Deficiency of somatic hypermutation of the antibody light chain is associated with increased frequency of severe respiratory tract infection in common variable immunodeficiency

Common variable immunodeficiency (CVID) is a heterogeneous primary immunodeficiency characterized by acquired quantitative IgA and IgG deficiency and recurrent infections.^1-4  IgM levels are normal or reduced. The most common infections involve the upper and lower respiratory tract followed by gastrointestinal infections, but susceptibility, severity, and location of infections vary considerably among patients.^5-8  The most important features of the disease are recurrent bacterial pneumonias often leading to bronchiectasis, fibrosis, and eventually to respiratory insufficiency, the leading cause of death among patients with CVID. Autoimmune and allergic manifestations are common.^5,8  Cancer occurs frequently in CVID.^5,9  In 176 patients with CVID, Mellemkjaer et al¹⁰  found a 10-fold increased incidence of malignant lymphoma and stomach cancer.

Several attempts have been made to subgroup patients with CVID based on clinical or immunologic parameters,^11,12  but no generally accepted criteria exist. Laboratory evidence for isolated B- and T-cell malfunction has been presented, and pathogenetic mechanisms may vary between patients.^13-24  Circulating B cells are normal or reduced in numbers, but rarely absent. Hypogammaglobulinemia results from failure to produce sufficient numbers of functional plasma cells. Recent data suggest that circulating memory B cells may be an important source of plasma cells.²⁵  It is therefore interesting that the generation of memory phenotype (CD27⁺) B cells is reduced in most patients.^11,14,26,27  The plasma level of IgG is a poor predictor of disease severity because patients with slightly decreased levels of IgG may be highly susceptible to infections and benefit from immunoglobulin substitution therapy.²⁸  Thus, reliable prognostic criteria are currently lacking and qualitative aspects of the produced antibodies should be considered.

Impaired somatic hypermutation (SHM) of IgG was described in a minor subpopulation (23%) of CVID patients using hot-spot–specific restriction enzymes after polymerase chain reaction (PCR) amplification of V3-23-IgG heavy-chain transcripts.^29,30  Thus, the normal coupling of isotype switching and SHM may fail in some patients. SHM is essential for the generation of normal long-lived plasma cells and memory B cells secreting and encoding high-affinity antibodies, respectively. SHM normally occurs in the germinal centers, where high-affinity variants are selected in a process involving antigen-coated follicular dendritic cells and T cells. By this process affinity may increase 10- to 1000-fold.^31-36  The finding that IgG-secreting cells in some patients lack somatic mutations indicates that the IgG produced in these patients is of reduced quality with low affinity for the eliciting antigens. However, affinity maturation by SHM does not depend on isotype switching and may occur in B cells producing IgM.^37-39  The failure to use affinity-maturated antibody within all isotypes might be a better predictor of a qualitative immunodeficiency than the mutations of isotype-switched genes.

In this study, SHM of the κ light chain was investigated in patients with CVID as a measure of affinity maturation irrespective of isotype and correlated to clinical features of the disease. We developed a quantitative restriction enzyme-based hot-spot mutation assay (IgκREHMA). Using this technique, impaired SHM was detected in 77% of CVID patients and shown to correlate with frequency of severe respiratory tract infection (SRTI). Quantification of the level of SHM may turn out to be an important parameter in the clinical management of patients with CVID and improve our understanding of the cellular mechanisms underlying the disease.

Thirty-one patients (13 men and 18 women) with CVID were included. All were referred to the Department of Clinical Immunology for evaluation of their immunodeficiency either during the study period (n = 26) or previously (n = 5) if frozen mononuclear cells were available. The inclusion criteria were primary hypogammaglobulinemia with a decrease of IgG to less than 5 g/L with IgA less than 0.7 g/L or IgM less than 0.5 g/L and fulfilling the following criteria: (1) onset of immunodeficiency later than 2 years of age and (2) exclusion of defined causes of hypogammaglobulinemia. Heparinized blood samples were collected and mononuclear cells isolated by density gradient centrifugation, frozen, and stored in liquid nitrogen until use. Samples were available from the time of diagnosis and prior to IgG substitution therapy for 17 patients; 14 patients were diagnosed several years earlier (median, 11 years; range, 3-26 years) and all were in substitution therapy at the time of the sampling. Frozen cell samples available from other time points were included to study the course of SHM.

The median age of the patients at the time of diagnosis was 35 years (range, 8-73 years) with histories of increased frequency of infections for 5 years (range, 0-48 years) prior to diagnosis. At the time of the first sample tested for SHM the median age was 41 years (range, 8-75 years). As controls, blood was obtained from 53 healthy adults after informed consent was given (9 men and 44 women; median age, 41 years; range, 23-65 years). Three patients were younger (ages 8, 14, and 22 years) and 3 older (ages 66, 74, and 75 years) than the controls.

Twenty cord blood samples were included as negative controls, donated after informed consent was obtained from the delivering women. The study was approved by the Ethics Committee of Copenhagen and Frederiksberg Counties.

cDNA synthesis. Mononuclear cells were thawed, and mRNA was isolated from 3 to 6 × 10⁶ cells using the Dynabeads mRNA DIRECT kit (Dynal, Oslo, Norway) as instructed by the manufacturer. cDNA synthesis was performed by the Perkin Elmer reverse transcription PCR kit using an oligo-dT reverse primer (DNA Technology, Aarhus, Denmark) as instructed by the manufacturer.

PCR. PCR was performed using fluorochrome-coupled primers specific for the signal peptide region and the third framework region (FR3), respectively, of the κ light-chain gene V_κA27 (IGKV3-20 in ImMunoGenetics (IMGT) nomenclature). The primers were: TET-coupled (tetrachloro-6-carboxyfluorescein) 5′-CAGAGGGAACCATGGAAA-3′ for the signal peptide region and FAM-coupled (6-carboxyfluorescein) 5′-CCACTGCCACTGAACCTGT-3′ for the FR3 region (Oswel DNA Services, Hampshire, United Kingdom; Figure 1). The PCR conditions were: primers 0.1 μM, MgCl₂ 1.5 μM, × 1 PCR buffer (50 mM KCL, 10 mM Tris (tris(hydroxymethyl)aminomethane)–HCl), platinum Taq-polymerase 1.25 U (Invitrogen, Life Technology, Taastrup, Denmark) in a reaction volume of 50 μL. Amplification conditions were 94°C for 2 minutes, 25 cycles of 94°C for 1 minute, 58°C for 1 minute, and 72°C for 1 minute with a final elongation at 72°C for 10 minutes.

Cleavage reaction. The 271–base pair (bp) PCR product was digested by the restriction enzyme Fnu4HI (Medinova Scientific, Hellerup, Denmark) recognizing 3 sites (GCNGC) in the unmutated gene product, one in the signal peptide/first framework region and 2 adjacent (AGCAGCAGC) in the CDR1 hot-spot region at codons 29-31 (classification by Kabat; 30-32 in IMGT nomenclature) of the variable region CDR1 (Figure 1). Prevention of cleavage in the CDR1 hot spot would demand either at least one mutation in both GCs of codons 29 and 31, or at least one mutation in the G or C of codon 30. The reaction conditions were as follows: 16 μL PCR product, 3.75 μL × 1 PCR buffer, and 2.5 U Fnu4HI incubated at 37°C for 2 hours. A parallel control digestion using 5.0 U DdeI (Medinova Scientific) instead of Fnu4HI was included as control for gDNA (genomic DNA) contaminating the cDNA, which would reveal such contamination by the presence of a FAM-coupled fluorescent 131-bp fragment. Throughout the study no DNA contamination was seen.

Fragment analysis by capillary electrophoresis. A preparation of 1 μL of an internal size standard (Prism Genescan 350 TAMRA, Applied Biosystems, Naerum, Denmark), 13 μL formamide, and 2 μL of the cleavage reaction product was made prior to fragment analysis by capillary electrophoresis (Applied Biosystems prism 310 genetic analyzer). A FAM-coupled fluorescent fragment of 106 or 109 bp indicated cleavage in the hot spot, whereas a length of 244 bp indicated cleavage in the signal peptide site but not in any of the sites in the CDR1 hot spot due to one or more mutations in the latter (Figure 1). The fraction of PCR products with mutations preventing cleavage was calculated as the proportion between the 244-bp fragment peak area and the sum of the peak areas of 244-, 106-, and 109-bp fragments (Figure 2).

Absence of a TET-coupled fluorescent 163/166-bp fragment was used as an internal control demonstrating complete cutting by the Fnu4HI nuclease. Throughout the study failure of the restriction enzyme was not seen.

PCR was performed as described using primers without fluorochromes. The PCR products were cloned into a PCR II vector (TOPO cloning kit; Invitrogen, Life Technologies) as instructed by the manufacturer and transformed into competent cells (One Shot Competent cells; Invitrogen, Life Technologies). Colonies from transformation vials grew on Luria-Bertani (LB) agar plates with ampicillin/ITPG/X-Gal at 37°C for 18 hours. Plasmids were extracted and purified (GFX Micro Plasmid Prep Kit; Amersham Pharmacia Biotech, Uppsala, Sweden), and V_κA27 inserts sequenced by the automatic 310 genetic analyzer using Dye Terminator Kit, AmpliTaq polymerase and primers supplied by the kit.

Case records from the 31 CVID patients were studied by 2 medical specialists unaware of the mutational status of the patients. Four clinical scoring systems (I-IV) were used to evaluate the severity of disease and complications, I to III describing lower respiratory tract involvement and IV describing diarrhea. Score system I divided the patients based on the number of SRTIs recorded within the last 2 years prior to initiation of IgG substitution or diagnosis in case substitution was not initiated immediately (1, no SRTI; 2, ≤ 2 SRTIs per year, 3, > 2 SRTIs per year). SRTI was defined as symptoms of lower respiratory tract infection, elevated numbers of leukocytes and C-reactive protein, body temperature higher than 38.5°C, or antibiotic therapy, and severity of symptoms requiring hospitalization. Score system II described the presence and severity of sequelae to the disease: bronchiectasis, functional dyspnea, reduced forced expiratory volume at 1 second (FEV₁) at time of diagnosis (none, 1; light, 2; severe, 3). Score system III described the same sequelae but over the entire period of observation. Score system IV classified the patients in 2 groups without or with intermittent or daily diarrhea at the time of diagnosis. Basic data included age at onset of symptoms, age at diagnosis of CVID, age at investigation by IgκREHMA analysis, age at registration of the data from case records, and plasma concentrations of IgG, IgA, IgM, and IgG subclasses.

To analyze the expression of CD27 on B cells (memory B cells), fresh peripheral blood mononuclear cells (PBMCs) were incubated with fluorescein isothiocyanate (FITC)–conjugated anti-CD27, phycoerythrin (PE)–conjugated anti-CD19, and peridinin chlorophyll protein (PerCP)–conjugated anti-CD3 (Becton Dickinson, Heidelberg, Germany) and analyzed by fluorescence-activated cell sorting (FACS) on a FACScan (Becton Dickinson). For analysis of the mutation fractions of CD19⁺/CD27⁺ B cells, frozen PBMCs were thawed and enriched for B cells using CD3 depletion with Dynabeads M-450 (Dynal) as instructed by the manufacturer. The cells were incubated with allophycocyanin (APC)–conjugated anti–human CD19 and R-phycoerythrin RPE-conjugated anti–human CD27 (Dako, Glostrup, Denmark) and analyzed on a FACSCalibur (Becton Dickinson). CD19⁺/CD27⁺ cells from 2 patients (women, aged 34 and 49 years) and 2 healthy adults (women, aged 34 and 48 years) were sorted into carrier cells (monocyte cell line U937). The purity of the sorted cells was tested with FACS (CVID patients 60% and 78%, healthy adults 95% and 86%, respectively). The sorted cells were used as a template in the REHMA analysis as described (see “IgκREHMA”).

Plasma concentrations of mannose-binding lectin (MBL) and MBL (mbl2) genotypes (wild-type A allele, mutant D; codon 52, mutant B; codon 54 and mutant C; codon 57) were assessed by conventional enzyme-linked immunosorbent assay (ELISA) and PCR restriction fragment length analyses.^40,41 

Confidence limits for the median were calculated based on the binomial distribution. Reference interval for IgκREHMA results was defined by the values of the second smallest and the second largest value from 53 healthy adults (equal to 96% confidence limits). The Mann-Whitney U test and Kruskal-Wallis test were used to test for different medians of 2 or 3 samples, respectively. P less than .05 was considered statistically significant.

To investigate the status of κ light-chain transcripts in patients with CVID, compared to controls, we developed a quantitative method, V_κA27-specific restriction enzyme based hot-spot mutation assay (IgκREHMA). The method, illustrated in Figure 1, takes advantage of 2 Fnu4HI restriction sites, both included in the CDR1 hot spot. Mutations in both of these restriction sites, or at least one in the central GC in codon 30 shared by both sites, are needed to abrogate cutting by Fnu4HI. Figure 2 shows representative fragment length analyses for 2 CVID patients (a high- and a low-mutating individual), one adult control, and a cord blood sample. Calculated values are given for fraction of V_κA27 transcripts with at least one mutation in the CDR1 hot spot, preventing cutting by Fnu4HI restriction nuclease.

IgκREHMA values in blood samples from 53 healthy adults ranged from 27% to 67%, yielding a calculated 96%-based reference interval for the assay of 28% to 62%. No effect of age was noted within this reference material (Spearman rank correlation coefficient ρ= 0.1, P = .5). The IgκREHMA values of 20 cord blood samples were all 0%.

IgκREHMA values in the 31 CVID patients ranged from 0% to 73% in the first blood sample taken after diagnosis. The median was 7.5%, which was significantly lower than control values (P < .0000001). This difference also held if patients younger (n = 3) or older (n = 3) than the age span of the controls (23-65 years) were omitted (medians, 8.5% and 7%, respectively; P < .0001). Mutations were undetectable (0%) in 4 (13%) of the patients and below the reference interval in 24 (77%; Figure 3). Two patients had values above the reference interval. If only values from patients tested within the first year of diagnosis and before institution of substitution therapy (n = 17) were considered, the IgκREHMA values ranged from 0% to 65% with a median of 9%. These values did not differ significantly from the remaining patients tested at later time points (P > .1).

An important question is whether the levels of hot-spot mutations are stable in CVID patients or change with time. Table 1 lists values for 13 patients from whom several frozen cell samples were available, covering at least 2 years of observation after diagnosis. Patients who had low values in the first sample continued with low values for the entire period (patient nos. 1, 5, 7, 9, 10, 12, 13, 21, 22, and 23 with 3-9 years of observation). Three patients (nos. 2, 17, and 19) who initially had a normal value continued so for 1, 10, and 3 years of observation, respectively, but 2 of them eventually acquired subnormal values. Overall, a slight declining trend was noted by comparing initial values with last values (P = .01, Wilcoxon matched-pairs test), but the decline was very slow. For the 2 patients with supranormal values, the samples were taken 6 years after diagnosis (no. 14) and at the time of diagnosis (no. 11), respectively.

From 2 patients (a low and a highly mutated) and 2 controls, 62 cloned sequences were obtained. Sixty had maximal homology with the published germline gene V_κA27 confirming the gene specificity of the IgκREHMA. Two sequences had maximal homology with V_κL6, which in turn has 98% homology with V_κA27 in the investigated regions (signal peptide, FR1, CDR1, and FR2).⁴² Figure 4 illustrates the frequency and distribution of mutations in V_κA27 sequences from these 4 individuals. For comparison, the same number (11) of randomly selected sequences was used from each individual. Among all 60 V_κA27 sequences, the average prevalence of mutations per nucleotide (codons 1-66, 198 nucleotides) was 3% and 2% for the 2 controls, 0.2% and 4% for the 2 CVID patients, and 0% for the cord blood sample, respectively. The 3 GC pairs critical for cutting by Fnu4HI on average had mutation prevalences per nucleotide of 17% and 14% (controls), 2.8% and 20% (patients), and 0% (cord blood). Also in the poorly mutating CVID patient, mutations were preferentially affecting the CDR1 hot spot (Figure 4). The mutation prevalences of the sequences correlated well with results of the IgκREHMA analysis (Figure 5).

Based on the case reports, the 31 CVID patients were classified clinically by medical specialists unaware of the mutational data. The patients were scored with respect to SRTI within the last 2 years prior to diagnosis (score system I), complications of SRTI (II, III), and presence of diarrhea (IV). Figure 6 shows that patients suffering from SRTI (groups 2 and 3) had significantly lower levels of SHM than patients without reported SRTI (group 1, P < .005). A similar correlation was found when only patients tested for hot-spot mutations within 1 year of the time of diagnosis were analyzed (n = 17, P = .015). Less than half the patients had complications of SRTI (II and III) at diagnosis (n = 13) or during follow up (n = 14) and the levels of SHMs showed no correlation to the scores (P > .38). Thirteen patients suffered from diarrhea, but the presence of diarrhea was neither correlated to light-chain mutations (P = .8), nor to IgG levels at diagnosis, nor to MBL status (data not shown). A correlation of IgκREHMA results with the plasma concentration of IgG at diagnosis was noted (Spearman ρ= 0.5, P = .003), as well for IgA (ρ= 0.47, P = .008) and IgM plasma levels (ρ= 0.39, P = .03). IgG level, however, was a poor predictor of SRTI score (P = .07, Kruskal-Wallis). A correlation of IgA and IgM level with SRTI score was, however, noted (P = .02 and P = .015, Kruskal-Wallis). Neither IgκREHMA results nor SRTI score correlated with the concentrations of B or T lymphocytes in the blood, the B-cell fraction, or age at onset of symptoms (data not shown).

Thirteen of the patients (42%) were heterozygotic for one of the 3 known structural mutations of the mbl2 gene (genotype AB, AC, or AD) and had reduced levels of plasma MBL (median, 148 μg/L; range, 0-1648 μg/L) compared with the remaining patients (genotype AA, 2832 μg/L; range, 384-12 096 μg/L; P < .0001). Heterozygosity (open circles in Figure 6) was associated with high SRTI score (I; P = .009, Mann-Whitney U test). However, the correlation between the prevalence of SHM and SRTI score was also seen when only patients with normal MBL genotypes were analyzed (n = 18, P = .006; filled circles in Figure 6). It was noted that among those with SRTIs, the level of SHM was similar, whereas the patients with MBL deficiency almost exclusively clustered in the most severely affected group (P = .049, Fisher exact test).

Flow cytometric analysis of CD27 expression on B cells was performed in 18 CVID patients and compared with the results of the IgκREHMA analysis. With a few exceptions, the fraction of CD27⁺ B cells in the circulation correlated well with the fraction of mutated κ transcripts (Figure 7; Spearman ρ= 0.6, P = .01). REHMA analysis was performed on FACS-sorted (Fluorescence-Activated Cell Sorter) CD19⁺/CD27⁺ B cells from 2 patients with 6% and 3% CD19⁺/CD27⁺ cells, respectively, and 2 healthy adults. The fractions of mutated κ transcripts of the CD19⁺/CD27⁺ B cells were reduced in the CVID patients compared with healthy adults (Figure 7B).

The clinical importance of affinity maturation of antibody responses is currently unknown. Patients with reduced levels of SHMs due to mutations of the activation-induced cytidine deaminase gene or CD40 ligand gene are highly susceptible to infections including recurrent bacterial infections of the respiratory tract.^43,44  In both conditions, however, isotype switching is also inhibited and contributes to the increased susceptibility to infections.

SHMs may increase the affinity of the antibody 10- to 1000-fold.^31-36  Sufficient binding of antibody to a given pathogen should therefore be possible with much less affinity-maturated antibody than with the unmutated version. Because the total levels of circulating immunoglobulins are restricted, SHM allows for more specificities in the immune antibody repertoire or for longer duration of antibody-mediated memory or both. In humans, who are dependent on antibody-mediated protection for decades after immunization, affinity maturation may prove necessary to combine long-lasting immunity with a sufficiently broad antibody repertoire.

Immunoglobulin mRNA obtained from the blood is predominated by transcripts from circulating plasmablasts and plasma cells.⁴⁵  In healthy adults, about 1000 immunoglobulin-secreting cells are found per milliliter of blood (approximately 1% of B cells).⁴⁶  Almost all (> 90%) of the Igκ transcripts carry SHMs, which tend to concentrate in certain hot spots located in CDR1 in particular. The V_κ gene A27 is the most commonly used light-chain gene and normally accounts for about 10% to 15% of V_κ transcripts in the circulation.^47-49  In this study, V_κA27 transcripts from healthy adults demonstrated SHMs with an average of 4 mutations per transcript equivalent to 2% of the nucleotides (codon 1-66). The V_κA27 CDR1 contains a strong hot spot consisting of 3 overlapping RGYW motifs; this motif is known to be preferentially targeted by SHM.⁵⁰  On the average, 16% of the GC nucleotides of the target sequence AGCAGCAGC were mutated, which is 7 times the average level for nucleotides of the V region. The sequence contains 2 overlapping Fnu4HI restriction sites both of which were modified by mutations in 27% and 31% of the sequences of the 2 healthy adults, respectively (Figure 3). Therefore, these restriction sites were exploited in this study to quantify mutation.

We demonstrate reduced levels of SHM of V_κA27 transcripts in 77% of CVID patients. In comparison, Levy et al²⁹  and Bonhomme et al³⁰  found reduced levels of SHM in IgG VH3-23 transcripts in 23% of CVID patients. The reason for this discrepancy is most likely the fact that the IgG-based method measures mutations only in isotype-switched cells. Even if T/B interactions are malfunctioning, it is possible that the (few) B cells, which obtain sufficient T-cell help for isotype switching, also mutate to a normal extent. In fact, Piqueras et al²⁷  found normal levels of mutation in IgG transcripts from switched memory B cells from patients with CVID. This would yield a normal IgG mutation rate despite a possible skewing of the overall antibody response from mutated IgG to unmutated IgM. In light chain–based SHM analysis, however, the cells donating the mRNA are not necessarily isotype-switched, and low levels of mutations may reflect a reduction of SHM within the individual isotypes as well as a shift from highly mutated IgG- or IgA-secreting cells to less mutated (or unmutated) IgM-secreting cells. Thus, the IgκREHMA result reflects the overall tendency to use mutated (affinity-maturated) versus unmutated antibody in ongoing antibody responses.

If affinity maturation is important for protection against infections, it should be revealed by a propensity for infections in individuals with low levels of light-chain mutations. Indeed, we found a good correlation between low levels of light-chain mutation and a high frequency of SRTIs in this study. Mutation rates appeared more important than IgG levels at the time of diagnosis because the latter did not correlate significantly with the susceptibility to SRTI though a trend for more infections with lower IgG levels was noted (P = .07). Structural mutations in MBL, however, did predict more SRTI episodes. Nevertheless, low IgκREHMA levels remained well correlated with the frequency of SRTI even when patients with MBL deficiency were excluded. This was also the case when patients with lower than median (1.7 g/L) IgG levels were excluded from analysis (P = .04). These results indicate an independent impact of somatic mutation on the susceptibility to SRTI.

It was somewhat surprising not to find a correlation of Igκ mutation to sequelae of lung disease. This could, however, be explained by the extreme variation in time from the first symptoms of immunodeficiency to measurement of SHM (0-51 years; median, 14 years) providing very different conditions for complications to occur. Indeed, the 4 patients with sequelae despite Igκ mutation values above the lower limit of the reference interval had had symptoms for a median of 40 years (range, 9-51 years). The lack of correlation between SHM and prevalence of diarrhea could be related to the poor correspondence between the function of the systemic humoral immune system and that of the small intestine as reported by Herbst et al.⁵¹ 

Important questions relate to whether reduced somatic light-chain hypermutation is a fixed feature characterizing a subgroup of patients or rather a dynamic feature changing with the course of the disease. The relative stability of the mutation levels over several years demonstrated in Table 1 may support the first possibility and makes it meaningful to evaluate SHM and probably to use it prognostically at the time of diagnosis. Normal values were in one case seen over a period of 10 years, and in others 6, 7, and 18 years after diagnosis and, in several cases, more than 40 years after onset of symptoms (reported increased susceptibility to upper respiratory tract infections). Likewise, patients with low values remained so for up to 9 years of observation without any patient demonstrating recovery of SHM. Nevertheless, 3 of 3 patients who initially presented with normal mutation levels and who were followed for more than 2 years did eventually acquire reduced levels of mutation suggesting that an inherent but slow progression with time of this qualitative immunodeficiency may be the rule in CVID. Bonhomme et al found that hypomutation of isotype-switched immunoglobulin genes is correlated to late onset of disease.³⁰  In our study, we did not find this correlation for light-chain mutation.

The correlation of light-chain mutation to the fraction of B cells that carried the memory marker CD27 (Figure 7), however, could indicate that the light-chain mutation levels (unlike mutation of switched heavy chains) might relate to the CVID subgroups proposed by Warnatz et al¹¹  or Piqueras et al²⁷  based on the absence of different memory subpopulations. However, the finding of reduced levels of light-chain mutations among CD27⁺ cells sorted from CVID patients (Figure 7) indicates that a low number of memory cells per se does not explain the lgκREHMA results and points to defects in the SHM machinery itself as at least a part of the explanation.

The influence of MBL deficiency on the predisposition to SRTIs in CVID patients has not previously been reported and should be confirmed in larger patient populations. This finding suggests that MBL may be an important protective factor in CVID patients. Indeed, the 60% of patients with more than 2 episodes of SRTI per year carried structural mutations in the MBL gene while approximately 34% of the background population are heterozygotes.^40,41  This fits well with the demonstration that MBL deficiency increases the susceptibility to respiratory infections in age groups with poorly developed humoral immunity as demonstrated in Greenland infants.⁵²  It may explain the finding of Mullighan and colleagues that MBL deficiency correlated with early onset of CVID.⁵³  We cannot, however, confirm this relation in the present study because neither age at onset of symptoms, nor age at clinical diagnosis correlated significantly with MBL status. This discrepancy could be related to different sampling of patients in the 2 studies.

This study shows that failure of SHM is more common in CVID patients than previously recognized and indicates that poor affinity maturation may predispose to lower respiratory tract infections, which are of paramount importance for the prognosis of these patients. Furthermore, it suggests that deficiency of MBL may also contribute independently to this predisposition. Measurement of light-chain SHM and MBL could prove useful in the management of this disease. Inquiries into mechanisms necessary for SHM might shed light on the pathogenesis of CVID.