SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. References

Common Variable Immunodeficiency Disorder (CVID) is a complex disorder that predisposes patients to recurrent and severe infections. Immunophenotypic classification schemes were developed to categorize patients with CVID into phenotypic and prognostic groups based on different memory B cell subsets. Whether the B cell subset analysis is stable over time has not been investigated. B cell phenotyping in patients with CVID (n = 15) and sex- and age-matched controls (n = 26) were carried out according to the three B cell classifications. Patients with CVID were evaluated monthly over 6 months. Controls were assessed once during the study. We scored how often each patient was assigned to the same group within each classification. The Freiburg classification assigned patients to the same group at a rate of 73% and the Paris classification at 88%. The EUROclass classification of smB− versus smB+ was at 90%. The two subclassifications [(smB-21low or smB-21norm) and transitional B] were at 87% and 97%, respectively. The level of naïve B cells measured in all patients with CVID during the 6-month evaluation was the most stable B cell subset. We conclude that all classifications systems show considerable variability, but the EUROclass classification was the most reliable scheme for our 15 CVID and 26 healthy cohorts. Our results indicate that phenotypic classifications within CVID will be difficult while there is variability of commonly used assays.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. References

Immunophenotypic classification schemes such as Freiburg (2002) [1], Paris (2003) [2] and EUROclass (2008) [3] were developed to categorize patients with Common Variable Immunodeficiency Disorder (CVID) into phenotypic and prognostic subgroups based on different memory B cell subsets. B cell subset number analyses have been linked to granulomatous disease, splenomegaly, lymphadenopathy and a trend towards autoimmune diseases in CVID [1-3]. Given the potential of these methods for identifying patients for targeted surveillance, we investigated the reliability of these classification schemes when the same patients were remeasured over time. To our knowledge, such reliability analyses have not been previously reported.

The Freiburg classification scheme [1] differentiates B cell subsets based on CD19, IgM, IgD, CD27 and CD21 expression (Fig. 1A). Patients with less than 1% B cells in the peripheral blood are excluded from this classification. Group I are patients with severely reduced numbers of switched memory B cells (CD19+ CD27+ IgM− IgD−). Patients in group II have normal counts of switched memory B cells. Group I is further divided into group Ia with an increased proportion of CD21low B cells and group Ib with normal percentages of CD21low B cells. Patients with CVID in group Ia showed higher prevalence of splenomegaly, granulomatous disease [4] and autoimmune cytopenias compared with other patients [1, 5].

image

Figure 1. The three memory B cell classification schemes. (A) The Freiburg classification [1]. (B) The Paris classification [2]. (C) The EUROclass classification [3]. PBL, peripheral blood lymphocytes.

Download figure to PowerPoint

The Paris classification scheme assigns patients according to the naïve/memory B cell phenotype [2]. The IgD and CD27 expression assigns patients to one of the three subtypes (Fig. 1B); MB0 are patients with almost no memory B cells, MB1 are patients with defective switched memory B cells but normal non-switched memory B cells and group MB2 consists of patients with normal B cells. Patients within the MB0 group have higher prevalences of splenomegaly, lymphoid proliferation and granulomatous disease compared with patients in other groups [2].

The EURO classification (Fig. 1C) is a two-tier scheme. The study was initiated to develop a consensus of the Freiburg and Paris classification schemes [3]. Like the Freiburg scheme, patients with less than 1% B cells are excluded from this classification. The patients with more than 1% B cells are divided into smB− and smB+. A further subclassification divides smB− and smB+ into CD21low or CD21norm. Within the smB− patient group, a subgroup is characterized by the expansion of transitional B cells (smB-Trhi) based on the expression of CD38 and IgM. Expansion of CD21low B cells represents the strongest marker for splenomegaly, and patients with smB-Trhi more often present lymphadenopathy [3].

In this study, we explored two different aspects of the three B cell classification schemes. Firstly, we questioned whether the three classification schemes were reliable in assigning the same patient over time to the same group. Each classification scheme was scored according to how reliably the same patient was assigned to the same subgroup every month over a 6-month period. Secondly, the 6-month time frame also allowed us to assess the stability of the different B cell subset values over time.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. References

Study population

Fifteen adult patients with CVID (10 male, five female; aged 28–68, mean 46 years) were selected from those attending the Immunology day clinic for immunoglobulin (IVIG) replacement therapy at Auckland City Hospital. Blood samples were taken prior to IVIG treatment. Diagnosis of CVID had been made according to the European Society for Immune Deficiency Disorders (ESID), the Pan American Group for Immune Deficiency (PAGID) and the Clinical Immunology Committee of the International Union of Immunological Societies (IUIS) [6]. Control blood samples (n = 26) were collected from healthy personnel and students working at Auckland City Hospital.

Informed written consent was obtained from each individual prior to participation. The consent forms and procedures were approved by the institutional review board of the Ministry of Health (MEC/06/10/134).

Flow cytometry

Peripheral blood was collected in lithium heparin tubes prior to IVIG replacement therapy. Flow cytometric analysis of the peripheral blood was undertaken within 24 h of the venepuncture as described previously [7]. Analyses were performed by a single operator (WK). Patients with a B cell proportion less than 1% of peripheral blood lymphocytes were excluded, as was reported in the Warnatz and Wehr studies [1, 3]. The exclusion criteria were chosen due to the difficulty of accurately determining the surface phenotype on small numbers of cells. Each patient with CVID was assessed monthly for 6 months. The healthy donors were evaluated once during the study.

Whole blood was stained with anti-CD27-PerCP-Cy5.5, anti-IgD-FITC, anti-IgM-APC, anti-CD38-PE, CD21-FITC (all from BioLegend, San Diego, CA, USA) and anti-CD19-APC-H7 (BD Pharmingen, San Jose, CA, USA). The lymphocytes were gated on size using forward scatter (FSC) versus side scatter (SSC). B cells were gated by CD19 expression.

Stained cells were analysed on LSRII Flow Cytometer (BD Biosciences, San Diego, CA, USA), and the data processed using FCS Express 3 (de Novo software, Los Angeles, CA, USA). Percentages for the Paris and EUROclass classifications are based on CD19+ cells, while percentages for the Freiburg classification are based on PBL.

Classification assignment

Each patient was assigned to the appropriate group as described previously according to their immunophenotype [1-3]. Patients were assigned to a group within each classification on six occasions. The patient group with the most ‘hits’ (at least 4 of 6 data points) was regarded as the ‘correct’ patient group for a particular patient. Ninety data points were collected for the 15 recruited patients for this study. The reliability of the phenotypic classification was assessed on the percentage of how often these patients were classified accordingly in the ‘correct’ group over 6 months. Healthy individuals were also classified according to the three classification schemes.

Statistical analysis and graphs

The consistency of the classification was summarized by the proportion of times the patient was classified into the same class. The 95% confidence interval of the proportion was derived using GraphPad Prism 5. The specificity of the classification was summarized by the proportion of times, and the normal controls were classified into the expected class.

Variability of the B cell subset measurements was assessed by comparing the coefficient of variation (CV) values. CV values were calculated by dividing the standard deviation by the mean B cell values over 6 months for each patient in each B cell subset. The CV value is the standardization of the standard deviation by removing the mean as a factor of variability [8]. Average CV values are presented as median (IQR). GraphPad Prism 5 was used to compute the results.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. References

Freiburg classification scheme

Allocations for all patients and healthy donors are presented in Fig. 2A (switched memory B) and Fig. 2B (CD21low). Nine patients were classified in group I and four in group II. Two patients were not classified because three data points were within group I and three within group II. Of the nine patients in group I, eight patients were Ia and one patient could not be classified based on the level of CD21low cells as this patient was classified within each group three of 6 months. The healthy controls were expected in group II with normal B cell levels. The controls were classified as group Ia (12%), group Ib (15%) and group II (73%). This indicates a specificity of 73% for the Freiburg classification. Analyses of how often each patient was classified within the same group showed that the Freiburg classification was on average reliable in 73% (Table 1) of the cases, 95% CI (64, 83).

Table 1. Comparison of the consistency of different classification schemes in classifying the 15 patients with CVID over 6 months
PatientFreiburg (%)Paris (%)EUROclass (%)
smBCD21low
  1. Each patient was classified according to the ‘correct’ patient group as described in the methods. The percentages given are based on the consistency of the assignment of the patients, within this correct group, over 6 months.

167676767
2831008367
35050100100
45010083100
51001008383
68383100100
7831008367
867100100100
9678310083
101001006783
11505010083
125010010083
138310010067
1483100100100
158383100100
Mean73889186
image

Figure 2. Freiburg classification analysis over 6 months based on (A) the level of switched memory B cells and (B) CD21 expression. (A) The threshold level of switched memory B cells in PBL is 0.4%. I: less than 0.4% switched memory B cells in PBL, II: more than 0.4%. (B) Ia: more than 20% CD21low B cells, Ib: less than 20% CD21low B cells. Each data point represents one monthly flow cytometry result.

Download figure to PowerPoint

Paris classification scheme

Seven patients with CVID were MB0 based on the level of CD27-positive cells (Fig. 3A). The remaining patients were classified based on the level of switched memory B cells into MB1 (four patients), MB2 (three patients) and one patient could not be classified (Fig. 3B). Healthy controls were expected to be in MB2 with normal levels of switched memory B cells. The proportion of normal donors in MB0, MB1 and MB2 were 19%, 23% and 58%, respectively. This result indicates a specificity of 58% for the Paris classification. Paris classification was on average reliable in 88% (Table 1), 95% CI (78, 98).

image

Figure 3. Paris classification over 6 months based on (A) CD27 expression and (B) the level of switched memory B cells. (A) The threshold level for CD27 expression on B cells is 11%. MB0: less than 11% CD27+ B cells; the other patients are then classified based on the level of switched memory B cells. B - The threshold level of switched memory B cells is 8%. MB1: less than 8% switched memory B cells; MB2: more than 8% switched memory B cells. Each data point represents one monthly flow cytometry result.

Download figure to PowerPoint

EUROclass classification scheme

The EUROclass profiles three levels of cell surface protein expression in the process of segregating patients into distinct groups (Fig. 1). Four patients with CVID were smB−, and 11 patients were smB+ (Fig. 4A). Patients with SmB− and smB+ were segregated based on the level of CD38lowCD21low B cells (Fig. 4B). One patient was smB-21low, and all other patients were smB-21norm. The level of transitional B cells was assessed in the four patients who were smB−. They were all smB-Trnorm (results not shown). All 26 healthy individuals were smB+, and hence, transitional B cells quantification was not required. All 26 healthy age-matched controls were smB+ because they have normal levels of switched memory B cells. Most (97%) of the healthy controls were smB-21norm because they were not expected to have elevated levels of CD21low B cells.

image

Figure 4. EUROclass classification based on (A) the level of switched memory B cells and (B) CD21 and CD38 expression. (A) The threshold level of switched memory B cells is 2%. smB−: less than 2% switched memory B cells; smB+: more than 2% switched memory B cells. (B) The threshold level of CD38lowCD21low expression in B cells in the EUROclass classification is 10%. Patients with more than 10% CD38lowCD21low B cells are classified as smB-21low, while the other patients are classified as smB-21 normal. Each data point represents one monthly flow cytometry result.

Download figure to PowerPoint

The EUROclass classification showed reliability in 91% of the cases based on the level of switched memory B cells (Table 1), 95% CI (84, 98). Classification of both groups (smB− and smB+) into smB-21low or smB-21norm was reliable in 86% of the cases, 95% CI (78, 93). The classification of patients with smB for transitional B cells was reliable in 97% of the cases, 95% CI (90, 100).

Stability of the B cell subsets

The level of naïve B cells measured in all patients with CVID during the 6-month evaluation was the most stable B cell subset (Fig. 5). The median CV value was 3.9 (IQR = 2.0–5.9), which was the lowest of all B cell subsets. The other subsets were less stable over time as evidenced by the higher CV and min/max values.

image

Figure 5. Coefficient of Variation (CV) values of the B cell subsets. CV values of the different B cell subsets were calculated in 15 patients with CVID over 6 months. Median values are presented as the line within the box, with the box extending from 25th to 75th percentiles. Whiskers show the minimum and maximum values. Dotted line represents the 5% level. One CV value was taken out of the transitional B cell subset as this was an outlier with a value of 245% which was due to five of six values being 0.00 and one value being 0.04.

Download figure to PowerPoint

We compared the switched memory B cells as determined by Freiburg and Paris/EUROclass schemes. The median CV value was higher in the Freiburg analysis than the Paris/EUROclass analysis [median CV values: 41.3 (IQR = 32.5–52.6) versus 24.0 (IQR = 20.7–37.5), respectively]. The level of activated CD21low B cells was similar when CD38 was used as a second marker (EUROclass) than when CD21 was only used (Freiburg) to determine this subset [median CV values: 50.4 (IQR = 37.3–62.2) versus 47.8 (IQR = 34.3–58.2)].

The levels of smB-Tr cells and plasmablasts were less stable than the levels of marginal zone B cells, switched memory B cells and CD21low B cells over time in this group of patients.

Correlation of B cell classification groups and clinical phenotypes

Splenomegaly, lymphadenopathy, autoimmune cytopenias and granulomatous disease have been linked to groups within the three classification schemes [1-3]. We found no correlation of such clinical phenotypes in our patient cohort prior to commencement and during the study period (data not shown).

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. References

The variation in lymphocyte subset measurements can be influenced by a number of factors including diurnal variation, age of sample, the disease state of the patient, the cell storage temperature, cell preparation and instrument settings [9, 10]. Age of sample variations were minimized by processing and staining blood samples within 24 h of venepuncture. In addition, immunophenotyping was performed by the same person on the same flow cytometer in an attempt to further minimize data deviation. Intra-assay variability was assessed for one patient by processing two samples in parallel. Minimal variation was observed between the samples (data not shown), suggesting variability seen is not likely due to assay variability in our cohort; however, this needs further validation. B cell subset analysis by a single operator over 6 months would however be unusual in a routine diagnostic laboratory. Multiple operators may add a further layer of variability to these complex assays.

Ficoll-separated cells were analysed in the Freiburg and Paris classification schemes [1, 2]. In the EUROclass trials, some centres used the whole blood method, while others used Ficoll-separated cells for their analyses [3]. We used whole blood immunophenotyping because of the potential use of these classification schemes in a diagnostic setting. The whole blood lysis technique saves considerable labour time and has previously demonstrated its equivalence to Ficoll separation [5, 7].

In this report, we evaluated the reliability of each classification scheme in assigning patients into their respective groups. Six monthly evaluations were carried out in 15 patients with CVID. We found that EUROclass (91 and 86%) is more reliable than Freiburg (73%) in assigning patients according to switched memory B cells (Table 1). The differences between these schemes is that the level of switched memory B cells is assessed as a percentage of peripheral blood lymphocytes in the Freiburg scheme, whereas the EUROclass scheme measures the level of switched memory B cells as a percentage of B cells. Variability in B cell levels (Fig. 5) may therefore have potential for more variation in the Freiburg scheme.

The CD21low subset was evaluated in the Freiburg and EUROclass schemes. Both classifications produced similar reliability of 84–85% in our cohort. The EUROclass CV values for both schemes are similar although a wider spread of values was observed in the EUROclass (Fig. 5). Inclusion of anti-CD38 in EUROclass did not seem to alter the CD21low levels although CD38 is a marker for activated CD21low cells.

The 26 age-matched controls provide additional information about the different classification schemes. In this study, the EUROclass appears to have the highest specificity among the three protocols evaluated. All controls were smB+ with normal levels of switched memory B cells. All but one was smB-21norm with no increased levels of CD38lowCD21low. B cell number variability in the Freiburg scheme can result in reassigment of patients from group I into group II. Such variable designation could be due to the low threshold level of 0.4% memory B cells. Our cell preparation from whole blood could be another possible explanation for the slight differences in B cell levels.

We evaluated the stability of the different B cell subset values in a 6-month period in this CVID cohort. Six monthly data were collected in each B cell subset for each patient. We calculated the coefficient of variation (CV) for each B cell subset in all patients to facilitate valid comparison (Fig. 5). The measurement of naïve B cells was the most stable over time in our patient group with a median of 3.9 (IQR = 2.0–5.9).

Clinical phenotypes of patients with CVID have been correlated with B cell subsets and classification groups [11], but we were unable to correlate any of these clinical phenotypes with their respective classifications.

We conclude that the EUROclass classification was the most reliable scheme among the three tested in a diagnostic setting for our 15 CVID and 26 healthy control cohorts. Variability in B cell memory subset numbers may hinder phenotypic subclassification of patients with CVID and needs to be explored further in a larger CVID cohort.

Acknowledgment

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. References

We thank the patients for participating in these studies for the benefit of others. We acknowledge the management of Auckland District Health Board for their support. We thank Dr Maia Brewerton for helpful discussions. We thank the A+ Charitable Trust for ongoing support and Immunodeficiency Foundation of New Zealand (IDFNZ) for their assistance.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. References
  • 1
    Warnatz K, Denz A, Drager R, Braun M, Groth C, Wolff-Vorbeck G, et al. Severe deficiency of switched memory B cells (CD27(+)IgM(-)IgD(-)) in subgroups of patients with common variable immunodeficiency: a new approach to classify a heterogeneous disease. Blood 2002;99:154451.
  • 2
    Piqueras B, Lavenu-Bombled C, Galicier L, Bergeron-van derCruyssen F, Mouthon L, Chevret S, et al. Common variable immunodeficiency patient classification based on impaired B cell memory differentiation correlates with clinical aspects. J Clin Immunol 2003;23:385400.
  • 3
    Wehr C, Kivioja T, Schmitt C, Ferry B, Witte T, Eren E, et al. The EUROclass trial: defining subgroups in common variable immunodeficiency. Blood 2008;111:7785.
  • 4
    Al Kindi M, Mundy J, Sullivan T, Smith W, Kette F, Smith A, et al. Utility of peripheral blood B cell subsets analysis in common variable immunodeficiency. Clin Exp Immunol 2012;167:27581.
  • 5
    Berglund LJ, Wong SW, Fulcher DA. B-cell maturation defects in common variable immunodeficiency and association with clinical features. Pathology 2008;40:28894.
  • 6
    Conley ME, Notarangelo LD, Etzioni A. Diagnostic criteria for primary immunodeficiencies. Representing PAGID (Pan-American Group for Immunodeficiency) and ESID (European Society for Immunodeficiencies). Clin immunol (Orlando, Fla.) 1999;93:1907.
  • 7
    Ferry BL, Jones J, Bateman EA, Woodham N, Warnatz K, Schlesier M, et al. Measurement of peripheral B cell subpopulations in common variable immunodeficiency (CVID) using a whole blood method. Clin Exp Immunol 2005;140:5329.
  • 8
    Reed GF, Lynn F, Meade BD. Use of coefficient of variation in assessing variability of quantitative assays. Clin Diagn Lab Immunol 2002;9:12359.
  • 9
    Belouski SS, Wallace D, Weisman M, Ishimori M, Hendricks L, Zack D, et al. Sample stability and variability of B-cell subsets in blood from healthy subjects and patients with systemic lupus erythematosus. Cytometry B Clin Cytom 2010;78:4958.
  • 10
    Levering WH, van Wieringen WN, Kraan J, van Beers WA, Sintnicolaas K, van Rhenen DJ, et al. Flow cytometric lymphocyte subset enumeration: 10 years of external quality assessment in the Benelux countries. Cytometry 2008;74:7990.
  • 11
    Sanchez-Ramon S, Radigan L, Yu JE, Bard S, Cunningham-Rundles C. Memory B cells in common variable immunodeficiency: clinical associations and sex differences. Clin immunol 2008;128:31421.