AU ml −1 . Low response was defined on the Roche anti-RBD Ig assay after evaluation of the serological response to vaccine in healthy controls. A cutoff for low was defined as the upper value of the bottom decile of healthy controls.

T-SPOT DISCOVERY SARS-CoV-2 assay
The magnitude of the T cell responses was measured using the T-SPOT DISCOVERY SARS-CoV-2 assay by Oxford Immunotec (https://www. tspotdiscovery.com/). Peptide pools representing the full spike (S) proteins, subunits S1 and S2, nucleocapsid and membrane, plus positive (phytohaemagglutinin) and negative controls were used to stimulate 250,000 PBMCs separated from fresh whole blood. IFNγ-secreting T cells were enumerated on an automated plate reader. Final values were calculated by subtracting the negative control and multiplying by 4 to define the number of IFNγ-secreting T cells per 10 6 PBMCs. Values greater than or equal to four IFNγ-secreting T cells per 10 6 PBMCs were defined as a positive response. In the renal cohort (hemodialysis with immunosuppression and hemodialysis), the full spike peptide pool was not included in the assay at all timepoints. To generate equivalent data, the S1 and S2 values were combined, and a cutoff of four IFNγ-secreting T cells per 10 6 PBMCs was used for positivity. There was strong correlation between S1 + S2 and full spike pools in this assay (r = 0.90, P < 0.0001) (Extended Data Fig. 8).

IFNγ T cell ELISpot assay
Frozen PBMCs were thawed, and the Human IFNγ ELISpot Basic Kit (Mabtech, 3420-2A) at Oxford University laboratories was used, as in ref. 65, but using 200,000 cells per well in duplicate. For antigens, we used overlapping peptide pools (18-mers with 10 amino acid overlap, mimotopes) representing ancestral spike (S1 and S2), Omicron (B.1.1.529 and BA.1) spike (S1 and S2), ancestral membrane and nucleocapsid SARS-CoV-2 proteins and pools of ancestral or Omicron 18-mer peptides covering the mutated regions of Omicron at a final concentration of 2 μg ml −1 . The mean spots of the negative control wells were subtracted from the test wells and then multiplied by 5 to give antigen-specific responses expressed as SFU per 10 6 PBMCs. Total spike responses were defined by adding S1 and S2 responses together.

Meso Scale Discovery IgG, IgA and IgM binding assays
IgG, IgA and IgM responses to SARS-CoV-2, SARS-CoV-1 and seasonal coronaviruses were measured using a multiplexed Meso Scale Discovery (MSD) immunoassay: V-PLEX COVID-19 Coronavirus Panel 2 Kit (K15369U-2) from Meso Scale Diagnostics. IgG was measured in all group 1 participants at all available timepoints; IgA and IgM were measured in ulcerative colitis, Crohn's disease, liver transplant, autoimmune liver disease, cirrhosis, inflammatory arthritis and ANCA-associated vasculitis disease groups at all available timepoints. A MULTI-SPOT 96-well, 10-spot plate was coated with four SARS-CoV-2 antigens (spike, RBD, nucleoprotein and NTD) and SARS-CoV-1 spike trimer as well as spike proteins from seasonal human coronaviruses, OC43, HKU1, 229 E and NL63 and BSA. Antigens were spotted at 200−400 μg ml −1 (MSD Coronavirus Plate 2). Multiplex MSD assays were performed as per the manufacturerʼs instructions. To measure IgG, IgA and IgM binding antibodies, 96-well plates were blocked with MSD Blocker A for 30 min. After washing with washing buffer, plasma samples diluted 1:1,000-10,000 in diluent buffer were added to wells, along with MSD standard or undiluted MSD internal controls. After a 2-h incubation and a washing step, detection antibody (MSD SULFO-TAG Anti-Human IgG, IgA or IgM Antibody, 1/200) was added. After washing, MSD GOLD Read Buffer B was added, and plates were read using a MESO SECTOR S 600 Reader. The standard curve was established by fitting the signals from the standard using a four-parameter logistic model. Concentrations of samples were determined from the electrochemiluminescence signals by backfitting to the standard curve and multiplied by the dilution factor. Concentrations are expressed in AU ml −1 . Cutoffs were determined for each SARS-CoV-2 antigen (spike, RBD, nucleoprotein and NTD) based on 64 pre-SARS-CoV-2 pandemic sera (average concentration + 3× standard deviation for IgG, IgA and IgM binding) (Supple mentary Table 22). As samples were from UK individuals with low probability to have been exposed to SARS-CoV-1, a cutoff for SARS-CoV-1 spike was similarly determined.

Microneutralization assay
Sera were serially diluted in DMEM supplemented with 1% FBS from an initial dilution of 1:10 to 1:10,000. Equal volumes of diluted sera and SARS-CoV-2 virus (approximately 100 foci-forming units (FFU)) were combined and incubated for 30 min. Viruses used in this assay were Victoria (VIC01) and Omicron (B.1.1.529 (BA.1)). After incubation, 100 μl of Vero E6 cells (4.5 × 10 5 per milliliter) was added to each well, and virus was allowed to infect the cells for 2 h at 37 °C and 5% CO 2 , followed by the addition of 100 μl of carboxymethyl cellulose (1.5%) to each well. The plates were incubated for a further 20-22 h at 37 °C and 5% CO 2 . All assays were carried out in triplicate.
Cells were washed with 200 μl of DPBS and then fixed with paraformaldehyde 4% v/v (100 μl per well) for 30 min at room temperature. Cells were permeabilized with Triton X-100 (2% in PBS) and then stained for SARS-CoV-2 nucleoprotein using a human monoclonal antibody (FB9B). Bound antibody was detected after incubation with a goat anti-human IgG HRP conjugate (Sigma-Aldrich) and, after TrueBlue Peroxidase Substrate (Insight Biotechnology) addition, imaged using an ELISpot reader. The IC 50 was defined as the concentration of compound that reduced the FFU by 50% compared to the control wells.

Clinical data
Clinical data were collected electronically from participating sites using a REDCap (Research Electronic Data Capture) database held at the CRCTU. Data collected on trial entry included sex, ethnicity, body • A post-boost sample was available.
• Data relating to age, sex, prior COVID status and vaccination type were available. • Samples were taken within the same timeframe defined for the OCTAVE participant samples.
After sorting, the healthy control group was matched to the OCTAVE patient group based on prior COVID status and vaccine type using a proportional matching method. The relevant participants for each group were then randomly selected from the available control data to match the four analysis groups: • Complete OCTAVE dataset (deep immunophenotyping and serology groups together) • OCTAVE deep immunophenotyping only • OCTAVE serology group only • OCTAVE renal disease group only A separate analysis group was created for the renal disease group, as these patients did not have the full spike peptide pool tested on the T-SPOT DISCOVERY at all timepoints due to the timing of recruitment.

Statistical analysis
Dataset preparation. A healthy control data pool comprising 986 participants (231 PITCH study and 755 UKHSA dataset) was sampled to create the healthy controls analysis group by matching, as closely as possible, the proportions of age, sex, prior COVID status and vaccination type, as observed in the OCTAVE recruits. ELISpot assay raw data required processing before analytical use; the control blank readings were subtracted from the sample readings, and any negative numbers were replaced with zero; the new values were then multiplied by 4 to give cell counts per million.

Analysis
Descriptive statistics data are presented as number of observations with percentages and medians with interquartile ranges (IQRs), unless indicated otherwise. Data were questioned using two-sided statistical tests, including Kruskal-Wallis H test (as data non-normal) with post hoc Dunn's testing for pairwise comparisons, χ 2 and Fisherʼs exact tests where required. Where multiple testing occurred, P values were adjusted using Bonferroni corrections. Pearson's correlations and logistic regressions investigated data relationships. COVID-19-free time post-second vaccination was estimated using the method of Kaplan and Meier (1958). Time was taken from date of the second vaccine to date last seen. COVID-19 events were taken as the date of confirmed SARS-CoV-2 infection or date of death confirmed as being COVID-19 related. The number of SARS-CoV-2 infection events per 1,000 d was calculated by using (total number of SARS-CoV-2 infections) / (total number of days at risk post-second vaccination) × 1,000 = number of SARS-CoV-2 infection events per 1,000 d. COVID-19 disease severity was defined as asymptomatic; symptomatic without hospitalization; hospitalization with COVID-19 without requirement of oxygenation; hospitalization with COVID-19 with requirement of oxygenation; admission to ITU with COVID-19; or death related to COVID-19. Data were visualised using various plot types: box and whisker with overlaid data points using the jitter function to aid interpretation; vertically stacked bar charts of percentages of response types (non-responder, low-responder and responder); scatter plots with Pearson's correlation analysis line; connected scatter plot showing changes; and matrix (panel) correlation plots illustrating assay results. All analyses used the statistical package Stata version 17.0 (StataCorp), GraphPad Prism (version 9.4.0) or R (version 4.2.1) with RStudio 2022.02.3.

Statistics and reproducibility
The OCTAVE trial is a multi-center, multi-disease, prospective observational cohort trial. Statistical analyses were completed, independently replicated and compared to initial results. No discrepancies were found. The sample size was based on an estimation of the number of participants who could be recruited within the short space of time required. An effect size calculation, based on a t-test, was done to provide information as to whether the number of recruits would be sufficient. Missing data were excluded from the analyses. No other data were omitted. As an observational study, experiments were not randomized, and investigators were not blinded to allocation.

Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability
Participant data and the associated supporting documentation will be available for requesting within 6 months after the publication of this manuscript. Details of our data request process are available on the Nature Medicine Article https://doi.org/10.1038/s41591-023-02414-4 CRCTU website. Only scientifically sound proposals from appropriately qualified research groups will be considered for data sharing. The decision to release data will be made by the CRCTU Director's Committee, which will consider the scientific validity of the request, the qualifications and resources of the research group, the views of the Chief Investigator and the trial steering committee, consent arrangements, the practicality of anonymizing the requested data and contractual obligations. A data-sharing agreement will cover the terms and conditions of the release of trial data and will include publication requirements, authorship and acknowledgements and obligations for the responsible use of data. An anonymized encrypted dataset will be transferred directly using a secure method and in accordance with the University of Birmingham's IT guidance on encryption of datasets. Information on data requests, including a contact address and expected timeframe of requests, can be found at https://www.birmingham.ac.uk/research/ crctu/data-sharing-policy.aspx.

Code availability
Code is available upon reasonable request through the same process as data, as detailed on the CRCTU website: https://www.birmingham. ac.uk/research/crctu/data-sharing-policy.aspx.