Lymphoproliferative disease in antibody deficiency: a multi-centre study

Authors


Dr M. M. Gompels, Consultant Immunologist, Immunology and Immunogenetics, North Bristol NHS Trust, Southmead Hospital, Bristol BS10 5NB, UK.
E-mail: mark.gompels@north-bristol.swest.nhs.uk

SUMMARY

We have undertaken a retrospective study of antibody deficient patients, with and without lymphoma, and assessed the ability of specific polymerase chain reaction (PCR) primers to determine if the detection of clonal lymphocyte populations correlates with clinical and immunohistochemical diagnosis of lymphoma. We identified 158 cases with antibody deficiency presenting during the past 20 years. Paraffin-embedded biopsy specimens or slides were available for analysis in a cohort of 34 patients. Of these patients, 29 had common variable immunodeficiency, one X-linked agammaglobulinaemia, one X-linked immunoglobulin deficiency of uncertain cause and three isolated IgG subclass deficiency. We have confirmed that lymphoma in antibody deficiency is predominantly B cell in origin. Clonal lymphocyte populations were demonstrated in biopsies irrespective of histology (16/19 with lymphoma and 11/15 without). Isolated evidence of clonality in biopsy material is therefore an insufficient diagnostic criterion to determine malignancy. Furthermore, our data suggest that clonal expansions are rarely the result of Epstein–Barr virus-driven disease.

INTRODUCTION

Primary antibody deficiencies are rare, the most frequent being common variable immunodeficiency (CVID), with an incidence of two per 100 000 people in Europe and North America [1]. Immunologically, CVID patients have low serum immunoglobulins (usually IgG and IgA) and recurrent pyogenic sinopulmonary infections. The risk of gastric carcinoma and of lymphoma is increased in these patients, compared to the general population [2,3]. The risk of lymphoma has been shown to be highest in female subjects in the 6th decade of life [2,4]. As with other immunosuppressed patients, most of the lymphomas have been reported to be of B cell phenotype and may be associated with Epstein–Barr virus (EBV). Few examples of T cell non-Hodgkin's lymphoma (NHL) exist.

The diagnosis of lymphoma in patients with CVID and other primary antibody deficiency conditions is particularly challenging. The histology of lymphoid tissue from antibody deficiency patients without suspected malignant change varies from being microscopically normal to grossly abnormal. Where the histology is abnormal the germinal centres can be absent or poorly developed, and there are none of the features of the normal reactive lymph node [5]. Furthermore, stimulation by an infection can result in an atypical lymphoid hyperplasia which, together with the lack of germinal centres, can be indistinguishable from lymphoma on ordinary light microscopy. There are often low numbers of B cells present with only some expressing surface immunoglobulin. The differentiation of these B cells may be markedly impaired, especially the IgG and IgA producing elements. A subgroup of patients with depletion of the naive CD4+ CD45RA+ T cell population have more severe clinical disease and in addition develop granulomatous changes in lymphoid tissue [6]. Up to 15% of CVID patients will develop granulomatous lymphadenopathy [1]. In X-linked agammaglobulinaemia (XLA) there are no germinal centres and lymphoid architecture is known to be grossly disturbed. In IgG subclass deficiency little is known about the lymphoid structures, although it is known that IgG levels do not correlate with infections.

The diagnosis of lymphoma rests on its histological characteristics and while immunochemistry has allowed greater interpretation, it is not definitive. Clonal rearrangements of IgH (immunoglobulin heavy chain) and TCR (T cell receptor) genes serve as markers of disease and are the hallmark of lymphoid tumours. However, the validity of clonality as a marker of lymphoma in antibody deficiency is not resolved, as monoclonal proliferations have been reported in these patients without progression [7,8]. Until recently antigen receptor gene rearrangement studies have been applicable only to fresh or frozen tissue, which has not always been available for analysis. The availability of polymerase chain reaction (PCR) techniques to amplify DNA extracted from paraffin sections has allowed us to undertake a retrospective study of biopsies from antibody deficient patients, with and without lymphoma, for the presence of T and B cell clones. The data have been correlated with the clinical and immunohistochemical diagnosis of lymphoma and the eventual clinical outcome.

MATERIALS AND METHODS

Identification of patients

The study was UK-based and conducted by a group of major centres involved in the management of patients with antibody deficiency. We identified a cohort of antibody deficient patients from (a) the UK audit of antibody deficiency [9,10] and (b) communication with major centres treating antibody deficiency and review of their case-load (see authorship and associated study group for participating centres).

In each case the diagnosis of antibody deficiency was confirmed to conform to international criteria (low immunoglobulins, recurrent infections, failure of antibody production) [11]. In order to exclude secondary cases, where a primary lymphoma resulted in an antibody deficiency, patients were included in the study only if there was a documented low immunoglobulin level in the 2 years prior to lymphoma, or a typical history of antibody deficiency (e.g. 2 years history of infections requiring multiple courses of antibiotics).

We identified 158 cases with antibody deficiency presenting during the past 20 years in whom a total of 273 biopsy reports on lymphoid tissue were traceable. Availability of suitable paraffin blocks or slides for PCR analysis reduced this to 64 samples from 34 patients. Twenty-nine patients had CVID, one XLA, one X-linked immunodeficiency of uncertain cause and three isolated IgG subclass deficiency.

Histological diagnosis

Morphological identification was performed on haematoxylin and eosin and Giemsa-stained sections [reviewed blind (B. A.)]. Lymphomas were classified according to the REAL classification, using a routine panel of antibodies [12]. Immunostaining was performed on 3–4 µm sections mounted onto coated slides, dried vertically at room temperature for 20 min and then at 60°C for 30 min. Sections were de-waxed, dehydrated and treated in 5% hydrogen peroxide in alcohol to block endogenous peroxidase activity. Antigen retrieval was performed according to the particular antibody to be used. The antibodies used included CD3, CD10, CD15, CD20, CD30, CD79a, Ki67, BCL-2, TIA-1 (slides pretreated in a pressure cooker with citrate buffer), βF1 (TCRβ) (slides pretreated with pronase), CD4, CD8 (slides pretreated in a pressure cooker with EDTA buffer), MIB1 (slides pretreated in a microwave with citrate buffer) and CD21, CD57, CD68, MAC387, S100 (slides pretreated with trypsin). Following pretreatment, sections were incubated for 10 min in 10% normal goat serum to reduce non-specific binding. Sections were incubated with primary antibodies diluted in the wash buffer (TRIS chloride pH 7·6) for 1 h. Antibody binding was detected using the streptavidin biotin detection system (Dako Duet, Dako Ltd, Ely, UK) and visualization of positive cells was achieved with DAB (BioGenex, San Ramon, CA, USA). Sections were counterstained with Carrazzi's haematoxylin.

Epstein–Barr virus

Formalin-fixed paraffin embedded sections were used for in situ hybridization (EBV polyprobe kit, Novacastra Laboratories, UK) to detect EBV-encoded short RNA species (EBER 1 and EBER 2) following the manufacturer's protocol. An EBV positive control, consisting of a known EBV positive case, and negative control were assayed in the same run.

PCR analysis of TCRG, TCRB and IgH gene rearrangements

DNA was isolated from paraffin wax embedded biopsy material as described previously [13]. TCRG PCR analysis was adapted from McCarthy et al. [14] and used primers to detect VG1, VGII, VGIII and VGIV genes in recombination with JG1·3/JG2·3 and JG1·1/JG2·1 genes. TCRB PCR analysis was performed using VBJB2 and DB1JB2 primer combinations [15]. TCRB and TCRG PCR products were electrophoresed through a 10% polyacrylamide gel. These TCR PCR methods were established using DNA from 16 clonal T cell tumours, characterized previously by Southern blot analysis, as clonal for TCRG and TCRB loci. In these experiments clonal TCRG gene rearrangements and TCRB VJ and DJ gene rearrangements were detected in 94% and 71% of cases, respectively (data not shown).

IgH FR2 PCR analysis was performed using a seminested PCR amplification method [16]. IgH FR3 PCR analysis was performed using FR3 and JH primers [17]. PCR products were electrophoresed through a 5% (FR2) or 10% (FR3) polyacrylamide gel. Our detection rate is 88% for demonstrating B cell clonality in a series of 40 immunohistologically characterized B cell neoplasms including 24 B cell leukaemias (22 B cell chronic lymphocytic leukaemia, one B cell prolymphocytic leukaemia and one hairy cell leukaemia) and 16 B cell non-Hodgkin's lymphoma (data not shown).

For all analyses, clonal bands are defined as one or two narrow sharp intense bands visible on polyacrylamide gels after electrophoresis, whereas polyclonal PCR products appear as a smear within the appropriate size range. Oligoclonal populations are defined by three or more distinct bands.

Sensitivity

DNA from known positive clonal controls were diluted serially (comprising 100, 50, 40, 30, 20, 10, 5, 2 and 1% of tumour DNA) with DNA from reactive polyclonal controls. Following appropriate PCR amplification, clonal products could be detected at a level of 2% clonal DNA on a polyclonal background smear for all PCR protocols.

Statistics

Data were analysed using Mann–Whitney U-test or Fisher's exact test, as appropriate. Differences were considered significant if P < 0·05.

RESULTS

We were able to assess tissue from 34 patients (18 female and 16 male), of whom 19 patients had progressed to lymphoma (follow up 1·2–16·0 years) and 15 had not (follow up 7·1–30·4 years). Of those patients with a lymphoma, 16 (seven female, nine male) had a diagnosis of CVID, two subclass deficiency (female) and one XLA (male). Of those patients without lymphoma, 13 (eight female, five male) had a diagnosis of CVID, one of IgG subclass deficiency (female) and one X-linked immunoglobulin deficiency of uncertain aetiology (male).

Figure 1 shows the age and gender of patients at time of diagnosis of lymphoma. The median age of diagnosis of lymphoma was 45 years (range 20–69); for females it was 46 years (range 20–50, n = 9) and males 42 years (range 26–69, n = 10) (not significant, P = 0·72). The median age at time of suspected lymphoma (as indicated by time of first biopsy) and diagnosis of CVID or subclass deficiency was comparable in women with and without lymphoma (Table 1). In men there was a significant difference between the age at first biopsy in those with and without lymphoma (P = 0·03 for all data, P = 0·04 if XLA excluded), but not at the age of diagnosis of antibody deficiency. This is likely to relate to a skewed distribution of ages as shown by the minimum age for antibody deficiency (6 versus 24 years, see Table 1).

Figure 1.

Age and gender at diagnosis of lymphoma. Data (*) include all patients presenting with lymphoma during the study period, including one patient in whom the histology was indicative of lymphoma, but equivocal (see text).

Table 1.  Analysis of age at diagnosis of antibody deficiency (XLA excluded), gender and lymphoma occurrence. Data (*) include all patients presenting with lymphoma during the study period, including two patients in whom the histology was indicative of lymphoma, but equivocal (see text)
GenderAge in yearsLymphoma
NoYes*All patients
FemaleMedian age (range) at
antibody deficiency
diagnosis
35 (18–49)34 (9–62)34·5 (9–62)
Median age (range) at
first biopsy
42 (18–49)46 (7–62)44·5 (7–62)
MaleMedian age (range) at
antibody deficiency
diagnosis
26 (6–40)37·5 (24–64)29 (6–64)
Median age (range) at
first biopsy
28·5 (9–40)42 (26–64)40 (9–64)
Median age (range) at
antibody deficiency
diagnosis
 27·5 (6–49)37 (9–64)34 (6–64)
Median age (range) at
first biopsy
 36 (9–49)45·5 (7–64)42 (7–64)

Diagnostic biopsies from 16 of the 19 patients with lymphoma were available for histological and immunohistological review (Table 2). Biopsies from 15/16 of these patients were confirmed to have histology consistent with lymphoma (Table 2). Biopsies from one patient were considered equivocal for lymphoma on review (patient 2). Lymphomas were predominantly of B cell origin (12 patients). In two patients the histology was consistent with T cell lymphoma. Diagnostic biopsies from the remaining three of 19 patients were not available; however, biopsies taken prior to the diagnosis of lymphoma were of reactive or normal histology (Table 3).

Table 2.  Clonality data and histopathology on biopsies from patients who had lymphoma
PatientGenderAge at biopsy
(years)
Site of biopsyHistology result (REAL
classification, where applicable)
Intensity of B
cell stain (CD20)
TCRBTCRGIgH
  1. EBV PD = Epstein–Barr virus proliferative disorder. LN = lymph node, BM = bone marrow, d = died from lymphoma. IgH gene rearrangements were assessed using FR2 and FR3 specific primers (see Methods). TCR gene rearrangements were assessed using VBJB, DBJB, VGJG1·3/2·3 and VGJG.1·1/2·1 primers. Code for interpretation F = no product, P = polyclonal, i.e. normal pattern of gene rearrangement, C = clonal and O = oligoclonal. Clonal for DBJB only, †clonal for VBJB only, ‡clonal for VGJG1·3/2·3 only, ^ clonal for VGJG.1·1/2·1 only, ∼ clonal for FR2 only, ‘’clonal for FR3 only. Intensity of B cell stain: + ++ strong, ++ moderate, + weak, 0 negative (absence of B cells), ND not done (B-cell numbers not available on immunostaining).

1dF34LungDiffuse large B cell lymphoma+ + +PPF
39MuscleDiffuse large B cell lymphoma+ + +PPC
2dM42BMPossible lymphoma+PPP
BMPossible lymphomaNDPP
3F45Lymphoid mass
invading muscle
Follicular and diffuse lymphoma,
low grade
+ + +PPC
4M69SpleenB cell lymphoma, low grade+ +PC**
5dF46LNHodgkin's disease+ +COP
BMHodgkin's disease+ +COF
LNHodgkin's disease+ +PP
LNHodgkin's disease+ +PP
6M61SpleenB cell lymphoma, low grade+ +PPC
7dM41Respiratory mucosa
with tumour
Anaplastic large cell lymphoma
T cell, high grade
+CPC
8dM26LNSmall B cell lymphoma+ + +PPF
GI tractUncertain, small biopsyNDC*CF
BMSmall B cell lymphoma+ + +PPF
LNSmall B cell lymphoma+ + +C*F
GI tract-rectalSmall B cell lymphoma+ + +PPF
9dF24LNNormal+ + +PP
30LN?EBV PD/Diffuse large B cell lymphoma+ + +PPP
LNDiffuse large B cell lymphoma+ + +PP
31LNDiffuse large B cell lymphoma+ + +PC
10dM34LNDiffuse large B cell lymphoma+CP
11F51LNDiffuse large B cell lymphoma+ + +PPC
12F47Lymphoid massProbable peripheral T cell lymphoma0POC
13dF49Lymphoid infiltrateDiffuse large B cell lymphoma+ +PPC
Lymphoid infiltrateDiffuse large B cell lymphoma+ +PCC
Lymphoid infiltrateDiffuse large B cell lymphoma+ +PP
Lymphoid infiltrateDiffuse large B cell lymphoma+ +PCC
Lymphoid infiltrateDiffuse large B cell lymphoma+ +PC
14dM60LNDiffuse large B cell lymphoma+ +PPC
15dF 7Respiratory mucosaNormal+PP
20BMLymphoplasmacytoid lymphoma,
B cell low grade
+ + +PC**
16dF63ParotidB cell lymphoma, low grade+ + +PPC
Table 3.  Clonality data and histopathology on biopsies from patients who developed lymphoma, but which was not evident at time of biopsy. Analysis and abbreviations as Table 2
PatientGenderAge at biopsy
(years)
Sites of biopsyHistology resultIntensity of B
cell stain (CD20)
TCRBTCRGIgH
17M43LNNormal0PPF
45LNNormalNDPPF
LNNormal0PPF
18dM52LNReactiveNDPPC
19M40LNReactiveNDPPP
LNReactiveNDPPP

Biopsies from 15 patients who did not develop lymphoma during the study period were consistent with reactive, normal or granulomatous histology (Table 4). B cells were detected in most, but not all, biopsies.

Table 4.  Clonality data and histopathology on biopsies from patients without lymphoma. Analysis and abbreviations as Table 2
PatientGenderAge at biopsy
(years)
Site of
biopsy
Histology result (REAL
classification, where applicable)
Intensity of B
cell stain
(CD20)
TCRBTCRGIgH
20M 9Respiratory − lungLymphoid hyperplasia+ + +PP
21F45GI tract − rectalNormal+ +PC‡
22M36GI tract − duodenalNormal+ + +PP
23F42LNNormal+ + +PPP
24F44SpleenReactive+ + +PPP
SpleenGranuloma+ + +PPP
LNReactiveNDPPP
25F41GI tract − duodenalNormal+PO
26M47LNGranuloma+ + +PP
LNGranuloma+ + +PP
27F64GI tract − duodenalReactive+ + +PCP
GI tract − gastricReactive+ + +POP
GI tract − gastricReactive+ + +POP
28F48Respiratory − lungGranulomaNDPP
53GI tract − jejunalNormal+ +PP
29M18GI tract − jejunalReactiveNDP
30F19LiverReactive0PPF
SkinReactive+ +PC‡F
31F27LNReactive+ +C*PP
32F35LNNormal+ + +PP
36SpleenReactive+ + +FPF
33M28GI tract − jejunalNormal0PPF
LNGranuloma+ + +PP
34M29SpleenGranuloma0PPF
LNNormalNDPP

EBV was sought by in situ hybridization in 17 biopsies and was not detected in either the lymphoma biopsies (10/17) or the nonlymphomatous biopsies (7/17).

DNA isolated from all biopsies was analysed for T and B cell clonality. Results for patients confirmed with lymphoma are summarized in Table 2. Clonal IgH products were found in 13 patients. In biopsies with B cell lymphoma IgH clonal products were found in 11/12 patients, with no product detected in 1/12 patients. Clonal IgH PCR products were also detected in two biopsies with T cell histology. Clonal TCR gene rearrangements were detected in nine patients (six with B cell lymphoma, one with T cell lymphoma and one with HD).

Results for DNA isolated from non-diagnostic biopsies from three patients who presented later with lymphoma are summarized in Table 3. Clonal IgH PCR products were detected in one patient.

Clonal analyses of isolated DNA from biopsies of patients without lymphoma are summarized in Table 4. Clonal products were found in biopsies for 11 patients. TCR clonal products were found in five patients and IgH clonal products in seven patients. Therefore, in biopsies from patients who did not have lymphoma at time of biopsy (Tables 3 and 4) IgH clonal products were seen in 12/18 patients.

The survival of those with lymphoma was poor, with only seven of 19 (37%) alive at follow-up, whereas of those without lymphoma all were alive at conclusion of the study (P = 0·0002).

DISCUSSION

In this report we have studied a large cohort of biopsies from 34 antibody deficient patients, predominantly with CVID, who were well characterized by both histology and immunophenotypic analyses. TCR and IgH clonality was determined in an attempt to delineate malignant from benign lymphoproliferations. Diagnosis of antibody deficiency was made on the basis of recurrent infections and low immunoglobulin levels. Care was taken in patient selection to exclude secondary cases, where a primary lymphoma resulted in an antibody deficiency.

The majority of our cases have occurred since 1980. Our study does not show the previous associations with regard to the age and gender at diagnosis of lymphoma [2,4]. In our population we have a greater occurrence of lymphoma in the male population than found previously [18]. We have also found that lymphoma is occurring at a slightly earlier age, with median age 46 years, in the female population (Fig. 1). Previous reports have found a strong association with female patients in their 50s [19,20]. With differences in management the presentation pattern of lymphoma may be changing. On the basis of our study, the risk of lymphoma must be considered raised in both sexes and in all age groups.

The diagnosis of lymphoma in CVID patients can be complicated by the fact that lymphoid tissue from these patients is histologically abnormal, with frequently absent or poorly developed germinal centres. Furthermore, stimulation by infection can result in histology that is difficult to differentiate from lymphoma. The rarity of CVID patients and the difficulty of histological interpretation may lead to erroneous diagnoses of lymphoma. On reviewing the available published literature on lymphoma in CVID subjects there have been two series published that looked specifically at the histological features. In these reports patients with antibody deficiency were selected from 1973 to 1990 [20] and 1959 to 1991 [21].

In the series of Cunningham-Rundle et al. [20], 10 of 117 CVID subjects developed lymphoma, of whom seven were B cell lymphomas. Nine of 10 were of diffuse morphology. These data were updated in a later report [18], extending the data to 19 NHL in 248 CVID patients. These were all B cell in type. The series also included three cases of Hodgkin's disease and one of Waldenstrom's macroglobulinaemia. Sander et al. [21] describe a series of 30 biopsies from 17 patients, with only two cases of lymphoma. In contrast, the biopsy data in the present study demonstrated 16/34 to have lymphoma, with 12/16 patients having B cell lymphoma (six classified as diffuse large B cell lymphoma) (Table 2). Three further reports include cases of CVID. Frizzera et al. [22], in a paper on the incidence of malignancy in 35 primary immunodeficiency patients, includes four CVID patients but with no breakdown as to type, EBV or BCL-2 status. A report by Filipovich et al. [23] from the immunodeficiency cancer registry reviews four cases of B cell lymphoproliferative disease in association with EBV in CVID patients who were treated with chemotherapy. Monoclonality was demonstrated in one case.

More recently, Canioni et al. looked at seven cases of lymphoma in a series of 28 children with lymphoproliferative disorders in primary or post-transplant immunodeficiency. Two CVID patients were included in the series but neither had lymphoma. Although clonal rearrangements were found in many of these cases, they were not considered helpful [24].

Association of EBV with lymphoid lesions in patients with secondary immunodeficiency is well documented; however, the situation with patients with primary antibody deficiencies is less clear. Our data suggest that lymphoma in antibody deficiency is rarely associated with EBV positivity. In the series by Sander et al. [21], described above, two of five cases were ultimately classified as lymphoma. EBV RNA was sought on one case and was positive. However, EBV RNA was also sought in four of the seven cases of atypical hyperplasia and found to be present in two cases. We found no cases positive in the present study. In the second series [20], EBV was not assessed and gene rearrangement studies were available in only a few cases.

We performed TCR and IgH PCR analysis on DNA isolated from paraffin-embedded biopsy specimens. PCR clonal products were found in 16 biopsies with lymphoma histology, but also in 13 biopsies with reactive histology. Clonality was found in 16/19 patients with lymphoma, or who later developed, lymphoma and in 11/15 patients not considered to have lymphoma (P = 0·72) (see Table 2). There were no differences between the two groups when TCRB, TCRG and IgH clonality were considered independently (respective P-values 0·50, 0·80 and 0·21). Therefore, in this study the presence of clonal populations does not appear to be a reliable index of malignancy. This finding is similar to a recent study of a heterogeneous group of children with primary immunodeficiencies, where TCR and IgH rearrangements were not helpful in distinguishing polymorphic lymphoproliferative disorders from lymphomas associated with EBV [24].

One of the limitations of using PCR for clonality studies is that amplification of DNA from restricted T or B cell populations can result in pseudoclonality due to low representation of target lymphoid DNA. This is particularly relevant to primary antibody deficiencies where B cell numbers may be low (i.e. patients 12, 17, 30, 33 and 34), especially in the analysis of biopsies with reactive histology. All biopsies were reviewed after completion of immunohistology. Significant B cell populations, where sufficient material was available for assessment, were found in 36/38 biopsies from lymphoma patients (Tables 2 and 3) and 13/15 biopsies from non-lymphoma patients (Table 4). Therefore, the occurrence of clonal or oligoclonal populations does not appear to be the result of low target numbers in the majority of biopsies. Thus, in patients without lymphoma, clonality may be demonstrated in those with adequate B cell numbers, excluding B cell deficiency per se as the cause of clonality in this population. Furthermore functional B cell clones have been confirmed by sequence analysis in six cases with lymphoma and four cases without, and the B cell origin of these B cell clones is being investigated by somatic mutational analysis (data not shown, manuscript in preparation).

The presence of oligoclonal populations in some patients is of interest. T and B cell oligoclonal populations have been well documented in a variety of autoimmune disorders [25,26]. Furthermore, oligoclonality has been observed in primary immunodeficiencies such as Omenn's syndrome, where a link with autoimmunity has been proposed [27,28]. Patients with CVID often have autoimmune manifestations. However, because ours was a retrospective case series focusing on those patients with suspected lymphoma, it was considered inappropriate to consider the autoimmune status of our patients. We feel such a study would be best conducted prospectively.

It would appear that clonal lymphocyte populations occur in CVID patients without lymphoma and furthermore we have shown in some cases that clonal population may precede the development of lymphoma. In fact the mortality of patients with lymphoma and CVID is high. In our series we observed a mortality of 63% during a follow-up period of 1·2–16·5 years.

In conclusion, we have confirmed that lymphoma in CVID is predominantly B cell in origin. However, evidence of clonality in biopsy material is insufficient evidence to determine malignancy as clonal expansions are also found in reactive tissue. Furthermore, our data suggest that these are rarely the result of EBV associated disease.

ACKNOWLEDGEMENTS

This study was supported by grants awarded by the Primary Immunodeficiency Association and the Cancer Research Council, which the authors acknowledge with thanks. Furthermore the authors would like to thank Dr Karan-Jane Palmer, Susan Harris and Dr Louise Lavender for their technical assistance.

Ancillary