SARS-CoV-2 Vaccine Responses in Individuals with Antibody Deficiency: Findings from the COV-AD Study

The immunological correlates of protection against SARS-CoV-2 infection and severe COVID-19 are not yet known. The passive acquisition [1] or development of anti-SARS-CoV-2 spike glycoprotein antibodies following infection [2‐4] confers significant protection against future disease and, in some cases, facilitates viral clearance in individuals that fail to mount effective immune responses following infection [5‐9].

Vaccination against SARS-CoV-2 is the most effective public health intervention to prevent severe morbidity and mortality from COVID-19 in the general population [10‐12]. A meta-analysis of vaccine efficacy studies has suggested that neutralising antibody levels are strongly associated with protection from symptomatic infection [13]. However, it is well recognised that patients with immunodeficiency may not respond optimally to vaccination. To date, SARS-CoV-2 vaccine immunogenicity and efficacy has not been comprehensively studied in individuals with primary and secondary immunodeficiency; preliminary studies suggest seropositivity rates following vaccination vary between 20.0 and 83.0% [14‐18]. Given the significantly increased risk of morbidity and mortality from COVID-19 that these patients face [19, 20], understanding the immunogenicity and efficacy of vaccines in this population is of critical importance.

COVID-19 in patients with antibody deficiency (COV-AD) is a multi-site UK study that aims to: (i) determine the prevalence of asymptomatic and symptomatic SARS-CoV-2 infection in patients with primary and secondary antibody deficiency, (ii) determine how frequently SARS-CoV-2 viral persistence occurs in patients with primary and secondary antibody deficiency and (iii) characterise the immune response of these patients following SARS-CoV-2 infection and vaccination. This manuscript presents an interim analysis of 320 participants in the COV-AD study to describe responses to the primary course of vaccination and the risk of vaccine breakthrough and viral persistence.

The results of 320 participants in the COV-AD study were available for interim analysis (Table 1). The median age of participants was 58.5 years and 40% (n = 128/320) were male. The median interval between the first and second vaccine dose was 76 days; 42.1% (n = 135/320) of participants received the Pfizer BioNTech 162b2 vaccine and 55.0% (n = 176/320) the AstraZeneca ChAdOx1 nCoV-19 vaccine.

Eighteen participants (n = 18/320, 5.6%) had suffered PCR-proven SARS-CoV-2 infection prior to study entry; these participants were significantly younger (52.0 vs. 59.0 years, p = 0.02) than individuals who remained SARS-CoV-2 infection-naive (Table 2). Only one participant remained SARS-CoV-2 PCR positive on study entry. Eleven participants (n = 11/18, 61.1%) returned negative nasopharyngeal swabs at the time of study entry and six participants declined further investigation. No specific immunological characteristics defined the population with apparent viral clearance: 4 patients had CVID, 2 other primary antibody deficiencies, 1 Good’s syndrome, 1 XLA and 3 secondary immunodeficiencies; 54.4% (n = 6/11) made no persistent serological response to infection (measured at study enrollment) and 36.4% (n = 4/11) no serological response to subsequent vaccination. T cell responses were assessed in five of the six seronegative individuals by interferon-gamma ELISPOT: 100% (n = 5/5, median spots/10⁶ cell = 158) mounted responses against spike peptide pools and 60% (n = 3/5, median spots/10⁶ cells = 45) to the nucleocapsid peptide pools demonstrating T cell immunity may compensate for the absence of humoral immunity under some circumstances.

One hundred and sixty-eight participants were sampled 1 to 2 months after their second vaccine dose using venous or DBS collection. The overall seropositivity following vaccination in this cohort was 54.8% (n = 92/168) and the median IgGAM ratio of seropositive individuals was 2.81 (positive defined as ratio > 1.0), with comparable results in groups sampled by DBS and venous blood (Fig. 1A). By comparison, overall seropositivity in 205 healthy participants from the COCO study was 100.0% with a median IgGAM ratio of 5.51. There was no significant difference in the percentage of individuals who were seropositive, or the magnitude of the antibody response between participants who had previously had SARS-CoV-2 infection and those who were infection naive (Fig. 1B). The Pfizer BioNTech 162b2 vaccine was associated with significantly greater seroconversion (65.7% vs. 48.0%, p = 0.03) and antibody responses (IgGAM ratio 3.73 vs. 2.39, p = 0.0003) in comparison to the AstraZeneca ChAdOx1 nCoV-19 vaccine (Fig. 1C). Serological responses from both vaccines display significant waning over time (2-way ANOVA, p = 0.001) but recipients of the Pfizer BioNTech vaccine displayed better preservation of antibody responses (2-way ANOVA, p < 0.0001) (Fig. 1D). Age did not significantly affect the magnitude of antibody responses or seroconversion following vaccination (Fig. 1E). Humoral responses following vaccination were variable amongst participants with a range of immunodeficiencies (Fig. 1F). As expected, serological responses were not detected in patients with X-linked agammaglobulinaemia (XLA); however, 52.2% of individuals with common variable immunodeficiency mounted a serological response to vaccination. Seropositivity was 75.0% in individuals with other primary antibody deficiencies (excluding XLA and CVID) and 100.0% in individuals with specific polysaccharide antibody deficiency. Variability was also observed amongst individuals with secondary immunodeficiencies regardless of aetiology. Thirty-one participants were sampled before and after their second immunisation permitting comparison of pre and post vaccine responses (Fig. 1G): seropositivity increased from 29.0% following the first dose to 61.2% following the second dose; both vaccines increased the magnitude of the antibody response.

T cell responses to vaccination were studied in 91 infection-naive individuals following their second vaccine dose and 12 individuals with a history of PCR + proven SARS-CoV-2 infection (Fig. 2A). In responses to a peptide pool derived from the SARS-CoV-2 spike protein, 46.2% of infection-naïve participants (n = 42/91) mounted a positive T cell response and a further 11.0% (n = 10/91) mounted a borderline response. In contrast, 91.7% (n = 11/12) of individuals with prior PCR positive infection mounted a positive T cell response to pooled spike peptides and 8.3% (n = 1/12) mounted a borderline response, as defined by this assay. In response to the SARS-CoV-2 nucleocapsid peptide pool, 8.8% (n = 8/91) of infection-naïve participants demonstrated a detectable T cell response and 1.1% (n = 1/91) mounted a borderline response compared to 66.7% (n = 8/12) and 8.3% (n = 1/12) respectively, in the prior-infection group. Individuals who had suffered previous PCR + SARS-CoV-2 infection mounted a significantly greater post-vaccination T cell response to the spike protein than those who were infection naive; no significant difference was observed for the nucleocapsid protein (Fig. 2A). All eight individuals with no prior history of PCR-proven SARS-CoV-2 infection who had positive T cell responses to nucleocapsid peptides also mounted above average responses to the spike peptide pools (> 100 spot forming units/10⁶ PBMC) suggesting a minority of individuals may have had asymptomatic infection, or mild symptomatic COVID-19 that was incorrectly attributed to other causes.

T cell responses directed towards the spike protein were comparable between the Pfizer BioNTech 162b2 and AstraZeneca ChAdOx1 nCoV-19 following the second dose of either vaccine (Fig. 2B) and persisted as time passed following vaccination (Fig. 2C). Participant age did not significantly influence the percentage of participants mounting a T cell response to the spike protein; a trend was observed towards greater magnitude responses in younger participants (Fig. 2D). A total of 57.9% of participants with common variable immunodeficiency disorder mounted a T cell response to the SARS-CoV-2 spike protein following vaccination with a wide range of responses detected in other primary and secondary immunodeficiencies (Fig. 2E). Both the Pfizer and AstraZeneca vaccines induced incremental T cell responses following the second vaccine doses in the majority of participants (Fig. 2F). A detectable T cell response was associated with significantly greater seropositivity following vaccination (79.5% vs. 53.8%, p = 0.009) and antibody responses of significantly greater magnitude (IgGAM ratio 3.08 vs. 2.14, p = 0.03) (Fig. 2G); however, no significant relationship was observed between the T cell response and peripheral CD19 + B cell numbers (Supplementary Fig. 2).

Participants that were seropositive post-vaccination had significantly greater serum IgM concentrations (Fig. 3A) and significantly larger numbers of CD19 + peripheral B cells (Fig. 3B) compared to those who were seronegative. There was no direct relationship between CD19 B cell numbers and IgM concentration (r² = 0.001, p = 0.53). Serum concentrations of both IgA and IgM were positively correlated with the magnitude of the antibody response following vaccination (Supplementary Fig. 1).

The functionality of antibodies was studied using in vitro, live virus neutralisation assays. Only 37% of participants with CVID (p = 0.0001) and 16% with primary antibody deficiency (p = 0.0003) displayed 50% viral neutralising activity or greater, compared to 100% of healthy controls (Fig. 4A). Neutralising capacity was not significantly impacted by prior SARS-CoV-2 infection status (Fig. 4B), type of vaccination received (Fig. 4C) or by participants’ age (Fig. 4D). The capacity of vaccine induced anti-spike IgG antibodies to bind the SARS-CoV-2 delta variant was significantly reduced compared to original Victoria strain (Normalised signal:noise ratio: 1.26 vs. 1.41, p < 0.0001). A total of 39.4% of individuals with detectable IgG responses against the Victoria strain fell below the threshold for positivity when the delta variant was substituted into the ELISA assay (Fig. 4E). Vaccine-induced IgG antibody binding was also significantly reduced to the Omicron variant of concern compared to original virus (Normalised signal:noise ratio 7.66 vs. 10.32, p < 0.0001) (Fig. 4F); however, no participants fell below the threshold for positivity.

Ten vaccine-breakthrough, PCR-proven infections have occurred this cohort up to 31/10/21 (median time from 2nd vaccine dose: 197 day); a further individual was infected between their first and second vaccine dose on the background of prior COVID-19 (Table 3). Eight participants reported new symptoms associated with acute COVID-19 above and beyond any chronic symptoms secondary to their immunodeficiency. A total of 90.0% (n = 9/10) of vaccine-breakthrough infections occurred in recipients of the AstraZeneca vaccine at a median interval of 120 days post second-dose and 70.0% occurred in individual who made no detectable humoral response to vaccination. One participant died of COVID-19, 3 months after receiving their second vaccine dose. This participant had a 31-year history of seropositive rheumatoid arthritis and secondary antibody deficiency (nadir IgG prior to immunoglobulin replacement: 0.97 g/L). Prior treatments for the underlying rheumatoid arthritis included oral corticosteroids, methotrexate, hydroxychloroquine and abatacept. At the time of SARS-CoV-2 infection, this participant was receiving daily oral prednisolone (9.5 mg/day) and had received rituximab 84 days prior to their first vaccine dose and 41 days after their second vaccine dose. No serological or cellular response to vaccination was detected in this participant at study enrolment.

Understanding the immunogenicity and efficacy of vaccinations is essential to guide global vaccination strategies and when to deploy non-pharmacological countermeasures to protect the immunologically vulnerable [19, 20]. Herein, we report the immunogenicity of the AstraZenca ChAdOx1 nCoV-19 and Pfizer BioNTech 162b2 vaccinations in patients with antibody deficiency, a cohort who have historically responded poorly to vaccinations [24‐26].

Overall, seropositivity following vaccination was 54.8%, significantly lower than healthy controls; comparable seropositivity was observed in the two largest subgroups of patients, common variable immunodeficiency (52.1%) and secondary immunodeficiency arising from haematological cause (55.8%). However, less than 10% of individuals with primary or secondary antibody deficiency made a neutralising antibody response equivalent to that of healthy controls following two doses of a SARS-CoV-2 vaccine. Furthermore, in individuals demonstrating a vaccine response, anti-spike IgG binding was significantly reduced against both the Delta and Omicron SARS-CoV-2 variants of concerns which are in widespread global circulation as of December 2021. Evidence suggests antibody binding is strongly associated with neutralising capacity [27]. These data suggest that vaccine-induced antibody responses are inadequate in the majority of individuals with antibody deficiency and additional strategies such as the use of prophylactic monoclonal antibodies to provide passive protection and antivirals are likely to be necessary to prevent severe disease.

T cell responses to vaccination displayed significant heterogeneity in this cohort as has been found in similar studies using identical laboratory methods [28]. The interferon-gamma release assay was originally validated to study T cell responses following natural infection, where it displays 98% sensitivity [29]. A total of 46.2% of infection-naïve COVAD participants and 91.7% of individuals with prior PCR-proven infection mounted a detectable T cell response following vaccination using this assay, compared to 54% of healthy individuals [30]. T cell responses in patients with evidence of previous SARS-CoV-2 infection were significantly greater than those detected following vaccination in infection-naive participants. The discordance between the detection of vaccine- and infection-induced T cell responses may arise from the duration, anatomy and magnitude of antigen exposure, differences in the immunological environment when antigen was presented, or assay-specific factors including differences in MHC restriction to the constituents of the peptide pool and the antigen-specific T cell repertoire in each circumstance.

Strong, polyfunctional T cell responses have previously been shown in an XLA patient following infection [5], and concordant with our study results, T cell response have been demonstrated in the majority of XLA patients following vaccination [17, 18]. However, across our antibody deficient cohort, there were no differences between the magnitude of the T cell response in individuals with or without a detectable peripheral B cell population. The clinical correlates of infection- or vaccine-induced T cell responses in patients with antibody deficiency, in particular, in the absence of humoral immunity remain uncertain. The absence of humoral immunity is a characteristic feature of individuals with prolonged SARS-CoV-2 infection [7]; however, robust T cell responses can limit the severity of disease in some individuals in the absence of humoral immunity as has been shown previously in patients with haematological malignancy [31]. Further studies are necessary to characterise the quality and breadth of T cell responses and its relationship to the development of effective humoral immunity following infection and vaccination in more detail.

With respect to vaccination strategies, we have shown that the Pfizer BioNTech 162b2 vaccine demonstrated significantly greater humoral immunogenicity in patients with antibody deficiency than the AstraZeneca ChAdOx1 nCoV-19 vaccination, a finding consistent with larger studies in healthy individuals [13, 32] and renal transplant recipients [28]. Furthermore, over 90% of vaccine breakthrough infections occurred in recipients of the AstraZeneca vaccine, 60.0% of whom made no serological response to the initial 2-dose vaccine schedule. Studies in the general population have suggested adenoviral-vectored vaccines demonstrate reduced vaccine-efficacy against severe disease when directly compared to mRNA vaccines (https://​www.​cdc.​gov/​mmwr/​volumes/​70/​wr/​mm7038e1.​htm). These observations support the use of mRNA vaccines in patients with antibody deficiency.

It could be argued that the deployment of a 3rd dose of vaccination in individuals that have not responded to a first dose is futile. However, we have found that serological and cellular responses to the SARS-CoV-2 spike protein were positively incremented by the second vaccine dose, in keeping with previous studies in patients with inborn errors of immunity [16] suggesting potential benefit from further doses. Our group have demonstrated the effectiveness of a 3rd primary immunisation in raising antibody levels against the delta and omicron variants of concern in a cohort of immunocompromised renal dialysis patients and provide preliminary evidence of the benefit of a heterologous vaccination strategy on serological responses to vaccination [33]. Further studies will be necessary to explore whether different vaccination combinations (homologous/heterologous) or dosing schedules may improve responses and efficacy in patients with primary and secondary humoral immunodeficiencies.

Existing studies have reported wide variation in the serological response to vaccination in patients with immunodeficiency: post-vaccine seroprevalence have ranged from 20.0 to 80.0% [14‐18]. The COVAD study is the largest reported study of patients with antibody deficiency and finds a seropositivity rate of 54.8% overall. At a cohort level, total B cell numbers were the principal determinate of serological response to vaccination, also in keeping with other SARS-CoV-2 vaccine studies [17]. Additional correlates of vaccine responsiveness remain to be elucidated: Salinas et al. demonstrated patients with CVID have a relative paucity of receptor-binding domain-specific, CD19⁺ CD24⁺ CD27⁺ B cells compared to healthy controls [18] and Hagin et al. were unable to demonstrate a common T-cell immunophenotype in vaccine non-responders beyond an inverted CD4/CD8 ratio [17]. Future work in COV-AD will employ detailed phenotypic and functional profiling to investigate potential correlates of vaccine immunogenicity and efficacy within this heterogeneous cohort.

This is a large study in a rare disease cohort and, although heterogeneous, we have had the opportunity to compare the immunogenicity of an mRNA and adenoviral-vectored vaccine in an immunodeficient cohort. To some extent, the generalisability of our study to the wider world is confounded by the extended UK vaccine schedule, which has not yet been widely adopted elsewhere. On the one hand, extension of the interval between first and second doses has been associated with greater neutralising antibody responses and enrichment of virus specific CD4 + T cells in healthy individuals [34], but shorter dosing intervals were associated with better humoral responses in a smaller study of patients vaccinated following treatment with B cell depleting agents [35]. There is an urgent need for further studies that explore how to maximise vaccine immunogenicity and efficacy in larger and heterogeneous cohorts of immune deficient patients.

In conclusion, we demonstrate profound impairment of serological responses following SARS-CoV-2 vaccination in patients with antibody deficiency and evidence of the superior immunogenicity of the Pfizer BioNTech 162b2 vaccine. These data highlight the ongoing risk of SARS-CoV-2 infection in antibody deficiency patients and should inform public health policy on vaccination strategies and other treatments to prevent morbidity and mortality.