Introduction
Common variable immunodeficiency disorders (CVID) form a heterogeneous group of immunodeficiencies likely encompassing several different etiologies [
1]. Humoral immunity is affected in all patients, whereas defects in cellular immunity vary [
2]. As pointed out in the
International Consensus Document (ICON) on CVID, although there is considerable knowledge regarding the immunological response to a few vaccines, for most of them, information is scarce or lacking [
3]. To this end, a rather unique feature of CVID is that vaccination is part of the diagnostic strategy. Common vaccines used include tetanus and diphtheria toxoids and antigens from
Haemophilus influenza type B, and
Streptococcus pneumoniae. The toxoids are proteins, whereas the bacterial vaccines are either polysaccharide-based or manufactured as polysaccharides conjugated to proteins. The majority of patients with CVID respond poorly to pure polysaccharide-vaccines but may mount weak responses to protein and conjugate vaccines [
4]. While such vaccinations serve two purposes, diagnostic and prophylactic, the response may be difficult to interpret due to previous exposure to these bacteria in the past as well as previous vaccinations.
An important consideration is also the fact that CVID patients regularly are on immunoglobulin replacement therapy (IGRT). This means that for almost all infectious agents, there is a background of specific antibody (Ab) levels in the commercial immunoglobulin preparations, which makes it almost impossible to study primary immune responses without confounding contributions from the IGRT. Given these known problems with available vaccines, all SARS-CoV-2 vaccines are exceptional since the antigens used are new. Furthermore, it is known from the analysis of anti-spike IgG antibodies in the immunoglobulin preparations used at the time of the study that there were no detectable levels present [
5‐
7].
Patients with CVID exhibit a variable susceptibility to Coronavirus disease 2019 (COVID-19), where those with underlying lung diseases, such as bronchiectasis and interstitial lung disease, have the highest risk [
8]. In contrast, there are several case series that report a more moderate risk [
9,
10]. However, the question of immune responses following COVID-19 vaccination in patients with CVID remains largely unanswered. We recently completed a large clinical trial with a prospective design, in which 449 patients with different immunocompromising disorders and 90 controls were vaccinated with the Pfizer-BioNTech BNT162b2 mRNA vaccine. In total, 50 patients with CVID were included in the study and 68.3% seroconverted [
11], which is in line with a previous study from Israel (
n=12 patients with CVID) [
12]. Notably, a more recent and larger study from Italy could only detect spike-specific IgG in 20% of seronegative COVID-19 vaccinated patients with CVID (
n=41). It was found that patients with a previous history of COVID-19 infection had higher titers after vaccination, suggesting that a natural infection may cause a more efficient priming of the immune system than the first vaccine dose [
13]. Another study from Italy with 14 patients with CVID showed that most patients seroconverted after COVID-19-vaccination, but with lower titers than in healthy controls (HC) [
14]. Hence, previous studies illustrate the fact that vaccine responses in patients with CVID are highly variable. Importantly, there is limited knowledge about the relation between immunological parameters and vaccine responses. Thus, we hypothesized that specific cellular subsets that are used to define CVID subgroups, including the EUROclass consensus definition, could correlate with outcome of COVID-19 vaccine-induced immune responses in patients with CVID [
15]. To test this, we studied clinical and immunological data from patients with CVID, who were included in our recently completed COVID-19 vaccine trial [
11], and correlated the results with seroconversion rates and spike IgG titers at day 35 (2 weeks after the second dose). Notably, the present study revealed that patients with an increased frequency of CD21
low B-cells had a significantly higher risk of seroconversion failure.
Discussion
The main finding in this report was the observation that a high frequency of the CD21
low B-cell subset significantly correlated with a poor primary humoral immune response. While this marker previously has been used to identify subsets of patients with CVID [
15,
20‐
22], to the best of our knowledge, this is the first time that the frequency of CD21
low B-cells was found to significantly correlate with the capacity to mount a humoral immune response to a true neoantigen.
Additionally, the EUROclass consensus definition clearly defined a group of responders to vaccination. Thus, all patients with CVID having detectable levels of B-cells switched memory B-cells >2% and a normal percentage of CD21low B-cells seroconverted in this study, whereas those patients with either switched memory B-cells <2% or an increased frequency of CD21low B-cells (>10%) exhibited a poor vaccine response. This finding confirms that the EUROclass definition can be useful in clinical practice for identifying patients at risk for a poor response to a protein antigen expressed in the form of an mRNA vaccine.
Next, the CD4
+ T-cell phenotypic profiles of the patients were analyzed and specific cellular subsets were related to the vaccine response. Notably, reduced levels of CD4
+, but not CD8
+, T-cells were associated with a poor vaccine response. In the CD4
+ T-cell pool, mainly elevated levels of naïve CD4
+ T-cells (CD45RA+, CCR7+) corresponded to an optimal vaccine response. Interestingly, patients with a poor vaccine response had lower levels of central memory (CM) cells, particularly CM Th2 and Th17 cells (Figure
3F). These findings suggest that the subgroup of CVID-patients with reduced levels of specific T-cell subsets, previously connected to autoimmune complications, also may suffer from a poor vaccine response to mRNA vaccination [
21,
23].
We also assessed the cellular immune response in a subset of patients with CVID, using an ELIspot IFN-γ release assay as well as a spike-specific AIM CD4 T-cell assay (FACS). Overall, our results show that patients with CVID exhibited poor cellular immunity, with only a fraction of patients reaching the same response as HC. This indicates that patients with CVID have a combined deficiency in both the cellular and humoral immune response to mRNA vaccination.
The observed increased frequency of the CD21
low B-cells among nonresponders to COVID-19 vaccination is of interest since this population has been implicated in immune responses. Thus, CD21
low B-cells have been suggested to represent a distinct population, which is transiently induced 2 to 4 weeks after immunization and recently egressed from germinal centers [
24]. While carrying CD27, they are not classical memory cells and express high levels of the transcription factor T-bet [
24‐
26]. T-bet–expressing cells are also induced by chronic viral infections, including HIV and hepatitis C [
27,
28]. CD21
low B-cells are also implicated in autoimmunity, such as in systemic lupus erythematosus (SLE)[
29,
30]. This raises the question whether CD21
low B-cells are
directly implicated in the poor vaccine response or if they merely serve as a
marker of a subnormal response.
Since the frequency, but not the absolute number, of CD21
low B-cells correlates with vaccine-induced immunity (Suppl fig.
1), we favor the idea that CD21
low is a useful marker but that this subset only indirectly contributes to poor immunity. In line with this reasoning, it was previously reported that the increased proportion of CD21
low cells in CVID is the indirect result of reduced naïve B-cell numbers [
31]. In fact, we also observed that reduced numbers of naïve B-cells were associated with a poor vaccine response (Suppl figure
1). Furthermore, it was recently found that CD21
low cells can serve as competent antigen-presenting cells in an allogeneic co-culture system [
32], suggesting that an elevated proportion of these cells is not directly implicated in poor vaccine-mediated immunity.
Germinal center B-cells are necessary for the induction of potent humoral immunity against SARS CoV-2 mRNA vaccines [
33]. Interestingly, patients with CVID may have dysfunctional and irregular germinal centers, which are associated with an increased percentage of CD21
low B-cells in the circulation [
34]. In addition, patients with CVID exhibit profound defects in the maturation of pre germinal center B-cells [
35]. Thus, it is possible that an increased percentage of CD21
low B-cells could serve as a marker for dysfunctional germinal center reactions in CVID patients and effectively predict nonresponders to mRNA vaccination.
This study has several strengths. First, the data were collected from a bona fide prospective clinical trial with the scientific and regulatory rigor inherent to this design, which contributed to high data quality, low dropout rates and minimal risk of selection bias. Second, by studying the immune reaction to a neoantigen and thereby avoiding problems with pre-existing immunity, this study, as well as others studying SARS-CoV-2 responses, is likely unique [
36]. Thus, while the effect of influenzae vaccination has been investigated in patients with CVID [
37], pre-existing immunity makes it difficult to differentiate between primary and recall responses. Finally, the study is one of the largest studies on patients with CVID in relation to COVID-19 vaccination, which allowed subgroup analyses with enough power to obtain statistical significance and clinically meaningful outcomes. However, certain subgroups were too small to allow for robust analyses. Thus, the low immune responses noted in two patients with TNFRSF13B-mutations will need confirmation in larger follow-up studies with more individuals, including both heterozygous and homozygous carriers.
Despite these strengths, several weaknesses need to be acknowledged. First, the T- and B-cell data was collected up to 4 years prior to the vaccine study, which could lead to lower precision in the analyses. However, the immunological aberrations in patients with CVID do not fluctuate much but rather remain over time [
15]. For example, when checking our own records, where available, most individuals with an increased frequency of CD21
low B-cells sustain this immunological phenotype over at least 5 years (data not shown). Thus, despite this potential confounder, our data are valid and clinically relevant. Further, due to logistical constraints, we could only collect PBMCs from a subset of patients, which makes it difficult to draw strong conclusions on T-cell immunity from subgroup analyses. Another limitation is the lack of mechanistic insight for the observed influence of the percentage of CD21
low-B cells and how this can lead to a poor immune response. While there has been considerable development in the understanding of this subset, many aspects remain elusive [
26]. At this point, we favor the idea that an elevated percentage of CD21
low B-cells is an indirect marker for disturbed B-cell–mediated immunity in patients with CVID.
A potentially interesting observation is the fact that in SLE, the CD21
low B-cell population has been suggested to represent an innate cell type related to TLR7 [
29,
38]. Given the recent finding that 1.8% of men below the age of 60 who develop life-threatening COVID-19 have a defective TLR7 [
39], there may also be other connections of importance for how CD21
low B-cells interact with SARS-CoV-2 vaccines. CD21
low B-cells have been shown to mount impaired calcium mobilization and NF-κB activation after B-cell receptor stimulation. This response could potentially be dysfunctional in a primary immune response against viral infection [
40], but given the fact that the percentage, but not the absolute number, of CD21
low B-cells is increased, we do not favor this idea. Thus, the mechanistic understanding of vaccine responses in patients with CVID requires further studies, where especially the role of different T-cell subsets for the development of the humoral immunity is taken into account.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.