SEARCH

SEARCH BY CITATION

Keywords:

  • antibody deficiency;
  • B cell subtypes;
  • common variable immunodeficiency;
  • flow cytometry;
  • memory B cells

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Patients, materials and methods
  5. Results
  6. Discussion
  7. Disclosure
  8. References

Abnormalities in peripheral blood B cell subsets have been identified in common variable immunodeficiency (CVID) patients and classification systems based upon their numbers have been proposed to predict the clinical features. We analysed B lymphocyte subsets by multi-colour flow cytometry (MFC) in a cohort of well-characterized CVID patients to look at their clinical relevance and validate the published association of different classification criteria (Freiburg, Paris and Euroclass) with clinical manifestations. CVID patients had a reduced proportion of total and switched memory B cells (MBC, swMBC) compared to normal controls (P < 0·0006). Patients classified in Freiburg Ia had a higher prevalence of granulomatous diseases (P = 0·0034). The previously published associations with autoimmune diseases could not be confirmed. The Euroclass classification was not predictive of clinical phenotypes. The absolute numbers of all B cell subsets were reduced in CVID patients compared to controls. There was a significant linear correlation between low absolute total B cells and MBC with granulomatous disease (P < 0·05) and a trend towards lower B cells in patients with autoimmune diseases (P = 0·07). Absolute number of different B cell subsets may be more meaningful than their relative percentages in assessing the risk of granulomatous diseases and possibly autoimmunity.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Patients, materials and methods
  5. Results
  6. Discussion
  7. Disclosure
  8. References

Common variable immunodeficiency (CVID) is a heterogeneous group of primary antibody deficiencies characterized by decreased levels of at least two immunoglobulin isotypes and constitutes the most common cause of hypogammaglobulinaemia. It usually manifests during the second, third or fourth decade and its prevalence is about one in 50 000 [1–3]. The term was first introduced in 1973 to differentiate this entity from other defined primary antibody deficiency syndromes [4].

Recurrent infections, autoimmune diseases, other forms of immune dysregulation such as granulomatous inflammation and malignancies are the most common manifestations in this group of patients [5–8]. CVID is generally regarded as a defect of immunoglobulin synthesis and regular immunoglobulin replacement is usually necessary, although not always sufficient to prevent infective episodes [9,10]. The broad variability of clinical manifestations in these patients may reflect distinct pathogenic mechanisms. This has made classification of the patients difficult. Although a few genetic defects that may cause CVID have been discovered recently, the majority of CVID patients suffer from as-yet unidentified genetic or acquired abnormalities [11–15].

Classification schemes have been proposed depending upon the relative frequency of various memory B cell (MBC) subsets and their association with different clinical manifestations. Warnatz et al. defined three subgroups (Ia, Ib or II), referred to collectively as the Freiburg classification, to explore the germinal centre-dependent and peripheral differentiation of B cell subsets by analysing class-switched MBC (swMBC) and CD21 expression, respectively [16]. Similarly, Piqueras et al. suggested a Paris classification based on the assessment of class-switched MBC and total CD27+ B cells. It classifies CVID patient into three groups, MB0, MB1 or MB2. Autoimmunity complications were increased in patients in the MB0 and MB1 groups compared to those in the MB2 group, and granulomatous disease and splenomegaly were found more often in the patients in the MB0 group. Although not precisely overlapping, MB2 patients are most similar to group II patients and MB0/MB1 patients most closely match the group I patients [17]. Finally, the Euroclass was introduced after analysis of more than 300 European CVID patients. It included some of the previous classification aspects plus new criteria to define CVID patients [18]. However, the usefulness of the Euroclass classification system has not been validated in independent cohorts. In this study, we analysed peripheral blood B lymphocyte subsets in a cohort of well-characterized CVID patients in order to look for association of various B cell subsets with different clinical manifestations and to validate the published association of different classification criteria (Freibrug, Paris and Euroclass).

Patients, materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Patients, materials and methods
  5. Results
  6. Discussion
  7. Disclosure
  8. References

Patients

Fifty-three consecutive CVID patients attending the immunodeficiency clinic at the Royal Adelaide Hospital were recruited into our study. Clinical details of infectious episodes [sinusitis, bronchiectasis, pneumonia, skin infection and central nervous system (CNS) infection], allergic manifestations, autoimmune and granulomatous diseases, specific antibody response to pneumococcal carbohydrate vaccine and baseline immunoglobulin level were ascertained by the clinician assessing the patient. The presence of granulomas was recorded based upon the available histopathological data and radiological studies. This was a cross-sectional study and no routine screening studies were incorporated. The diagnosis of CVID was determined by a decrease in serum immunoglobulin (Ig)G (greater than 2 standard deviations below normal range), the presence of recurrent bacterial infections and exclusion of other causes of hypogammaglobulinaemia [19]. Immunoglobulins were measured by nephelometry (Beckman-Coulter, Fullerton, CA, USA). The response to pneumococcal vaccine was measured by looking at the increase in the pneumococcal antibody titres; total immunoglobulin (Ig)G and IgG2 measured by enzyme-linked immunosorbent assay (ELISA) (The Binding Site Ltd, Birmingham, UK) from baseline after 4–6 weeks following a 23-valent pneumococcal polysaccharide vaccine. A greater than fourfold rise in titres of both specific IgG and IgG2 from baseline were considered adequate responses.

Flow cytometry

Peripheral blood was collected in lithium heparin for analysis by multi-colour flow cytometry (MFC). Analysis of different B cell subsets was performed by using a combination of B cell markers. At room temperature, 100 μl of whole blood in two tubes was processed using ImmunoPrep Reagent System (reagents A and B; Beckman-Coulter, Fullerton, CA, USA), following a three times wash and whole-blood lysing method. The washing was performed with 2 ml Dulbecco's phosphate-buffered saline with 5% bovine calf serum (BCS). Then tubes were centrifuged at 1000 g for 30 s using a DiaMed DiaCent-12 centrifuge. Tubes were stained for 10 min with antibodies to CD19-extracellular domain (ECD), CD27-phycoerythrin (PE), IgM-PCy5 and IgD-fluorescein isothiocyanate (FITC)-PE for the first tube. The second tube was stained for CD19-ECD, CD21-PE, IgM-PCy5 and CD38-FITC. Finally both tubes were fixed with 250 μl formaldehyde solution fixative and were analysed within 24 h of processing. Data acquisition and analysis were performed on a FC500 flow cytometer (Beckman-Coulter). The lymphocyte gate as defined by forward- and side-scatter was analysed with CD19 and CD27 to define the MBC and non-MBC populations and also against CD19 and CD21 to define CD19+ CD21lo B cells. Then, the MBC gate was analysed with IgM and IgD to define IgM-only MBC, marginal zone-like B cells, switched MBC and IgD MBC. The CD27- B cell gate was also analysed with IgM and IgD to define naive B cells. Combined staining for CD19, CD21, CD38 and IgM permits the distinction of transitional B cells (CD19+CD21loCD38++IgM++) and plasmablasts (CD19lo CD21lo CD38++ IgM-). Accordingly, patients were classified further into various subgroups based on the published classification criteria; Freiburg (Ia, Ib, II), Paris (MB0, MB1, MB2) and Euroclass.

Statistical analysis

Wilcoxon's test was used to compare the B cell subsets and clinical manifestations. Logistic regression was used to determine the significance of B cell subsets (absolute number) as predictors of the different clinical and laboratory parameters. Fisher's exact test was used to test whether Paris and Freiburg classifications were associated with the presence of granulomatous disease or autoimmune diseases. A two-tailed P-value of 0·05 was used to indicate statistical significance. All calculations were performed using SAS version 9·2 (SAS Institute Inc., Cary, NC, USA).

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Patients, materials and methods
  5. Results
  6. Discussion
  7. Disclosure
  8. References

Clinical features and immune phenotyping of CVID patients

Thirty-nine control subjects were analysed for the derivation of normative values of different B cell subsets. The values did not differ significantly from the published normal ranges and thus it was thought that the patients could be assigned to the respective classification systems without instituting a bias in the system. Fifty-three CVID patients (38 female, 15 male), with a median age of 57 years (range of 9–87 years) were recruited into this study (Table 1). The median age of patients at the time of diagnosis of CVID was 48 years (range 5–77 years). The long-term sequelae of recurrent infections in the form of chronic sinusitis and bronchiectasis were seen in 35 (66%) and 23 patients (43%), respectively. Allergic manifestations were present in 18 (34%) patients (eight with asthma, five with food allergy, three with drug allergy and two with honey bee venom allergy. The median IgE in 15 allergic patients was 9 kU/l (range <5–310). Autoimmune diseases were present in 12 (23%) patients (two psoriatic arthritis, two Sjögren disease, two with insulin-dependent diabetes mellitus (IDDM) type 1, two with autoimmune hypothyroidism, one with pernicious anaemia, one with seronegative arthritis, one with Goodpasture's syndrome and one with discoid lupus. Six patients (11%) had granulomatous diseases and/or splenomegaly (three with splenomegaly, one with granulomatous lung disease, one with CNS granuloma and one with intestinal granulomatous process). Although the most common cause of splenomegaly in CVID is a benign lymphoproliferative disease, it was clustered with granulomatous disease in the final analysis to increase patient numbers, as both have been shown to be associated with similar abnormalities in the peripheral blood B cell subsets [12,20]. Lymphadenopathy was present in one patient who also had splenomegaly.

Table 1.  Demographic, clinical and laboratory details.
Characteristics of patients (n = 53)
  1. Ig: immunoglobulin.

Female : male2·5:1
Median age (years)57
Median age at diagnosis (years)48
Infective sinusitis35 (66%)
Bronchiectasis23 (43%)
Allergy18 (34%)
Autoimmune diseases12 (23%)
Granulomas diseases6 (11%)
Lack of pneumovax response (n = 20)12 (60%)
Baseline IgG (g/l) 
 < 3(26%)
 3–6(74%)

Most patients had baseline IgG level between 3–6 gm/l (74%). Details about the functional antibody response to pneumococcal vaccine were available in only 20 patients. Eight patients (40%) showed an adequate response (rise in the pneumococcal titre ≥ 4 times the baseline) to vaccination, but were still given a diagnosis of CVID based on low total IgG, the presence of recurrent infections and exclusion of other causes of antibody deficiency, as described previously [21]. The disease severity in terms of infective sinusitis, bronchiectasis and granulomatous diseases was not significantly different between the vaccine responders and the non-reponders. The mean baseline IgG levels were not significantly different in patients with inadequate and adequate response to vaccination (5·2 g/l and 5·4 g/l, respectively).

B cell subset analysis by MFC

Analysis of B cell subsets expressed as percentage of total lymphocytes and B cells.  B cell subsets were analysed both as percentage and absolute count and compared to normal controls (Figs 1 and 2). The controls used for comparisons were greater than 18 years of age. However, there were six patients under 18 years in the study population for whom published reference ranges were applied [22]. There was a significant reduction in total MBC and swMBC in CVID patients (P < 0·0006), expressed as a proportion of B cells. The median percentage of B cells, MBC, swMBC, transitional B cells and plasmablasts was reduced in CVID compared to the controls (P < 0·05) (Fig. 1). Although the median percentage of CD21lo B cells was comparable to normal controls (P = 0·1113), all three patients with splenomegaly had high CD21lo cells; mean 27·10% (normal range −0·9–12·15%), as published previously [18]. The only patient with lymphadenopathy had high transitional B cells; 14·29% (normal range 0·7–3·9%), as has been reported previously [18]. It was observed that all patients with an inadequate functional antibody response (12 patients) had low MBC and swMBC. However, there were six patients with low MBC and swMBC levels who showed an adequate response to vaccination, and hence this finding did not reach statistical significance.

image

Figure 1. Analysis of B cell subsets in common variable immunodeficiency (CVID) patients (n = 53) and controls (n = 39) as proportion of lymphocytes and B cells (bars show maximum, 3rd quartile, median, 1st quartile and minimum values): the median percentage of B cells, memory B cells, switched MBC, transitional B cells and plasmablasts was reduced in CVID compared to the controls, P < 0·05. The median CD21lo B cells was comparable in CVID patients to normal controls; P = 0·1113.

Download figure to PowerPoint

image

Figure 2. Analysis of B cell subsets in common variable immunodeficiency (CVID) patients (n = 53) and controls (n = 39) as an absolute count (bars show maximum, 3rd quartile, median, 1st quartile and minimum values): there was significant reduction in the absolute number of total B cells, total and switched memory B cells, CD21lo cells, transitional B cells and plasmablasts in patients with CVID compared to controls, P < 0·05.

Download figure to PowerPoint

All the patients were classified into different subgroups based upon the Euroclass, Freiburg and Paris classifications. The association between Freiburg group Ia and granulomatous diseases was confirmed in this cohort (P = 0·0034) (Table 2).Granulomatous diseases were reported in 60% of patients in category Ia compared to 10·3% in category Ib and 0·0% in category II. However, there was no association with the Paris MB0 group (P = 0·27). In our group there was no association of autoimmunity with Paris group MB0 and MB1, contrary to previous published reports [17]. Euroclass was not helpful in segregating the patients according to clinical features.

Table 2.  Presence of granulomatous or autoimmune diseases according to Paris (MB0, MB1 and MB2) and Freiburg (Ia, Ib and II) classifications.
Classification systemGradeGranulomatous disease n (%)P-valueAutoimmune diseases n (%)P-value
ParisMB0 (n = 27)4 (14·8)0·276 (22·2)0·35
MB1 (n = 15)0 (0·0)5 (33·3)
MB2 (n = 11)2 (18·2)1 (9·1)
FreiburgIa (n = 5)3 (60·0)0·00340 (0·0)0·29
Ib (n = 29)3 (10·3)9 (31·0)
II (n = 19)0 (0·0)3 (15·8)

Analysis of B cell subsets expressed as an absolute count (Fig. 2).  The absolute numbers of various B cell subsets and their association with different clinical manifestations was also examined in this study. There was a significant reduction in the absolute number of total B cells, total and swMBC cells, CD21lo cells, transitional B cells and plasmablasts in patients with CVID compared to controls (P < 0·05). The proportion of MBC in total B cells was reduced in CVID patients (the ratio was 0·20 ± 0·22 in CVID patients compared to 0·28 ± 0·13 for controls, P = 0·03). The swMBC component of MBC was also reduced significantly in CVID patients compared to controls (0·30 ± 0·22 in CVID patients and 0·49 ± 0·15 in control patients, P < 0·001).

Logistic regression analysis of absolute numbers of different B cell subsets as predictors of various clinical manifestations revealed an increasing risk of granulomatous disease and/or splenomegaly with decreasing absolute numbers of total B and MBC cells [odds ratio (OR) = 1·30, 95% confidence interval (CI) 1·05–1·61, P < 0·05 for total B cells and OR 18·42, 95% CI 1·37–243·90), P < 0·05 for MBC cells]. There was a borderline association between autoimmune diseases and low B cell numbers (OR = 1·07, 95% CI 0·99–1·15, P = 0·07). There was no difference in the absolute B cell subset numbers in patients with and without chronic sinusitis, bronchiectasis or allergic diseases.

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Patients, materials and methods
  5. Results
  6. Discussion
  7. Disclosure
  8. References

Disturbances in the B cell homeostasis have been observed consistently in all the major published cohorts of CVID patients. Our cohort also showed significant alterations in the different B cell subsets, in keeping with the published literature. Total MBC and swMBC were found to be reduced significantly in CVID patients compared to controls. The validation of this finding in multiple cohorts could possibly warrant the inclusion of this additional parameter in the diagnosis of antibody deficiency. However, we observed a decrease in this subset in a few subjects (unpublished data) being investigated for suspected immunodeficiency who turn out to have normal immunoglobulin levels and vaccine responses. Hence, the utility of this parameter as a stand-alone test for antibody deficiency needs to be clarified further.

The major goal of the published classification schemes utilizing the variation in different peripheral B cell subsets has been to find a surrogate marker to identify clinical phenotypes. These studies reported consistently a variation in the B cell compartment in CVID patients and stratified them on the basis of B cell subset phenotype, as a percentage of either total lymphocytes, or B lymphocytes, or both. The Euroclass classification is the largest multi-centre cohort (303 patients) which distinguished CVID patients according to the total MBC, swMBC, transitional and CD21lo cells according to percentage of total B cells. They found that a severe reduction in swMBC cells was associated with higher risk of splenomegaly and granulomatous diseases. Low maturation marker CD21lo cells were found more in patients with splenomegaly, whereas patients with lymphadenopathy showed expansion of transitional B cells. They also validated the previously published classification schemes and found a higher incidence of splenomegaly and granulomatous disease in Paris MB0 (reduced total and swMBC) and Freiburg Ia (high CD21lo cells with reduced total MBC). However, in a separate Australian cohort, only Freiburg classification Ia was found to be useful in associating with clinical features of lymphoproliferation and autoimmune cytopenias with no utility of Paris classification [23]. In our cohort, we found that patients with granulomatous disease and/or splenomegaly were classified in Freiburg Ia, in keeping with the published association. However, we could also not confirm any other association of the clinical phenotypes with the Paris or the Euroclass classification schemes. Recent publications have reported abnormalities in the distribution of peripheral blood T lymphocytes which are not incorporated into the current classification systems [24]. Clearly, this is an area for further research.

The analysis of B cell subsets as proportion of B cells (i.e. percentages) gave different results compared to absolute numbers of these cells. We found that the absolute numbers of all B cell subsets were reduced significantly in CVID patients. Although absolute CD21lo cells were reduced in CVID patients, they were normal when analysed as a percentage of B cells. Moreover, patients with autoimmune diseases tended to have a lower number of total B cells (P = 0·07), while patients with granulomatous disease and/or splenomegaly had reduced absolute numbers of both MBC and swMBC (P < 0·05). The gradient of increase in risk of these complications with reduction in the B cell populations makes more logical sense, and possibly points towards a clinicopathological association. The absolute numbers of other B lymphocyte subsets, including the swMBC CD21lo cells, did not show any correlation with any clinical manifestation. Yong et al reported similar findings of high incidence of granulomatous and autoimmune diseases in a cohort of paediatric CVID patients with low total B cells, total MBC and swMBC [25]. However, in our cohort the reduction in swMBC was not significant (P = 0·09) in patients with granulomatous and autoimmune diseases.

A possible discrepancy in the association between the clinical phenotypes and the different classification systems could lie in the usage of relative distribution of the B cell subsets (as a percentage of total lymphocytes or B cells) rather than using the absolute numbers of these cells. The relative distribution of a particular lymphocyte subset can be affected by the subset not in question; it would either increase or decrease depending upon the presence of other subsets which are not of interest for that particular analysis. An analogy of this discrepancy can be found in patients with human immunodeficiency virus (HIV) infection, where absolute CD4 count rather than percentage was found to be a superior prognostic indicator for patients with CD4 counts lower than 200 cells/mm3[26].

There is only one published study at the time of preparation of this manuscript whose primary aim was to find a computational approach for phenotype analysis of CVID patients [27]. Ours is the first clinical study to examine the clinical utility of the Euroclass classification and validate the published associations independently. We did not find a significant clinical association with any of the Euroclass groups. This might be due to the relatively low number of patients in the current cohort. However, even if that is the case, this study tends to negate the clinical use of these classification systems in stratifying individual patients into different risk categories. Our data suggest that it may be more accurate to use absolute numbers of B lymphocyte subpopulations in future studies of CVID phenotypes. There is limited usefulness for these studies in routine clinical practice before these questions are answered.

Disclosure

  1. Top of page
  2. Summary
  3. Introduction
  4. Patients, materials and methods
  5. Results
  6. Discussion
  7. Disclosure
  8. References

There was no external source of funding for this trial. There is no reportable conflict of interest for any of the authors.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Patients, materials and methods
  5. Results
  6. Discussion
  7. Disclosure
  8. References