**4. Discussion**

The *IKZF1* gene (OMIM: #603023) encodes the IKAROS zinc-finger transcription factor. It is located on chromosome 7 (7q12.2) and has eight exons. The *IKZF1* gene can produce different isoforms due to alternative splicing, but the main "DNA-binding" form of IKAROS is isoform 1, whose structure is composed of an N-terminal DNA-binding domain, which is made up of four zinc-finger motifs (ZF1–ZF4), a central activation/repression domain, and a C-terminal dimerization domain (ZF5–ZF6) (Figure 3). IKAROS binds as homodimers or heterodimers to pericentric heterochromatin regions, promoting the expression of target

genes that play important roles in lymphocyte development, differentiation and function, and myeloid cell development [13].

**Figure 3.** Schematic representation of the structure of human IKAROS protein encoded by the *IKZF1* gene. The DNA-binding domain consists of four zinc-finger motifs (ZF1–ZF4) and the dimerization domain of two (ZF5–ZF6). The p.Arg162Gln variant found in our patient is highlighted in red and affects the ZF2 domain. Other amino acid positions of IKAROS variants identified in CVID patients are indicated. ZF: zinc-finger.

The first *IKZF1* mutations that were linked to human pathologies were somatic and associated with a worse prognosis in B-cell acute lymphoblastic leukemia (B-ALL). However, more recently, heterozygous germline mutations have also been related to B-ALL development [14]. On the other hand, germline heterozygous mutations in the *IKZF1* gene have been detected in patients with primary immune deficiency or inborn errors of immunity. The first de novo heterozygous *missense* mutation in the *IKZF1* gene was described by Goldman et al. in 2012, in a 33-week preterm infant with pancytopenia and loss of B cells, which shared phenotypic similarities with the *IKZF1-null* mouse animal model and that affected the ZF4 domain of IKAROS [15].

Subsequently, various types of germline mutations in the *IKZF1* gene have been identified in patients with a CVID-like phenotype, including missense mutations (all of which have been classified as pathogenic or probably pathogenic), frame-shift mutations, nonsense mutations, and intragenic deletions of *IKZF1* [15–22] (Figure 3). These mutations can act in different ways, giving rise to haploinsufficiency (HI), dominant-negative (DN), or dimerization-defective (DD) effects. However, most patients carrying germline heterozygous missense mutations that act in a HI manner and affect the DNA binding domain manifest CVID when they have symptoms [23,24].

The heterozygous missense R162Q variant of the *IKZF1* gene present in our patient was discovered for the first time in 2016 by Kuehn et al., in a European family composed of three generations of affected individuals, where the variant cosegregated with the disease in seven of the twelve members of the family [17]. Later, in 2017, Hoshino et al. also reported the R162Q variant in three generations of a family with two affected individuals from three carriers [20]. Both studies revealed an autosomal dominant inheritance with incomplete but high penetrance. However, in our patient, it appears for the first time as a de novo mutation.

The pathogenicity of the variant was demonstrated by Kuehn et al. [17], through functional analysis and subsequently corroborated by Hoshino et al. [20]. Flow cytometry data of the Kuehn et al. study showed that the amount of IKAROS in the T and B cells of the patient with the R162Q mutation in *IKZF1* was equal to controls. To determine the effect of the R162Q mutation, they used transfected cells. They noticed that the R162Q variant produced stable proteins, which could dimerize with wild-type (WT) protein forms and migrate to the nucleus, but they were not capable of binding to the pericentromeric target DNA region. To mimic the heterozygous state, Kuehn et al. used transfected cells with vectors expressing 100% WT IKAROS and transfected cells with vectors expressing 50% WT and 50% mutant IKAROS. They found that DNA binding was reduced by 38–74% in cells transfected with 50% mutant vectors, compared to cells transfected with 100% WT vectors. The heterozygous state and the haploinsufficiency mechanism explain the high but incomplete penetrance of this mutation. Finally, based on our in silico analysis using MutationTaster and other prediction methods, the R162Q variant causes a deleterious effect on the mutated IKAROS protein, supporting the previous results.

In the Kuehn study, individuals with the heterozygous R162Q variant were characterized by recurrent infections, mainly respiratory tract infections or otitis, and hypogammaglobulinemia. The age onset was before the age of ten in most of them. One of the patients died of pneumonia at the age of 74 and another patient developed B-ALL and died from a relapse at the age of 5. All of the patients presented a reduction in B cells and most of them a reversed CD4/CD8 T-cell ratio [17]. In the Hoshino study, one of the symptomatic patients, at seven years of age, had bacterial pneumonia with agammaglobulinemia and IgA vasculitis and analytically low levels of B cells, normal NK cells, and low CD8+ T cells. The other symptomatic patient had a history of thrombocytopenic thrombotic purpura (ITP) without dysgammaglobulinemia, with normal levels of B and NK cells, an increase in T follicular helper cells, and a reduction in CD4/CD8 T-cell ratio [20]. Our patient was diagnosed at an older age compared to the patients previously described with this variant. Analytically, he is also characterized by recurrent infections, a marked reduction in B cells, a reversed CD4/CD8 T-cell ratio, and very low immunoglobulins levels at the diagnosis, except for IgM.

The humoral response to the complete vaccination schedule with AstraZeneca and the booster with Moderna in our patient is considered very low. The study carried out by Atmar et al. showed circulating antibody levels against the Spike protein of SARS-CoV-2 after 1, 15, and 29 days of the booster dose in healthy individuals. This study showed that there are no major differences between homologous booster vaccination and heterologous booster. The normal humoral response after the regimen of two doses of Janssen (viral vector) and the booster of Moderna is estimated at 4560 BAU/mL of circulating antibodies, which is much higher than our patient [25]. In the same way, Munro et al. published an article that compared the immunogenicity of seven booster vaccines after the two doses from AstraZeneca and Pfizer-BioNTech (COV-BOOST) by means of, among others, the levels of anti-spike IgG measured by ELISA [26]. In the 98 healthy cases studied that were vaccinated with two doses of AstraZeneca and subsequent booster with Moderna, a mean of 31,111 ELISA laboratory units per milliliter (ELU/mL) was found at day 29 post-boost. According to the WHO international standard for COVID-19 serological tests to harmonize anti-spike SARS-CoV-2 assays [27], taking as reference the ELISA technique, BAU/mL = 0.142 × AU/mL. Thus, the results obtained by Munro are similar to those obtained by Atmar (4417 BAU/mL) and are higher compared to our patient. On the other hand, Amodio et al. conducted the first study in a cohort of 21 European patients with CVID. They evaluated the humoral response 21 days after the first dose and 7 days after the second dose with the Pfizer-BioNTech vaccine and found that patients with CVID could generate an antibody response, but at a lower magnitude than healthy controls [8]. It should be noted that some of these patients had a genetic alteration, but none with a mutation in the *IKZF1* gene. In the same way, Hagin et al. observed a positive response, but at lower levels than healthy controls in a cohort of 13 CVID patients treated with immunoglobulins, measuring the SARS-CoV-2 anti-protein S antibody response 2 weeks after the second dose of the Pfizer-BioNTech vaccine [9]. In addition, in the group of patients with CVID, they found a greater response in patients under 50 years of age. Bergman et al. showed that CVID patients with a lower percentage of switched-memory B cells or an increased percentage of CD21*low* B cells have a poor response to the Pfizer-BioNTech mRNA vaccine [28]. These immunophenotypic characteristics and results were obtained in our patient, although after the booster dose of the SARS-CoV-2 vaccine, he generated

a small increase in switched-memory B cells, coinciding with his greater production of antibodies after vaccination.

The T-cell response after the booster dose in our patient, despite being positive, is far from those found in the healthy population used as a control group in other studies and is similar to that of vulnerable and immunocompromised patients in whom the SARS-CoV-2-specific IFN-γ were quantified using the same commercial kits (Euroimmun) and techniques as in our case. Schwarz et al. quantified SARS-CoV-2-specific IFN-γ 7 days after the second dose of the Pfizer-BioNTech vaccine, finding that the mean concentration in people >70 years old was 707.3 mIU/mL, while in healthy young adults it was 2184 mIU/mL [29]. Ruether et al. compared the levels of SARS-CoV-2-specific IFN-γ 29 days after the second dose of the mRNA vaccine (Pfizer-BioNTech or Moderna) or vectorbased vaccine (AstraZeneca) in 3 groups: 19 healthy controls (mean around 1000 mIU/mL), 26 patients with liver cirrhosis (mean between 100 and 200 mIU/mL or borderline), and 82 liver transplant patients (mean below 100 mIU/mL or negative response) [30]. Our patient presents a loss of cellular response 3 months after the second dose with the AstraZeneca vaccine, and although the response is good after the booster dose with Moderna, it is presumably lower than in the healthy population and will be lost over time. This hypothesis may be demonstrated in subsequent controls on the patient and corroborated by different studies that are still underway.

The molecular heterogeneity of patients with CVID, the use of different techniques to measure the immune response to SARS-CoV-2 vaccines, and the use of different vaccines and vaccine regimens make it difficult to compare the results between studies, but reinforces the importance of vaccinating this vulnerable population, emphasizing the need to measure not only the humoral response but also the cellular response in these patients [7–9].
