In non-follicular lymphoproliferative disorders, IGH/BCL2-fusion is not restricted to chronic lymphocytic leukaemia

Authors


Correspondence: Pascale Felman, Laboratoire d'Hématologie Cellulaire-Centre, Hospitalier Lyon-Sud, Pierre Benite Cedex 69495, France.

E-mail:pascale.felman@chu-lyon.fr

Summary

The translocation t(14;18) and its t(2;18) and t(18,22) variants, which involve the BCL2 genetic hallmark for follicular lymphoma (FL), have been reported in several cases of chronic B-cell lymphoproliferative disease (CLPD) and frequently in chronic lymphocytic leukaemia (CLL). We describe here the clinical, morphological, immunological, cytogenetic and molecular findings from 37 cases of t(14;18)-positive CLPD, identified from our series of non-FL B-cell neoplasms (n = 993) that were routinely analysed in peripheral blood by conventional cytogenetics analyses. The FL diagnosis was excluded by morphology and immunology (the samples were CD10 negative in all cases). The BCL2 translocations were observed in 22 CLL cases, including 7 monoclonal B-cell lymphocytosis (MBL) cases re-classified according to the new International Workshop on CLL criteria, six small lymphocytic lymphoma (SLL) cases, 1 splenic marginal zone lymphoma (SMZL) case and eight cases of unclassifiable CLPD with overlapping CLL/MZL features. In the CLL cases, the IGH/BCL2 fusion was remarkably associated with trisomy 12 (13/22) and mutated IGHV status (20/21) and did not affect the outcome. Moreover, most of these CLLs harboured a low mutation load of BCL6 gene and unmutated FAS (CD95) loci, which points to a post–germinal-centre cellular origin.

Reciprocal chromosomal translocations, especially those involving the immunoglobulin genes, are uncommon though recurrent in chronic lymphocytic leukaemia (CLL) (Döhner et al, 2000). Different partner genes have been identified and included, particularly the BCL3 and BCL2 genes (Cavazzini et al, 2008). The t(14;19) translocation involving BCL3 is widely described in the literature and is often associated with atypias in morphology as well as in immunological profile, with unmutated IGHV genes and a poorer prognosis than common CLL (Huh et al, 2007; Martín-Subero et al, 2007; Cavazzini et al, 2008; Chapiro et al, 2008). The involvement of BCL2 for balanced immunoglobulin gene translocations in CLL has been less frequently reported (Sen et al, 2002; Lau et al, 2008; Put et al, 2009). Different clinical behaviours have been reported, which may be related to grouping the t(14;18) and t(14;19) translocations together in one report (Lau et al, 2008). The t(14;18)(q32:q21) translocation is considered to be a genetic hallmark of germinal centre (GC)-derived B-cell lymphomas and especially of follicular lymphoma (FL). The IGH/BLC2 translocation is observed in 80–90% of the all the FL cases analysed by conventional cytogenetics. The observation that some CLLs carry t(14;18) raises the question about their cellular origin, particularly whether they have transited through the GC, which is still debated (Chiorazzi & Ferrarini, 2011). In addition to t(14;18), aberrant somatic mutations found in the BCL6 and/or FAS (CD95) genes have been mentioned as evidence of cell transit through the GC (Grønbaek et al, 1998; Pasqualucci et al, 1998; Capello et al, 2000). Thus, identifying such mutations could provide additional information about the cellular origin of CLL with t(14;18).

In contrast to the BCL3 translocations that have been reported in CLL and in a heterogeneous group of B-cell malignancies (Huh et al, 2007; Martín-Subero et al, 2007; Chapiro et al, 2008), the IGH/BCL2 fusion, other than FL and DLCBL, has been described in only a few CLL series (Sen et al, 2002; Cavazzini et al, 2008; Lau et al, 2008; Put et al, 2009) and in occasional cases of mucosa associated lymphoid tissue (MALT) lymphomas (Nakamura et al, 2007; Palmedo et al, 2007). To more clearly define whether the IGH/BCL2 translocation is restricted to CLL or whether it may be observed in other chronic B-cell lymphoproliferative diseases (CLPD) with usual or frequent disseminated presentation, we analysed the prevalence of IGH/BCL2 translocations using conventional cytogenetics of a consecutive series of non-FL B-cell neoplasm studied in peripheral blood. In addition to the cytohistological, immunological, cytogenetic and clinical features, we studied the mutational status of the IGHV, BCL6 and FAS genes to precisely determine whether t(14;18)-positive CLL cases may have a particular histogenesis.

Materials and methods

Patients and samples

IGH/BCL2 translocations were identified in 37 cases from a consecutive series of non-FL CLPD (n = 993) cases routinely studied between 1997 and 2010 by morphological, immunophenotypical and conventional cytogenetic analyses. Either t(14;18)(q32;q21) or its variants, t(2;18)(p11;q21) and t(18;22)(q21;q11), were identified. The presence of the BCL2 gene was confirmed by fluorescent in situ hybridization (FISH) analysis. An FL diagnosis was excluded by morphology: the absence of centrocytes in peripheral blood (PB) or bone marrow smears (BM) and a CD10-negative immunophenotype (in all cases), followed by confirmation by BM (n = 11) or lymph node (LN) biopsy histology (n = 7). The clinical data were provided by the referring clinicians.

The initial staging procedures included age, stage and sites of involvement (PB count with cytological analysis), and the serum levels of lactic dehydrogenase (LDH) and β2–microglobulin. The CLL patients were staged according to the Binet system, and all treatment decisions were made according to National Cancer Institutes (NCI) criteria (Hallek et al, 2008).

Cytological review

The cytology was reviewed on available stored PB films (n = 33), BM smears (n = 3), or spleen imprints (n = 1) by two observers. Freshly made blood films stained with May-Grünwald-Giemsa were made at diagnosis in all cases but one. A manual 100-cell differential lymphocyte count was performed, and the cells were classified into three categories: CLL-type lymphocytes; atypical cells including lymphoplasmacytoid cells, cells with indented nuclei or cells with other abnormalities; and prolymphocytes.

Histological review and immunohistochemistry

Formalin- or Bouin-fixed, paraffin-embedded LN (n = 7) skin nodule (n = 1) and splenic tissue (n = 2) sections (haematoxylin/eosin stained) were reviewed by expert haematopathologists according to the current criteria (Isaacson et al, 2008; Mûller-Hermelink et al, 2008), as were resin-embedded BM specimens (n = 11). Standard immunohistochemical techniques were performed on the paraffin-embedded tissues according to the manufacturer's instructions, including heat-induced antigen retrieval and avidin-biotin-peroxidase detection using an automated immunostainer (Benchmark, Ventana Medical System/Roche Diagnostic GmbH, Mannheim, Germany). Appropriate positive and negative controls were run for each antibody. The immunohistochemical panel included B cell markers (L26/CD20*, CD79a, and clone HM57*), T cell markers (polyclonal CD3*, CD5, and clone 4C7**), anti-CD23 (clone 1B12**), anti-CD43 (clone MT1**), anti-BCL6 (clone PG-B6P*), and anti-CD10 (clone W8E7***) (*Dako, Carpinteria, CA, USA; **Novocastra, Lb Co. Ltd., Newcastle, UK; ***Becton Dickinson, San Jose, CA, USA).

Immunophenotype analysis by flow cytometry

Whole PB alone (n = 34), PB and LN (n = 1), or PB and spleen cell suspensions (n = 2) were stained using 2, 3 or 6 combinations of directly conjugated monoclonal antibodies. The samples were analysed on a Coulter EPICS XL flow cytometer (analysis with System II software or Expo32 ADC software**** for samples from 1997 to 2006) and on a BD FACSCanto II cytometer (analysis with BD FACS DIVA software*** for samples from 2006 to 2010). The following major antibodies were used: CD19 (HD37*, fluorescein isothiocyanate/phycoerythrin [FITC/PE]; J4·119****, phycoerythrin-Texas Red conjugate [energy coupled dye, ECD]; SJ25C1****, peridinin chlorophyll-A protein-Cyanin 5·5 [PerCP-Cy5·5]), kappa (FITC)*, lambda (PE)*, CD5 (DK23*, PE; BL1a***, PE-Cy7), CD10 (HI10a***, PE-Cy7), CD23 (MHM6*, FITC; EBVCS-5***, allophycocyanin [APC]), FMC7 (FITC)*, CD22 (4KB128*, PE), CD43 (DF-T1*, FITC), CD79b (SN8*, PE; CB31*, PE; SN8***, APC) CD20 (Bly1* FITC; L27***, APC-Cy7) and CD38 (HIT2***, PE) (*Dako,***Becton Dickinson and ****Beckman-Coulter, Hialeah, FL, USA).

The immunophenotype flow cytometry methods have been previously reported (Traverse-Glehen et al, 2008). Having 30% or greater positive cells was considered to be positive for all markers. The fluorescence level for the different parameters was also measured in arbitrary units, such as the mean fluorescence intensity (MFI). All of the cases were scored using the scoring system established for diagnosing CLL (Matutes et al, 1994).

Cytogenetic and FISH analyses

In all cases, the cytogenetic studies were performed as previously reported (Callet-Bauchu et al, 1999), and abnormalities were described according to the International System for Human Cytogenetic Nomenclature (ISCN) (Mitelman, 1995) Additional FISH experiments were performed on each of the patients according to standard methods, as has been previously reported (Gazzo et al, 2003). The following specific Vysis probes (Abbott Molecular, Des Plaines, IL, USA) were used: LSI ATM (11q22·3), TP53 (17p13·1), DS13S319 (13q14·3), CEP 12 (D12Z3), and the LSI IGH/BCL2 dual-fusion translocation probe.

Molecular analysis

Genomic DNA was extracted from the PB cells (n = 33) and spleen samples (n = 1) using the High Pure polymerase chain reaction (PCR) Template Preparation kit (Roche, Mannheim, Germany). The mutational status of the IGHV genes was determined by sequencing analysis, as has been previously described (Traverse-Glehen et al, 2005). The genes were considered to be mutated when they displayed <98% homology with the closest germline IGHV gene.

The BCL6 (region located downstream to the first noncoding BCL6 exon) and FAS (exon 8, exon 9 and 5′UTR) nucleotide sequence analyses were obtained by directly sequencing the amplified PCR fragments in 23 cases, as has been previously described (Traverse-Glehen et al, 2007).

Statistical analysis

The treatment-free survival (TFS) was defined as the time between the diagnosis and first-line treatment, and overall survival (OS) was defined as the time between the first diagnosis and death or the last follow-up. The Chi-square test was used to compare the frequencies of the cytogenetic abnormalities and the IGHV mutational level. A P value of < 0·05 was considered to be significant.

Results

All of the results are summarized in Table SI.

Sub-classification of cases

According to the cytohistology, immunophenotype, the International Workshop on CLL (IWCLL) 2008 guidelines (Hallek et al, 2008) and the World Health Organization current criteria (Mûller-Hermelink et al, 2008), these 37 cases were classified as follows: 22 CLL cases that were characterized by a Matutes score ≥3 in all cases but one, which had a score of 1 but was diagnosed as CLL by the LN biopsy sections; six small lymphocytic lymphoma (SLL) cases; and one case of splenic marginal zone lymphoma (SMZL) that was proven from a tissue specimen (Case 37). Eight cases were considered to be unclassifiable (U-CLPD) in the absence of convincing criteria, despite available tissue biopsies in two cases. Among the 22 CLL cases, seven cases, which presented with fewer than 5·00 × 109/l circulating clonal B lymphocytes, as demonstrated by immunophenotyping, without any lymphadenopathy and organomegaly (as defined by physical examination or computed tomography scans), and without cytopenias or disease-related symptoms, were re-classified as ‘monoclonal B-lymphocytosis’ (MBL) according to the new IWCLL criteria (Hallek et al, 2008). In the past, these seven cases would have been considered to be early stage CLL (Rai 0/I and Binet A). Among them, three developed CLL during the follow-up.

For better clarity, we decided to present our results in three parts. The first part includes (i) CLL/MBL cases that can be compared with CLL without a IGH/BCL2 translocation (referred in the present text as common CLL) and with previously reported cases of CLL with IGH/BCL2 translocation (Sen et al, 2002; Lau et al, 2008; Put et al, 2009), and (ii) SLL cases. The second part relates to the SMZL case, and the third consists of the U-CLPD cases.

CLL/MBL/SLL group

Clinical findings

There were 28 patients (16 men and 12 women), with a median age of 70·0 years (range, 50·0–86·0 years). In the CLL group (n = 15), all of the patients had a B-cell count higher than 5·00 × 109/l (median 18·46 × 109/l; range 6·21–178·60 × 109/l), which was in contrast to the 6 SLL (median 1·26 × 109/l; range 0·57–3·76 × 109/l) and 7 MBL patients (median 4·17 × 109/l; range 1·49–4·56 × 109/l). Five CLL patients had lymphadenopathy, and three had splenomegaly (two had both lymphadenopathy and splenomegaly). All of the SLL patients had lymphadenopathy, and one presented with a skin nodule. None of the SLL cases had splenomegaly. By definition, the MBL patients presented with neither lymphadenopathy nor splenomegaly. No CLL/MBL/SLL cases presented with anaemia (haemoglobin < 110 g/l). Only one CLL case presented with thrombocytopenia (platelet count <100·0 × 109/l). A monoclonal component was detected in 3 of the 13 CLL cases (2 IgG kappa, 1 IgG lambda), in the SLL case (IgA kappa) and in one of the MBL cases (IgM kappa).

Of the CLL cases, 12 presented with stage A (80%), two with stage B (13%) and 1 with stage C (7%).

Cytohistological results

The cytological features indicated a CLL diagnosis in all cases but one, which appeared borderline between CLL and MZL (Table 1, Fig 1). The cytology was typical (small lymphocytes with scanty cytoplasm and regular nuclei with clumped chromatin) in 13 cases. Eight cases with atypical cells (lymphoplasmacytoid cells or cells with indented nuclei) above 20% were considered as atypical CLL. Increased prolymphocytes were observed in two atypical CLL cases (12% and 15%) and in the borderline CLL/MZL case (15%). In the majority of SLL cases (5/6), the cytology was atypical, and one case showed prolymphocytes (10%). The BM smears showed infiltration in the 14 available cases (approximately 80% in the CLL cases and 30–50% in the SLL cases). The pattern of BM biopsy infiltration was nodular, with an associated interstitial component in all of the cases (n = 7) and with proliferation centres in two cases. True paratrabecular infiltrates, such as those seen in FL, were absent in all of the cases.

Figure 1.

Peripheral blood smears (May Grünwald Giemsa staining, ×100): Chronic lymphocytic leukaemia/monoclonal B-cell lymphocytosis (CLL/MBL) cases with typical CLL cytology (A1–3): small lymphoid cells with scanty cytoplasm and regular nuclei with clumped chromatin. Cases with atypical CLL cytology: nuclear indentation (B1–3), lymphoplasmacytoid cells (C1–3), and prolymphocytes (D). Lymph node CLL sections (haematoxylin-eosin-safranin staining, low magnification ×4): diffuse involvement with a typical proliferation centre (E, arrow).

Table 1. Morphological data
 CLL/MBL (n = 22) with IGH/BCL2 fusionSLL (n = 6) with IGH/BCL2 fusionU-CLPD (n = 8) with IGH/BCL2 fusionSMZL (n = 1) with IGH/BCL2 fusion
CLL (n = 15)MBL (n = 7)
  1. CLL, chronic lymphocytic leukaemia; MBL, monoclonal B-cell lymphocytosis; SLL, small lymphocytic lymphoma; U-CLPD, unclassified chronic B-cell lymphoproliferative disease; SMZL, splenic marginal zone lymphoma.

  2. a

    Hilar lymph node.

Cytology
Typical CLL8/155/71/63/8 
Atypical CLL6/152/75/6 
Borderline CLL/MZL1/15005/8 
MZL00001/1
Histology
Lymph nodeCLL (2/2) SLL (5/5)Unclassified (1/1)SMZLa
Spleen   Unclassified (1/1)SMZL

An LN biopsy was performed in two of the CLL cases, which were then concluded to be CLL due to effaced architecture by lymphoid infiltration with a diffuse growth pattern and some proliferation centres. The histology was similar in all of the SLL cases.

Immunophenotypical results

All of the CLL/MBL cases except two presented with a Matutes score of at least 4 (Table 2). One CLL case showed a score of 3, but CD5 and CD23 co-expression and dim CD20 expression enabled a CLL diagnosis despite strong SIg intensity (Rawstron et al, 2002; Morice et al, 2008; Matutes et al, 2010). Surprisingly, Case 15 presented with a score of 1 (only CD5 positivity), but the proposed CLL diagnosis, based on the PB smear, was proven on the LN biopsy. A lack of CD5 and CD23 was observed in 1 and 4 cases, respectively. CD20 was dimly expressed in 11 cases. In seven cases, the CD20 expression was defined as intermediate because it was stronger than that of common CLL but dimmer than that of other small B-cell lymphomas, such as MZL, mantle cell lymphoma (MCL) and benign B-cells. However, this finding remains consistent with a CLL diagnosis (Rawstron et al, 2002). In three of the cases, the CD20 expression was strong. Strong CD43 expression, which is another useful marker for CLL (Matutes et al, 2010), was present in 16 cases. All of the cases were CD27 positive.

Table 2. Immunological data
 CLL/MBL (n = 22) with IGH/BCL2 fusionSLL (n = 6) with IGH/BCL2 fusionU-CLPD (n = 8) with IGH/BCL2 fusionSMZL (n = 1) with IGH/BCL2 fusion
CLL (n = 15)MBL (n = 7)
  1. CLL, chronic lymphocytic leukaemia; MBL, monoclonal B-cell lymphocytosis; SLL, small lymphocytic lymphoma; U-CLPD, unclassified chronic B-cell lymphoproliferative disease; SMZL, splenic marginal zone lymphoma.

  2. *Matutes score = 1, but CLL histology on lymph node biopsy.

Matutes score
4–513/157/75/60
31/1501/61/8
<31/15a007/81
CD5+/CD23+10/157/76/62/8

The immunological profile of the SLL cases was typical for CLL (CD5+/CD23+/Fmc7-/CD22dim/CD79bdim/SIgdim) in all of the cases.

Cytogenetic results

A t(14;18) was observed in 18 of the CLL/MBL cases (82%), a t(18;22) was observed in three cases (14%) and a t(2;18) was observed in one case (4·5%) (Table 3). The t(14;18) translocation or variant was the sole aberration in three cases (14%). Trisomy 12 was observed at a high frequency (13/22; 59%) and was more frequent than in common CLL (16%) (< 0·05) (Döhner et al, 2000); however, del(13q) was observed at a similar frequency (seven cases, 33% vs. 55%). Its incidence was higher in CLL (11/15; 73%) than in MBL (2/7; 28%). Two cases showed an association between trisomy 12 and del(13q). No cases presented with del(11q) or del(17p).

Table 3. Cytogenetic data
 CLL/MBL (n = 22) with IGH/BCL2 fusionCommon CLLa (%)
  1. CLL, chronic lymphocytic leukaemia; MBL, monoclonal B-cell lymphocytosis.

  2. a

    From the CLL series reported by Döhner et al (2000).

  3. b

    (Chi-square test; < 0·05).

Trisomy 1213/22 (59%)b53/325 (16)b
del(13q)7/22 (33%)178/325 (55)
del(11q)0/2258/325 (18)
del(17p)0/2223/325 (7)

Only the t(14;18) was observed in the six SLL cases; it was the sole abnormality in two cases (33%). Trisomy 12 was observed in most cases (4/6). A del(13q) was identified in only one case. There were no cases with del(11q) or del(17p).

In 6 of the 17 CLL/SLL cases bearing t(14;18) and trisomy 12, 2 different clones were identified by the cytogenetic analysis. The t(14;18) was the primary change in three cases (three CLL). In three others (1 CLL, 1 SLL, and 1 MBL), it appeared in a derived clone in addition to the primary abnormality (trisomy 12 or others), and thus represented a secondary event.

Molecular results

IGHV mutation analysis was performed in 21 cases of CLL/MBL (Table 4). In one case, the IGHV sequences showed 100% sequence identity to the germline IGHV genes. Twenty cases (95%) presented with a mutation status that was homologous to the germline sequence by 81·6–96·0% (median 92·1%). Thus, the IGH/BCL2-postive CLL/MBL appeared mutated more often than has been previously reported in common CLL cases (55%) (< 0·05) (Hamblin et al, 2009). The most frequent IGHV family used was IGHV3 (17/21, 81%). In the SLL cases (five studied cases), there were two mutated and three unmutated cases.

Table 4. Molecular data
 CLL/MBL with IGH/BCL2 fusionCommon CLLaGC-derived B-cell lymphomasaSLL with IGH/BCL2 fusionU-CLPD with IGH/BCL2 fusionSMZL with IGH/BCL2 fusion
CLLMBLFLDLCBL
  1. CLL, chronic lymphocytic leukaemia; MBL, monoclonal B-cell lymphocytosis; SLL, small lymphocytic lymphoma; U-CLPD, unclassified chronic B-cell lymphoproliferative disease; SMZL, splenic marginal zone lymphoma; M, mutated; UM, unmutated.

  2. a

    Data from CLL without t(14;18) and GC-derived B-cell lymphoma have been previously reported in the literature (Grønbaek et al, 1998; Hamblin et al, 1999; Capello et al, 2000).

  3. b

    (Chi-square test, < 0·05).

  4. c

    FL and DLCBL cases with M-BCL6.

  5. d

    Unusual CLL case with skin tropism (study of FAS mutations) (Grønbaek et al, 1998).

M-IGHV13/14 (92%)b7/7 (100%)46/84 (55%)b2/5 (40%)8/8 (100%)1
UM-IGHV1/14 (8%)0/738/84 (45%)3/5 (60%)0/8
BCL6
M-BCL6/M-IGHV8/12 (65%)2/5 (40%)10/28 (35%)37–47%c40–80%c1/3 (33%)
UM-BCL6/M-IGHV4/12 (35%)3/5 (60%)5/28 (19%)1/2 (50%)2/3 (66%)1
UM-BCL6/UM-IGHV13/28 (46%)1/2 (50%)
M-FAS0/140/61/17 (6%)d6%21%000

Mutations in the BCL6 gene were detected in 8/12 CLL cases and were represented by single nucleotide substitutions in all cases (1–3 mutations). These eight cases harboured mutations in both the BCL6 and IGHV genes, and 4 CLL cases displayed mutations in the IGHV gene in the presence of germline BCL6 gene, as has already been reported for CLL without t(14;18) (Capello et al, 2000). The single unmutated IGHV CLL case was not available for BCL6 mutation analysis. Of the five available MBL cases, all of which had mutated IGHV, the BCL6 gene was mutated in two cases and unmutated in three cases. The BCL6 gene was unmutated in the two available SLL cases; IGHV was mutated in one case and unmutated in the other. No FAS mutations were detected in any of the available CLL/MBL/SLL cases.

Outcome

Of the 20 available CLL cases, nine were untreated. The other 11 patients (55%) received first-line treatment with chemotherapy and immunotherapy. Of these 11 patients (54%), 6 did not respond to their first-line therapy and received 2 or 3 additional lines of treatment. In the SLL cases, five patients (83%) received chemotherapy ± immunotherapy, and four cases received two lines of therapy. In the CLL/MBL patients, the median TFS was 39 months (range 0–104), which did not appear to be different from the TFS of CLL with a normal karyotype (49 months) or CLL with trisomy 12 only (33 months) (Döhner et al, 2000).

The median follow-ups of the CLL/MBL and SLL groups were 53·5 months (range 2·0–269·0) and 39·0 months (range 7·0–194·0), respectively. Eight patients in the CLL/MBL group died, with a survival range of 20·0 to 269·0 months (median 55 months). In the SLL group, two patients died (7·0 and 194·0 months after the initial diagnosis).

SMZL case

The case classified as SMZL involved a 59-year-old man who presented with splenomegaly, lymphadenopathy with anaemia (105 g/l), thrombocytopenia (106·0 × 109/l) and a high B-cell count (181·51 × 109/l). No monoclonal component was observed. After splenectomy, the patient was treated by immuno-chemotherapy (R-CHOP; rituximab, cyclophosphamide, doxorubicin, vincristine, prednisolone) and is still alive 13 months after the diagnosis.

The PB smear favoured the MZL diagnosis. It showed heterogeneous lymphoid cells (small lymphocytes with chromatin condensed in small clumps, lymphoplasmacytoid cells and 10% villous lymphocytes). The SMZL diagnosis was confirmed by hilar LN histology, which showed important involvement by small CD5, CD10 and BCL6 negative B-cells with a nodular pattern and without proliferation centres. The spleen section slides showed an enlarged marginal zone (Fig 2).

Figure 2.

Peripheral blood smears (May Grünwald Giemsa staining, ×100): The splenic marginal zone lymphoma (SMZL) case: a majority of small lymphoid cells with densely packed chromatin in small irregular clumps (F1–3). Lymph node and spleen sections (haematoxylin-eosin-safranin staining, low magnification ×4 and ×20; CD5 immunostaining): A hilar lymph node section for the SMZL case: involvement by CD5-negative B-cells with a nodular pattern (G, H) and without a proliferation centre (I); typical enlargement of marginal zone can be seen in the spleen section (J).

The immunological profile (in the PB and spleen suspension) corresponded to a typical SMZL profile (CD5-/CD23-/FMC7+/CD22bright/CD79bbright/SIgbright: a Matutes score of 0) with strong CD20 expression.

No recurrent SMZL abnormalities (Callet-Bauchu et al, 2005; Matutes et al, 2008), such as del(7q), trisomy 3 or trisomy 18, were detected.The IGHV gene was mutated (87·9% of homology). The IGHV3-23*01 or *04 gene was involved, as has already been reported for SMZL (Arcaini et al, 2009). This case displayed unmutated BCL6 gene and FAS loci, as expected (Mateo et al, 2001; Traverse-Glehen et al, 2007).

Altogether, the data were consistent with the SMZL diagnosis.

U-CLPD group

Clinical findings

There were 4 men and 4 women, with a median age of 71·5 years (range, 58·0–83·0 years), in the unclassifiable group. The median B-cell count was 12·10 × 109/l (range 2·99–34·16 × 109/l). Three patients had lymphadenopathy, and one had splenomegaly. The median haemoglobin concentration was 132 g/l, and was <110 g/l in one case. The median platelet count was 228·0 × 109/l without any cases <100·0 × 109/l. In the available cases (n = 6), no monoclonal component was detected.

Cytohistological results

In five cases (four blood, one spleen), the morphology did not allow a true distinction between MZL and CLL, whereas in the three other cases, the picture suggested CLL (Table 1). Three cases showed increased prolymphocytes (one case with CLL features). The BM smears (n = 4) revealed that the extent of involvement varied from 30 to 80% (median = 50%). The histology of the spleen and LN sections in Cases 30 and 31, respectively, did not contribute to the diagnosis; they showed a diffuse pattern of infiltration without proliferation centres.

Immunophenotypical results

The Matutes score was <3 in all of the cases but one (score 3, Case 36) (Table 2). This case displayed a non-CLL profile (CD23-/CD22+bright). In 3 cases (Cases 32, 33, and 35), the profile was consistent with MZL, showing a lack of CD5 and CD23 expression and a strong CD20 and/or SIg expression (two cases with borderline CLL/MZL cytology and one case with CLL cytology). In two cases (Cases 30 and 31), the profile was ambiguous, showing a strong CD20 and/or SIg expression but a CD5/CD23/CD43 co-expression that has been proposed as an atypical CLL profile (Rawstron et al, 2001). Two other cases (Cases 29 and 34) were CD5 positive, but in the absence of other CLL-typical features (such as CD23 positivity and a dim expression of SIg or CD20), the profile was more consistent with CD5 +  SMZL (Baseggio et al, 2010). If the presence of CD5, lack of CD23 and strong CD20 or SIg expression is generally taken as de facto evidence of MCL, no case presented with a cyto-histology evocative of MCL. In addition, no t(11;14) was identified. In summary, the immunological pattern in these U-CLPD cases overlapped with the CLL and MZL phenotypes but was closer to that of MZL in most cases (n = 5).

Cytogenetic results

A t(14;18) was detected in six cases (75%), and a t(18;22) was detected in two cases (the sole abnormality in one case). No case showed a t(2;18). Trisomy 12 was observed in seven cases (87,5%), and del(13q) was observed in three cases; trisomy 12 was found in each case. There were no cases with del(11q) or del(17p). None of the recurrent abnormalities seen in SMZL, such as del(7q), trisomy 3 or trisomy 18, were observed. No t(11;14)(q13;q32) was detected by karyotype analysis. In the two cases showing an association between trisomy 12 and t(14,18) in 2 different clones, the primary change was trisomy 12 in one case and t(14;18) in the other case.

Molecular results

All of the available U-CLPD cases (n = 7) were mutated, with a median IGHV sequence identity with the IGHV germline of 92·4% (range 83·9–96·4%) (Table 4). The IGHV3 segment was the most frequent IGHV family used (5/7; 71%) without selective IGHV gene usage.

Only one case (Case 32) had a mutated BCL6 gene, with 6 different mutations associated with a mutated IGHV. By contrast, the other three cases displayed a mutated IGHV but a BCL6 gene in the germline configuration. No FAS mutations were detected in any of the available cases.

Outcome

In the available cases (n = 6), all of the patients were treated by chemo- and immunotherapy. Four patients achieved a partial response to the first-line therapy, and two received second-line therapy. The median follow-up of this group was 66·5 months (range 13·0–156·0 months). Following treatment, four patients were alive, and three others had died (mean = 89·0 months, range 75·0–156·0 months).

Discussion

In this series of non-FL small B-cell lymphoma/leukaemia cases, the IGH/BCL2 rearrangement was rarely identified (3%); most cases corresponded to CLL/MBL/SLL (76%). In agreement with previous series (Sen et al, 2002; Put et al, 2009), the IGH/BCL2 fusion was remarkably associated with trisomy 12, and this association appeared more frequently than in common CLL series. The atypical cytology (increased number of lymphoid cells with irregular nuclear contours, plasmacytoid features or prolymphocytes) and/or an atypical CLL profile (lack of CD23 and intermediate/strong CD20 expression) is probably linked to trisomy 12, as has been previously reported (Matutes et al, 1996). This hypothesis is reinforced by our observation of the isolated t(14;18) CLL/SLL cases (n = 4), which showed a typical cytology in all of the cases but one (Case 24) and a typical immunologic profile in all cases but one (Case 7, lack of CD5). By contrast, the mutated IGHV status in the present CLL series (95% of CLL cases) appeared to result from the t(14;18) rather than from the trisomy 12, given that most of our CLL cases with isolated trisomy 12 were unmutated (65%) and belonged to a CLL series with an association between t(14;19) and trisomy12 (Martín-Subero et al, 2007; Cavazzini et al, 2008; Chapiro et al, 2008). Unlike the IGH/BCL3 CLL, there was no argument for aggressive disease in the present series of patients, as has already been suggested by other groups (Sen et al, 2002; Put et al, 2009).

In addition to the well-documented CLL/MBL/SLL cases, an IGH/BCL2 rearrangement was observed in a heterogeneous group of CLPD (n = 8). This group remained unclassifiable in the absence of characteristic cytology (borderline CLL/MZL, n = 5) and immunology (atypical CLL, n = 2, or closer to an MZL profile, n = 5). No case displayed a specific chromosomal abnormality signature, although the association of trisomy 12 with del(13q) favoured the diagnosis of CLL in three cases. In two cases (Cases 30 and 31), the histology of the spleen and LN sections did not contribute to the diagnosis, which illustrated the existence of borderline CLL/MZL cases. In the U-CLPD cases, however, a Matutes score below three could not formally exclude CLL diagnosis because one case (CLL/MBL/SLL group) in the present series with a score of 1 could be definitively classified as CLL by LN section due to the presence of proliferation centres. Moreover, in our personal series of CLL, seven histologically proven CLL cases had a score below 3. Interestingly, all seven of these cases presented with trisomy 12. On the other hand, these CD5-positive or negative U-CLPD with t(14;18) may have corresponded to MZL. First, 20% of the MZL cases expressed CD5 in the blood (Baseggio et al, 2010). Second, we report here the first case of t(14;18)-positive SMZL (histologically proven). In addition, occasional cases of t(14;18)-positive MALT lymphoma (one gastric and one cutaneous) have been previously reported (Nakamura et al, 2007; Palmedo et al, 2007). Altogether, the differential diagnosis between CLL and MZL in the U-CLPD cases presented here could not be made with certainty in the absence of typical cytology, a typical immunological profile, cytogenetic recurrent abnormalities or characteristic histology.

CLL is thought to develop from an experienced antigen B-cell, either pre-GC or post-GC, probably originating from the marginal zone (MZ) (Chiorazzi & Ferrarini, 2011). However, the presence of an IGH/BCL2-rearrangement in CLL may raise the possibility of derivation from GC B-cells because the bulk of circulating t(14;18) cells in healthy individuals has been recently shown to be FL-like B-cells (Roulland et al, 2006) and because IGH/BCL2 translocations are quite uncommon in MZL. However, some of the present findings argue against the GC-derivation hypothesis. First, although it is rare, MZL may develop from t(14;18)-bearing cells, as reported here for a case with SMZL and in other reports of IGH/BCL2 MALT lymphoma. Second, the IGHV mutation load was lower in most of the t(14;18)-positive CLL cases than in the GC-derived B-cell lymphoma cases (a median mutation load of 92·1% vs. 88·0%) (Gagyi et al, 2008). Third, the analysed FAS loci were unmutated in all of the CLL cases with an IGH/BCL2 translocation (Müschen et al, 2000). Finally, if most t(14;18)-positive CLL cases displayed a mutated BCL6 gene, illustrating that they have probably transited through the GC, this transit could be only transient because the BCL6 mutation load was lower than in GC normal B-cells or GC-derived B-cell lymphoma cells (FL, DLCBL) (Pasqualucci et al, 1998; Capello et al, 2000). Thus, transforming events leading to lymphoma/leukaemia could apparently occur in another site and probably in MZ. Furthermore, the lack of BCL6 mutations in 44% of our CLL/MBL/SLL cases despite the presence of IGHV mutations suggests a T-independent response outside of the GC, as has been proposed for splenic and nodal marginal zone lymphomas (Traverse-Glehen et al, 2007). Although the t(14;18) translocation is thought to occur in BM pre-B cells, in our 2 cases (1 CLL and 1 SLL) with unmutated IGHV gene that demonstrated a GC-independent pathway, the t(14;18) was detected as a secondary event; thus, it probably occurred at a site other than bone marrow, perhaps in the MZ.

In summary, the IGH/BCL2 fusion-positive CLL, as CLL without t(14;18), may originate from MZ-derived memory B-cells which consists of GC-independent B-cells and post-GC B-cells.

Acknowledgements

The authors are grateful to Drs Belhabri, Zurlinden (CH, Macon, France), Dion, Chapelle (CH, Aubenas, France), Salles, Kirschgesner (CH, Châlon sur Saone, France), and Fabre (CH, Bourgoin France) for providing us with some clinical data and samples used in this study.

Conflict of interest

The authors report no potential conflict of interest.

Authorship and disclosures

LB performed the flow cytometry analysis, analysed and interpreted the data, and wrote the manuscript. MOG collected and characterized the cases and performed the IGH/BCL2 FISH analysis. FP reviewed the histological specimens. ECB selected the cases and performed the cytogenetic research. SG performed the cytogenetic research. ATG reviewed the histological specimens and performed the IGHV mutational pattern analysis. MF reviewed the bone marrow biopsies and critically revised the manuscript. SH performed the molecular analysis. AV performed the molecular analysis. DM reviewed the cytological specimens. LJ viewed the cytological specimens. JPP analysed the cytogenetic data. GS analysed the clinical data, performed the statistical analysis and critically revised the manuscript PF designed the research, selected the cases, reviewed the cytological specimens, analysed the data and co-drafted the manuscript.

Ancillary