Kinetics of immune responses elicited after three mRNA COVID-19 vaccine doses in predominantly antibody-deficient individuals

Mass vaccination campaigns reduced COVID-19 incidence and severity. Here, we evaluated the immune responses developed in SARS-CoV-2-uninfected patients with predominantly antibody-deficiencies (PAD) after three mRNA-1273 vaccine doses. PAD patients were classified based on their immunodeficiency: unclassified primary antibody-deficiency (unPAD, n = 9), common variable immunodeficiency (CVID, n = 12), combined immunodeficiency (CID, n = 1), and thymoma with immunodeficiency (TID, n = 1). unPAD patients and healthy controls (HCs, n = 10) developed similar vaccine-induced humoral responses after two doses. However, CVID patients showed reduced binding and neutralizing titers compared to HCs. Of interest, these PAD groups showed lower levels of Spike-specific IFN-γ-producing cells. CVID individuals also presented diminished CD8+T cells. CID and TID patients developed cellular but not humoral responses. Although the third vaccine dose boosted humoral responses in most PAD patients, it had limited effect on expanding cellular immunity. Vaccine-induced immune responses in PAD individuals are heterogeneous, and should be immunomonitored to define a personalized therapeutic strategies.


PAD individuals mount heterogeneous immune responses to COVID-19 vaccination
These responses should be monitored to know if additional vaccine doses are needed A booster dose increased humoral responses but had limited effect on nonresponders

INTRODUCTION
As of May 2022, the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has affected more than 528 million people worldwide, and reached an overall death toll of 6.28 million (https://covid19.who.int/). Fortunately, mass vaccination has drastically reduced the number of SARS-CoV-2-infected individuals that require hospitalization. 1,2 Human inborn errors of immunity (IEI) encompass a diverse set of diseases characterized by monogenic germline mutations that result in increased susceptibility to infection, malignant phenotypes, autoimmune, autoinflammatory and allergic diseases, 3,4 mainly because of an impaired immune system and specific immunosuppressive treatments (e.g. B cell-depleting agents). IEI patients demonstrate high heterogenicity in their phenotype and clinical manifestation, even in those individuals with identical genetic alterations. 4 Although these patients were initially considered at risk of severe COVID-19, SARS-CoV-2 seroprevalence and COVID-19-related fatality rate is similar to immunocompetent individuals, and most patients develop mild COVID-19. 5 However, those IEI individuals that require ICU admission or die because of COVID-19 illness are usually younger than those in the general population. 5 Severe COVID-19 in IEI individuals has also been associated with comorbidities, including autoimmune or inflammatory complications, lung disease, or higher proinflammatory responses. 5,6 Because IEI comprise a highly heterogeneous disease group, severity and fatality rate differ among pathologies. 7 Particularly, severe combined

Patient characteristics
Twenty-seven PAD adult patients under immunoglobulin replacement therapy (IRT) were initially included in the current study. According to 2019 ESID criteria, 14 patients were classified into four groups: combined immunodeficiency (CID) (1/27), common variable immunodeficiency (CVID) (14/27), thymoma with immunodeficiency (TID) (1/27), and unclassified primary antibody-deficiency (unPAD) (11/27). However, to better define vaccine-induced immune responses in PAD subjects, we excluded four patients (two unPAD, and two CVID individuals) that were diagnosed with SARS-CoV-2 infection either before vaccination or during the length of this study. Of note, anti-nucleocapsid protein (NP) antibodies were detected in one CVID patient after the third vaccine dose. Although this patient did not report signs of SARS-CoV-2 infection, we cannot exclude the possibility of an asymptomatic infection. Therefore, this time point was not considered in our analysis. As expected, none of the remaining 22 SARS-CoV-2-uninfected PAD patients described below showed antibodies against Spike (Figures 1 and S1), receptor-binding domain (RBD) ( Figure S2) before immunization, or against NP (data not shown) at any of the analyzed time points, confirming their seronegative status. Healthy controls (HCs, n = 10) also showed lack of reactivity against NP during the length of the current study.
Main characteristics of SARS-CoV-2-uninfected PAD patients are described in Tables 1 and 2. No severe adverse effects were reported after mRNA-1273 vaccination. A reactogenicity increase was observed after the administration of second vaccine dose, but not after the third immunization (Table S1).
Conversely, 78% (7/9) and 100% (9/9) of unPAD individuals seroconverted at w4 and w8, respectively Figure 1B). Anti-Spike IgG levels in unPAD responders at w8 (181 G 149 AU/mL) were similar to those observed in HCs (191 G 83 AU/mL, p>0.99), and significantly higher than in CVID responders (p = 0.04, Figure 1C). A significant decrease in antibody levels was observed from w8 to w24 in both HC (191 G 83 AU/mL versus 100 G 81 AU/mL, p = 0.006) and unPAD groups (181 G 149 AU/mL versus 34 G 23 AU/mL, p = 0.004, Figure 1B). Remarkably, and even though unPAD antibody levels were similar at w8 to the HC group ( Figure 1C), these individuals showed lower IgG levels at w24 (p = 0.02, Figure 1D). Despite that, only 11% (1/9) of unPAD patients were below the detection limit at w24. Administration of the third vaccine dose boosted IgG levels in all unPAD patients (p = 0.008) to similar levels than those observed at w8 in the HCs (p>0.99, Figure 1E). We were unable to detect antigen-specific IgG in patients with CID or TID at w8 or w24 ( Figures 1B and S2). However, the CID patient seroconverted after the third dose (60 AU/mL), showing the potential benefit of this additional shot.
Anti-Spike IgG responses correlated with RBD-specific IgG levels ( Figure S2D). Because RBD is considered a major target of neutralizing antibodies (NAbs), 15,16 we evaluated the capacity of our 23 SARS-CoV-2-uninfected PAD vaccinees to neutralize the ancestral SARS-CoV-2 Wuhan-Hu-1 (WH1), and two additional variants of concern (VoC): Delta (B.  Figure 2A). However, this neutralizing activity was hardly detected in CVID and unPAD individuals. According to our binding data (Figure 1C), the NAb titers against WH1 waned over time in HC (p = 0.002) and unPAD groups (p = 0.004), even though they remained stable in CVID patients (p = 0.1, Figure 2B). Despite that, HCs showed higher levels of neutralization at w24 against WH1 than CVID group (p = 0.001) and against Delta and Omicron than unPAD (p = 0.019, p = 0.035, respectively) and CVID groups (p = 0.001, p = 0.006, respectively) ( Figure 2C). Of interest, whereas NAb titers decreased in unPAD individuals, and were sustained in CVID patients over time, a transient increase was observed in HCs at w24. After that, NAbs decreased in the absence of an additional vaccine dose ( Figure 2D). Anti-Omicron neutralization titers also waned over time, but were still detected in 50% ( Figures 2B and 2F). In line with previous binding data ( Figure 1B), poor neutralizing activity was observed in the CID patient at w28, probably because of the presence of low IgG levels.
We then evaluated anti-Spike IgG avidity in a subset of PAD patients who responded to vaccination, and observed that IgG avidity significantly increased in both unPAD and HC groups over time ( Figure 3A). A similar positive trend was observed in CVID patients. Of interest, although unPAD patients had similar levels of anti-Spike IgG to HCs ( Figure 1C), they showed reduced IgG avidity at w8 (0.26 G 0.07 versus 0.36 G 0.06, p = 0.044, Figure 3B). Conversely, these patients developed higher IgG avidity than HCs at w24 (0.58 G 0.1 versus 0.48 G 0.06, p = 0.049, Figure 3C). After the third dose, antigen-specific IgG avidity in CVID and unPAD patients continued to increase, reaching similar values (0.65 G 0.09 versus 0.66 G 0.14, p>0.99, Figures 3A and 3D).

Induction of low levels of SARS-CoV-2-specific cellular responses in PAD patients after COVID-19 vaccination
Next, we evaluated vaccine-induced cellular responses against Spike by IFN-g ELISpot and flow cytometry at w0, w8, w24, and w28 ( Figure S3A). All HCs showed high levels of Spike-specific IFN-g-producing cells at iScience Article w8 (50 G 28 SFC/10 5 cells) and w28 (84 G 55 SFC/10 5 cells), indicating that these responses were stable and, in some cases, increased over time ( Figure 4A). Conversely, IFN-g responses were detected in 67% (8/12) of CVID patients at w8 (23 G 20 SFC/10 5 cells, p = 0.01, Figures 4A and 4B). Six months later, we observed IFN-g responses in only 33% (4/12) of CVID individuals. Of interest, the administration of the third vaccine dose restored the frequency of CVID patients that showed IFN-g-producing cells to those levels observed at w8 (15 G 19 SFC/10 5 cells, Figures 4A and 4B). Similarly, 67% (6/9) of unPAD patients developed IFN-g-producing cells after two doses (21 G 20 SFC/10 5 cells, p = 0.05, Figures 4A and 4B), and five of them remained detectable at w24 (24 G 20 SFC/10 5 cells). The third vaccine dose had no effect in this group (p>0.1, Figures 4A and 4B). Of note, the magnitude of these responses in CVID and unPAD groups was lower than HCs at w8 (p = 0.007, p = 0.014, respectively) and w24, compared with HCs at w28 (p = 0.001, p = 0.044, respectively). After the third vaccine boost (w28), these responses were still lower than those observed in HCs at w8 (p = 0.001, p = 0.025, respectively, Figures 4C-4E). Although both CID and TID patients did not develop antigen-specific IgG after two doses, these individuals showed detectable IFN-g-producing cells at w8 ( Figure 4A). Particularly, the CID patient showed a large IFN-g-producing response at w8, which progressively declined until w24. No boost was observed in these patients after the third vaccine dose (Figures 4A and4B).
Remarkably, the third vaccine dose did not significantly boost CD4+T cell responses in unPAD patients (p>0.99, Figure 5E). However, a frequency increase of CD69+CD137+CD4+T cells was observed in CVID patients after the third vaccine dose (p = 0.04, Figure 5B), reaching similar values than those detected in HCs at w8 (p>0.99, Figure 5E). Despite that, lower frequency of CD25+OX40+CD4+T cells in CVID patients at w28 was observed when compared to HCs (p = 0.003, Figure 5E). Of interest, 92% of CVID (11/12, p = 0.001) and 100% of unPAD patients (9/9, p = 0.004) elicited CD25+CD8+T cells at w8, which remained stable in all groups ( Figure 6A). Despite that, the magnitude of CD25+CD8+T cells in CVID patients was significantly lower than in HCs at w8 (p = 0.003, Figure 6B). The administration of the third vaccine dose did not boost CD8+T cells responses in CVID or unPAD individuals ( Figure 6A), which remained significantly lower than in HCs at w8 (p = 0.002, Figure 6C). Similar results were observed in CID and TID patients ( Figure 6A).

DISCUSSION
Although the three-dose COVID-19 mRNA vaccine regimen has shown increased effectiveness over the two original doses in healthy individuals, 17 its effect remains largely unknown in PAD individuals. Here, iScience Article we characterized the immune responses elicited in 23 SARS-CoV-2-uninfected PAD patients after receiving three mRNA-1273 vaccine doses. Although our PAD cohort is mainly composed of unPAD and CVID patients, it also includes one CID and TID patient, which might be interesting because of the limited available information about how these individuals respond to COVID-19 vaccination. 18,19 According to previous reports, [10][11][12][18][19][20][21] our results showed that mRNA-1273 immunization was safe, and most patients developed Spike-specific immune responses. However, the heterogenicity of PAD disorders is reflected in the distinct immune responses elicited after vaccination. For example, although unPAD patients developed vaccineinduced humoral responses with similar kinetics to those elicited in HCs, they showed a faster decline of NAb titers over time, requiring a third vaccine dose to develop NAbs against Omicron, which was achieved iScience Article in HCs after two doses. Thus, COVID-19 vaccination efficacy may wane earlier in unPAD patients than in the general population. In contrast, only 67% of CVID individuals seroconverted at w8, and developed lower levels of anti-Spike IgG and NAb titers against all VoC than HC and unPAD groups. Of interest, antibody levels remained stable over time in most CVID responders. Although vaccine-induced IgA responses were detected in most unPAD and HC individuals, only one CVID patient elicited anti-Spike IgA. These results were not surprising, because most CVID individuals, including the ones in our cohort, demonstrate impaired IgA responses. 22 Notably, we detected low levels of anti-Spike IgM in two CVID patients who had not developed IgG or IgA responses. One of them showed low NAb titers against WH1 and Delta. Differences observed in the humoral responses among groups may be because of the enrollment of different B cell subsets. Particularly, the generation of antibody-secreting cells (ASCs) with distinctive half-life (short to intermediate versus long) could explain why antibody levels waned differently in PAD and HC groups. It has been previously described that the frequency of T and B cell subsets in unPAD individuals is comparable to HCs, 23 which could explain the similar kinetics observed between these two groups. Remarkably, although humoral responses in the CVID group were more heterogeneous, the stability of the antibody levels observed in responder individuals suggests that long-lived ASCs might be generated in a fraction of them. These results are noteworthy because it has been described that CVID individuals demonstrate a dysregulated B cell compartment, characterized by a reduction of class-switched memory B cells, and an increased frequency of atypical memory B cells (CD19+CD27-CD21-IgM-IgD-), which encompasses most Spike-specific memory B cells after COVID-19 vaccination. 11 Although the origin of these cells remains unclear, it has been postulated that they might derive from an extrafollicular B cell response, a T-independent response, or an early germinal center (GC) reaction. 11,24 Because the avidity of anti-Spike IgG responses gradually increased after the second vaccine dose in CVID responders, our results support a potential involvement of GCs and antigen-specific B cell selection in a subset of CVID patients, that could iScience Article contribute to the generation of long-lived ASCs. In line with that, COVID-19 vaccination induces a persistent GC reaction in healthy individuals, 25 and it is indeed possible to detect GCs and somatic hypermutations in CVID patients. 26,27 However, GCs in CVID individuals might be dysfunctional, which could explain the increased frequency of atypical memory B cells. 27 In addition to characterize the humoral responses in our PAD cohort, we defined the vaccine-induced cellular responses using two assays: IFN-g ELISpot, and the detection of activation markers by flow cytometry. 28 Both assays showed that cellular responses were sustained in unPAD and HC groups, and that the third vaccine dose had no effect on expanding the magnitude of previously-generated responses in unPAD individuals. Contrarily, although CD4+T cell responses decreased in CVID patients over time, these responses were boosted after the third immunization. Compared to HCs, CVID patients developed lower frequency of CD8+T cell responses that remained stable over time. Intriguingly, we observed a discrepancy in the magnitude of antigen-specific cellular responses detected by both techniques. We identified  iScience Article significantly lower levels of IFN-g-secreting T cells in CVID and unPAD patients compared to HCs after two and three vaccine doses. These differences were not detected by flow cytometry, except in OX40+CD25+CD4+T cells of CVID patients at w28, which were lower than HCs. Our results may suggest that although T cell responses could be generated after vaccination in unPAD and CVID individuals, their function (i.e. IFN-g production) might be impaired. Accordingly, Fernandez et al. described a lower proportion of IFN-g responses after stimulation with Spike-derived peptides in COVID-19 vaccinated CVID patients, 11 which could be a general particularity of CVID individuals. 29 Conversely, Hagin et al. 10 stated that vaccine-induced cellular responses in IEI patients and HCs were similar. Although IEI patients in this latter study and ours are different, ELISpot data obtained in the CVID group from Hagin et al., and the one presented herein seem comparable. It is possible that the discrepancy lies in that our HCs showed greater levels of IFN-g-producing responses than those described in Hagin et al. 10 Besides unPAD and CVID patients, we also analyzed the vaccine-induced immune responses in one CID and one TID patients. Although none of them showed anti-Spike antibodies, both patients developed Spike-specific cellular responses after two vaccine doses. Similarly, we identified detectable T cell responses in three of five CVID patients who did not elicit humoral responses. It has been previously described that individuals who do not develop humoral responses after vaccination (e.g. XLA individuals 11 or patients treated with anti-CD20 antibodies 30 ) can elicit antigen-specific cellular responses. Of interest, although the CID patient seroconverted after receiving a third dose, showing Spike-specific iScience Article IgG and anti-WH1 NAbs, we were unable to detect Spike-specific immune responses in two CVID patients after three COVID-19 immunizations. Immunosuppressive therapies have been associated with the development of poor humoral responses in patients with multiple sclerosis, 31 limiting the efficacy of COVID-19 mRNA vaccines. 32,33 However, of the two CVID subjects who did not elicit vaccine-induced immune responses, only one received immunosuppressive treatment. Thus, because of the small sample size, we cannot conclude the impact of these therapies in halting vaccine-induced immunity in our study.
Our study highlights the distinct response to COVID-19 vaccination elicited in PAD individuals. Although most individuals mount Spike-specific immune responses, there is a fraction of subjects that remained non-responders even after three vaccine doses. Therefore, immunomonitoring of these patients could provide insights about their immune status and the need of additional vaccine doses or other prophylactic approaches.

Limitations of the study
Although our results regarding immune responses in PAD patients are similar to other published work, [10][11][12]18,20,21 the heterogenicity of our cohort and its small size, in addition to other several factors (i.e. patient heterogenicity, methodology, time points, and administered vaccine), hinder direct comparison among studies.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

ACKNOWLEDGMENTS
We would like to thank all the patients who participated in this study, and the people that are supporting our research through the ''#YomeCorono'' crowdfunding initiative. This work was funded by the Departament de Salut of the Generalitat de Catalunya (SLD0016 to JB and Grant SLD015 to JC), the Carlos III Health Institute (PI17/01518 to JB and PI18/01332 to JC), Ministerio de Ciencia, Innovació n y Universidades (MI-CINN, PID2020-119710RB-I00 to CB), CERCA Program (2017SGR 252), and crowdfunding initiatives #Yomecorono, BonPreu/Esclat, and Correos. CAN was supported by a predoctoral grant from Generalitat deCatalunya and Fons Social Europeu (2020 FI_B_0742). EP was supported by a doctoral grant from National Agency for Research and Development of Chile (ANID: 72180406). This study was also supported by CIBER -Consorcio Centro deInvestigació n Biomé dica en Red (CB 2021), Carlos III Health Institute, Ministerio de Ciencia e Innovació n and Unió n Europea -NextGenerationEU. Funders had no role in study design, data analysis, decision to publish, or manuscript preparation. Graphical abstract and drawing in Figure 1 and S3 were created with BioRender.com.

OPEN ACCESS
iScience Article RESOURCE AVAILABILITY

Lead contact
Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Jorge Carrillo (jcarrillo@irsicaixa.es).

Materials availability
This study did not generate new unique reagents.
Data and code availability d All data reported in this article will be shared by the lead contact on request.
d This article does not report original code.
d Any additional information required to reanalyze the data reported in this article is available from the lead contact on request.

Study overview and human subjects
A prospective observational cohort-comparative study was conducted at the Hospital Universitari Germans