BCL6 gene rearrangement and protein expression are associated with large cell presentation of extranodal marginal zone B-cell lymphoma of mucosa-associated lymphoid tissue



Extranodal marginal zone B-cell lymphoma of mucosa-associated lymphoid tissue (MALT lymphoma) is an indolent B-cell lymphoma, which is often localized in the stomach. It is characterized by typical morphology, immunology, cytogenetics and expression profile. The coexistence of a large B-cell lymphoma and a MALT lymphoma in the gastrointestinal tract is defined as a composite lymphoma (ComL) and, as we have previously shown, is almost always the consequence of secondary transformation of MALT lymphoma. Here, we have analyzed a panel of seven MALT lymphomas, seven ComL and thirteen large cell variants of marginal zone B-cell lymphomas (MZBL) using FISH for the detection of rearrangements of IGH, MALT1, BCL6, BCL10 and FOXP1 and immunohistochemistry for Bcl6, Bcl10 and FoxP1. Translocations involving IGH were found in 10/27 lymphomas including two cases with IGH-BCL6 fusion and one with IGH-BCL10 fusion; in 7/10 cases, the translocation partner was not identified. Bcl10 and FoxP1 protein expression was heterogeneous throughout the series. Genetic rearrangements of BCL6 and Bcl6 protein expression were found almost exclusively in the large cell components of the ComL and the large cell extranodal MZBL (p = 0.2093 and p = 0.0261, respectively). These findings suggest Bcl6 as a marker for transformation of MALT lymphoma.

A large variety of B-cell lymphomas can primarily arise extranodally involving organs with constitutive or acquired lymphoid tissue. Among them, marginal zone B-cell lymphoma (MZBL) is the only entity occurring more frequently in organs than in lymph nodes, hence it deserves a separate place in the current WHO lymphoma classification,1i.e. extranodal marginal zone lymphoma of mucosa-associated lymphoid tissue (MALT lymphoma) (ICD-O 9699/3). While the large cell variant of MZBL in the lymph node is a canonized, albeit rare, lymphoma variant, the current WHO classification does not apply this term when it comes to name the large cell component of extranodal MZBL in composite lymphoma (ComL) of MALT. Actually, the WHO describes this situation as “diffuse large B-cell lymphoma (DLBCL) in the presence of accompanying MALT lymphoma”. The result of clonal progression of MALT lymphoma is, according to WHO, also to be called DLBCL: “Transformation of MALT lymphoma to DLBCL may occur”.1 This nomenclature reflects the fact that, on histomorphological and immunohistochemical grounds, the large cell component arising in MZBL is still poorly defined. While lymphomas like B-lymphocytic leukemia/lymphoma, mantle cell lymphomas and Burkitt lymphoma can be diagnosed easily in extranodal sites using morphology, immunohistochemistry and cytogenetics, this task is very difficult concerning the heterogeneous group of aggressive B-cell lymphomas. This is regrettable since clinical studies show that the DLBCL of the MALT has a far better cure rate than nodal DLBCL.2, 3 For example, in over half of the cases, regression of gastric DLBCL in an early localized stage occurs following Helicobacter pylori-eradication.4

The small cell and the large cell components of ComL of MALT have repeatedly been shown to be clonally related in terms of VDJ-rearrangement at the IGH locus.5, 6 Molecular-cytogenetic analyses showed common genomic alterations in both and additional aberrations in the large cell compartment yielding evidence for clonal malignant progression.6–10 By expression profiling, we were the first to molecularly define the MALT lymphoma and, additionally, the large cell component of ComL of MALT. The close resemblance of the profiles lead to the concept that the large cell component is the transformation of the small cell component. Furthermore, we were able to identify the majority of DLBCL of MALT as equivalents of the large cell component of ComL, too, since the profiles of these and those of the large cell components of ComL were close to identical. Against this background, we, in the following, use the term “large cell component” to name the DLBCL component of ComL and “large cell extranodal MZBL” for those DLBCL of MALT for which we have obtained the transformed MALT lymphoma signature by gene expression profiling.11

To further characterize the MALT lymphoma and its clonally related large cell variant, we re-examined our panel of 27 cryopreserved extranodal B-cell lymphomas (seven MALT lymphomas, thirteen large cell MZBL and seven ComL) by FISH for the most common chromosomal translocations in MZBL and by immunohistochemistry for protein expression levels of Bcl6, Bcl10 and FoxP1. In four ComL, we were able to analyze the small cell (SC) and the large cell (LC) components separately.

We show here that rearrangements of the BCL6 locus and strong expression of the Bcl6 protein are associated with large cell morphology of MALT lymphoma.

Material and Methods


In essence, the panel used for this study is the formerly published lymphoma series, which has been characterized by molecular cytogenetics and gene expression profiling.6, 7, 11 In brief, frozen samples of 27 primary extranodal B-cell lymphomas were taken from the tissue collection of the Institute of Pathology, Ulm University. Formalin-fixed paraffin-embedded (FFPE) tissue for immunohistochemistry was available for 21 lymphomas. The lymphomas were pseudonymized to comply with the German law for correct usage of archival tissue for clinical research.12 Approval for this procedure was obtained from the local ethics committee. This study included seven MALT lymphomas, five of these originating from the stomach, one from the lung and one from the large intestine, thirteen large cell MZBL, eleven of these originating from the stomach, one from the lung and one from the illeocoecum and seven ComL, one from the small intestine and six originating from the stomach. Lymphomas were classified on morphological and immunohistological grounds according to the WHO1 while mantle cell lymphoma, follicular and transformed follicular lymphoma and Burkitt lymphoma/lymphoma intermediate between Burkitt and DLBCL were excluded using additional appropriate FISH assays (see FISH section below). In all ComL, the small and the large cell components were microdissected and clonality was proven by identical VDJ rearrangements at the IGH locus verified by PCR according to standardized protocols.6 Separate FISH analysis of the small cell and the large cell compartments was possible in four ComL. Formalin-fixation and paraffin-embedding for immunohistochemistry were done only on tissue from lymphomas with sufficient amounts of cryopreserved material.

Interphase fluorescence in situ hybridization (FISH)

MALT lymphoma associated chromosomal translocations were investigated by interphase FISH on 3μm cryosections according to standard protocols.7 All lymphomas were negative by FISH for a chromosomal breakpoint affecting the BCL2 locus or the IGH-BCL2 fusion indicating the t(14;18)(q32;q21) using commercially available probes (LSI BCL2 Break Apart Rearrangement Probe and LSI IGH/BCL2 Dual Color, Dual Fusion Translocation Probe, Vysis/Abbott; Des Plaines, IL, USA) hence excluding follicular lymphomas. All large cell MZBL and ComL LC were negative by FISH for common chromosomal breakpoints affecting the MYC or CCND1 locus. (LSI MYC Break Apart Rearrangement Probe and LSI CCND1 Break Apart Rearrangement Probe, Vysis/Abbott).

All lymphomas were screened with dual-color break-apart probe sets for IGH, MALT1 and BCL6 (LSI IGH Break Apart Rearrangement Probe, LSI MALT1 Break Apart Rearrangement Probe and LSI BCL6 Break Apart Rearrangement Probe, Vysis/Abbott), BCL1013 (BCL10 FISH DNA Probe, Split Signal, Dako, Glostrup, Denmark) and FOXP1 (using BACs RP11-430J3 and RP11-321A23).14 Lymphomas with a signal pattern indicating a chromosomal breakpoint in the MALT1 locus were further examined with double color dual fusion translocation probes for the t(14;18)(q32;q21)/IGH-MALT1 fusion and the t(11;18)(q21;q21)/API2-MALT1 fusion (LSI IGH/MALT1 Dual Color, Dual Fusion Translocation Probe and LSI API2/MALT1 Dual Color, Dual Fusion Translocation Probe, Vysis/Abbott). Lymphomas with signal patterns indicating a chromosomal translocation affecting IGH and BCL6 or IGH and BCL10, respectively, were further investigated using in-house developed translocation probes for the detection of t(3;14)(q27;q32)/BCL6-IGH fusion15 and the t(1;14)(p22;q32)/IGH-BCL10 fusion.16. Break-apart probes for the loci of the immunoglobulin light chains kappa (IGK) and lambda (IGL)13 (IGK FISH DNA Probe, Split Signal and IGL FISH DNA Probe, Split Signal, Dako, Glostrup, Denmark) were applied on lymphomas with signal patterns indicating a chromosomal breakpoint affecting the BCL6-locus but showing an intact IGH locus. Cut-off values for each probe set were defined as described recently.17 In brief, the probes were hybridized to at least three reactive tonsils or lymph nodes, respectively. Mean values were determined and the cut-off values were set to mean value plus three times the standard deviation. Each lymphoma sample was evaluated by two independent observers, and at least 100 nuclei were counted. The cut-off values were 7% for the IGH/BCL2 Fusion probe, 8% for the BCL2 Break-apart probe, 10% for the MYC Break-apart probe, 6% for the CCND1 Break-apart probe, 13% for the IGH Break-apart probe, 5% for the MALT1 Break-apart probe, 8% for the BCL6 Break-apart probe, 6% for the BCL10 Split Signal probe, 11% for the FOXP1 Break-apart probe, 11% for the IGH/MALT1 Fusion probe, 11% for the API2/MALT1 Fusion probe, 5% for the BCL6/IGH Fusion probe, 5% for the IGH/BCL10 Fusion probe, 8% for the IGK Split Signal probe and 9% for the IGL Split Signal probe.


Immunohistochemical staining for Bcl6 (clone PGB6p, DAKO), Bcl10 (clone 151, a kind gift from Prof. Ming-Qing Du, Cambridge)18 and FoxP1 (Cell Signaling, Danvers, MA) was performed on Formalin-fixed paraffin-embedded tissue sections as described previously.19 In the lymphomas with no paraffin material available, staining for Bcl6 was performed on cryosections. Evaluation of immunostaining was carried out in a blinded fashion. In tissue sections, strongly stained immunoblasts or lymph follicles were regarded as positive intrinsic controls. Several categories were created according to the proportion of lymphoma cells displaying positive staining; those were “no staining detected”, “staining in up to 30%”, “staining in more than 30% and up to 70%” and “staining in more than 70%” of the total number of lymphoma cells analyzed (see also Table 3 and supplementary table Supporting Information).

Statistical analysis

The differences in frequencies of Bcl6 protein expression and genomic rearrangements between the small cell lymphomas/compartments and the large cell lymphomas/compartments were tested for statistical significance using Fisher's exact test. The p-value for significance was below 0.05.


Incidence of MALT-lymphoma-associated chromosomal aberrations

FISH results are summarized in Table 1. 13/27 lymphomas (48%) had at least one of the translocations or breakpoints investigated. One MALT lymphoma of the stomach (No.7) had a significant number of interphase cells with a signal constellation indicating the t(11;18)(q21;q21)/API2-MALT1 and another one originating from the lung (No.3) showed chromosomal breakpoints affecting MALT1 and IGH other than the t(14;18) (q32;q21). One ComL LC (No. 8, see Fig. 1) and one large cell MZBL (No.27) showed a signal constellation suggesting the t(3;14)(q27;q32)/BCL6-IGH, whereas two lymphomas (one MALT lymphoma originating from the ileocoecal region, No.5, and one large cell MZBL originating from the stomach, No.17) showed breakpoints affecting the IGH and BCL6 locus, but not the t(3;14)(q27;q32) (see also Table 3). No.5 had a split signal constellation of the IGK light chain in a subset of lymphoma cells, suggesting a t(2;3)(p12;q27)/IGK-BCL6. One ComL (No.9) had in the majority of lymphoma cells of the large cell compartment signal constellations suggesting the t(1;14)(p22;q32)/BCL10-IGH (Figs. 2a2c) and an additional breakpoint of BCL6. Three lymphomas (two large cell MZBL, Nos. 21 and 26) had a breakpoint of the BCL6 locus as the sole aberration; seven lymphomas in total (26%) showed a rearrangement of IGH with an unidentified translocation partner.

Table 1. FISH results of the different GI lymphoma specimen
inline image
Figure 1.

Lymphoma No.8; ComL with t(3;14)(q27;q32)/BCL6-IGH in the large cell component. (a) Hematoxylin-Eosin staining of lymphoma No.8; clearly distinguishable the small cell component in the upper left side of the panel and the large cell component in the lower right side. (b) Small cell compartment: FISH with a BCL6 break-apart probe; the lymphoma cells show fusion signals indicating an intact BCL6 gene (representative cells are highlighted by arrows). (c) Large cell compartment: FISH with a BCL6 break-apart probe; arrows pointing to separated red and green signals indicating a breakpoint of BCL6. (d) Large cell compartment: FISH with a BCL6/IGH-Translocation probe-set; arrows marking fusion signals indicating a t(3;14)(q27;q32).

Figure 2.

Lymphoma No.9; ComL with t(1;14)(p22;q32)/BCL10-IGH and additional breakpoint of BCL6 in the large cell component. (a) Large cell compartment: FISH with a BCL10 break-apart probe; arrows pointing to separated red and green signals indicating a breakpoint in the BCL10 gene. (b) Large cell compartment: FISH with a BCL10/IGH-translocation probe-set; arrows marking fusion signals indicating a t(1;14)(p22;q32). (c) Large cell compartment: immunostaining with an antibody against Bcl10, arrows pointing to single cells expressing high amounts of the protein.

Genomic status and protein level of Bcl6, Bcl10 and FoxP1

Results of the immunostaining for Bcl6, Bcl10 and FoxP1 are summarized in Table 2. Two MALT lymphomas (40%), eight large cell MZBL (89%) and two ComL LC (29%) among the analyzed lymphomas showed expression of FoxP1 protein in at least 30% of the lymphoma cells despite the fact that no genomic aberration was detected in any case by FISH.

Table 2. IHC results of the different GI lymphoma specimen
inline image

Thirty-three percent of all lymphomas showed at least weak expression of Bcl10 protein in more than 30% of the lymphoma cells. The ComL harboring the t(1;14)(p22;q32)/BCL10-IGH (No.9) in its large cell compartment showed a strong cytoplasmic and nuclear expression of Bcl10-Protein in a subset of these lymphoma cells (Fig. 2d, see also Table 3). Concerning expression of Bcl10 and FoxP1, there was no significant difference between small cell and large cell lymphomas.

Table 3. FISH and immunohistochemistry data of lymphomas with genomic aberrations of BCL6 in detail
inline image

Whereas Bcl6 expression was detectable in most ComL LC (86%) and in about half of the large cell MZBL (46%), none of the MALT lymphomas or ComL SC showed positive immunostaining. Among the lymphomas with intact BCL6 gene, 46% had moderate up to strong Bcl6 protein expression (including five ComL LC and six large cell MZBL). The FISH and IHC results for all lymphomas with genomic aberrations of BCL6 are listed in detail in Table 3 (for detailed information of all lymphomas see Supporting Information table). The two lymphomas harboring the translocation t(3;14) (q27;32) (Nos. 8 and 27) showed positive Bcl6 immunostaining at least in a subset of lymphoma cells. Of five other lymphomas with a chromosomal breakpoint affecting the BCL6 locus, one MALT lymphoma (No.5) showed no immunohistochemical staining, and three large cell MZBL (Nos. 16, 21 and 26) and one ComL LC (No.9) showed moderate up to strong Bcl6-expression. FISH analysis revealed extra copies of Bcl6 in three lymphomas (Nos. 5, 12 and 21) and a monoallelic loss in one lymphoma (No.18) independently whether the gene was intact or not. Protein expression did not correlate with the genomic status of BCL6 in these cases (see Table 3).

Comparing the status of all small cell lymphomas (MALT lymphomas and ComL SC) with that of all large cell lymphomas (large cell MZBL and ComL LC), the frequency of breakpoints in the BCL6 gene is higher in large cell lymphomas (6/20) than in small cell lymphomas (1/11). Concerning Bcl6 protein expression, we observed the same: 12/20 large cell tumors (tumor components) showed moderate up to high expression compared to 0/10 small cell tumors (tumor components). While differences in protein expression proved to be statistically significant (p = 0.0261), this was not the case for the occurrence of BCL6 breakpoints (p = 0.2093).


We have analyzed a series of 27 MALT lymphomas, ComL and large cell MZBL using FISH and immunohistochemical staining for BCL6, BCL10 and FOXP1 supposed to be associated with lymphomagenesis and progression.

Bcl10 is essential for B-cell development and specifically links the antigen receptor signaling to the inflammatory NF-κB pathway.20, 21 In MALT lymphoma, the BCL10 gene has been sporadically identified in the translocation t(1;14)(p22;q32) and in the variant translocation t(1;2)(p22;p12) involving either the immunoglobulin heavy chain or the kappa light chain promoter, hence deregulating BCL10 gene expression.20, 22 In our series, we have found that BCL10 was once involved in a t(1;14)(p22;q32). In this ComL we found an expression of the Bcl10 protein in a subset of lymphoma cells of the large cell component with cytoplasmic and nuclear staining. The intensity of staining in these cells was the strongest observed in all lymphomas studied suggesting an IGH promoter based effect on Bcl10 protein expression. We cannot completely exclude the possibility of the variant translocation t(1;2)(p22;p12), since there is no accordant validation of the used break-apart probe so far. The lymphomas with signal constellations indicating an intact BCL10 gene showed a heterogeneous expression of Bcl10 protein which was predominantly nuclear.

In some MALT lymphomas and rarely in nodal and extranodal DLBCL, FOXP1 is involved in the t(3;14)(p13;q32),23, 24 leading to over-expression of the protein.25 Since this translocation has been described predominantly in MALT lymphomas outside the gastrointestinal tract and the lung, our finding that none of the lymphomas in our series shows this aberration is in line with this data. Protein expression of FoxP1 was found in about 50% of DLBCL26 as well as in 29% of MALT lymphomas27 and was in both types of B-cell lymphoma associated with a poor prognosis. Goatly et al. found protein expression of FoxP1 in gastric MALT lymphomas and DLBCL also independently of an increased copy number of the gene or its involvement in the t(3;14)(p13;q32).25 In our panel, the FoxP1 expression was the highest in large cell MZBL (89%) despite an intact FOXP1 gene, hence supporting published findings and suggesting regulation mechanisms other than translocations involving the immunoglobulin genes.

Rearrangements involving the IGH gene were believed to be a rather rare event in MALT lymphomas arising from different sites. In studies with a large number of cases (n = 133 and n = 90, respectively), only 6 to 9% harbor this aberration.28, 29 On the other hand, Vinatzer et al. found translocations involving IGH in 59% of their MALT lymphoma panel (n = 29), containing mainly extragastric lymphomas.30 In our study, 43% of the MALT lymphomas and 75% of the ComL SC have an IGH-split. Looking at the data in detail, one can see that IGH-involved translocations occur mainly in extragastric lymphomas. Correspondingly, two of our three positive lymphomas were located in the lung and the large intestine, respectively. This fits the concept that the incidence of chromosomal translocations in MALT lymphoma may depend on the site of primary involvement.31

In their collection of gastric ComL and large B-cell lymphomas Nakamura et al. found a breakpoint within the IGH gene in 14% (14/27) and 22% (22/36), respectively.32 Apart from the fact that translocations involving c-MYC were not excluded in this study, our numbers fit this data since we found translocations involving IGH in 38% of the ComL LC and in 8% of the large cell MZBL. Although rare translocations involving c-MYC cannot be detected using the break-apart probe,33 the total number of IGH rearrangements other than the characteristic t(14;18)(q32;q21)/IGH-MALT1, the t(1;14)(p22;q32)/BCL10-IGH and the t(3;14)(p13;q32)/FOXP1-IGH is surprisingly high. This fact points to new IGH translocation partners in gastrointestinal B-cell lymphoma. What is more, in four ComL with separately analyzed small cell and large cell components, the IGH rearrangement was positive in three lymphomas in both compartments and is thus an early oncogenic event.

Bcl6 is a transcriptional repressor of P53 and modulates DNA damage-induced apoptotic responses in germinal center B-cells.34 It is currently regarded as a typical marker of germinal center-derived lymphomas. Therefore, the fact that we find the gene rearranged and the protein expressed in lymphomas with marginal zone differentiation sheds new light on this molecule. These observations add to the facts found in activated B-cell like DLBCL35 that Bcl6 expression is not restricted to germinal center derived lymphomas. Translocations involving BCL6 are found in only 2% of MALT lymphomas,36 in 10% of follicular lymphomas37 and in 40% of nodal DLBCL where they are thought to be associated with a poor prognosis.38, 39 In a study focusing exclusively on gastrointestinal B-cell lymphomas, 36% of the included extranodal DLBCL (n = 33) and 50% of the ComL (n = 6) showed rearrangements of BCL6.40 In our panel, we have one MALT lymphoma (14%), four large cell MZBL (31%) and LC of two ComL (29%) with a breakpoint within the BCL6 gene, this being in line with published data. The fact that rearrangements of BCL6 including the t(3;14)(q27;q32)/BCL6-IGH are almost exclusively present in large cell MZBL and in large cell compartments of ComL points to an association with transformation of MALT lymphoma.

Comparing FISH results showing deregulation of the BCL6 gene and immunohistochemical staining for Bcl6 protein in these lymphomas, we failed to find any correlation. Ye et al. made the same observation in their series of MALT lymphomas.36 Likewise, in some DLBCL and follicular lymphomas grade III with BCL6 rearrangement, no protein expression was detectable.41, 42 This might be explained by the pleiotropic functions of BCL6 during B-cell development and differentiation and thus various existing regulatory mechanisms for this gene. Nevertheless, in our series, Bcl6 protein expression was only present in the large cell MZBL and in the large cell components of ComL (Fig. 3), underlining its association with lymphoma progression.

Figure 3.

Lymphoma No.11; ComL with Bcl6 expression exclusively in the large cell component. (a) Hematoxylin-Eosin staining of the small cell component of lymphoma No.11 (ComL). Cells have a lymphocytic appearance (b) Hematoxylin-Eosin staining of the large cell component of the same lymphoma. (c) Immunostaining with an antibody against Ki-67. The large cell compartments show a considerably higher proliferation index than the small cell areas. (d) Immunostaining with an antibody against Bcl6. The small cell compartment is negative, while the large cell compartment shows strong expression of Bcl6 protein.

Taken together, rearrangements of BCL6 including the t(3;14)(q27;q32)/BCL6-IGH are far more frequent in large cell MZBL and in ComL LC than in small cell MALT lymphoma and SC of ComL. The difference is not statistically significant, most probably due to the small number of MALT lymphomas containing one lymphoma with t(3;14)(q27;q32)/BCL6-IGH. Expression of Bcl6 protein is exclusively present in the large cell MZBL and ComL LC and, thus, is significantly different between the groups of small cell and large cell lymphomas. Therefore, Bcl6 is an immunohistological marker for transformation of MZBL.

In conclusion, we have analyzed a series of 27 MALT lymphomas, ComL and large cell MZBL using FISH and immunohistochemical staining. In particular, we analyzed the small cell and the large cell components of four ComL separately. The percentage of translocations involving IGH and an unknown translocation partner was surprisingly high throughout the different lymphoma entities represented in our panel. Rearrangements of BCL6 and enhanced Bcl6 protein expression were found almost exclusively in the large cell MZBL and in the large cell components of the ComL, suggesting an association of this gene with transformation from small cell MALT lymphoma to its large cell variant.


The authors thank Prof. M.-Q. Du for kindly providing the Bcl10 antibody. They thank Claudia Becher for excellent technical assistance. FISH analyses for detection of IGH translocation partners were performed in the framework of the Verbundprojekt “Molekulare Mechanismen der malignen Lymphome” (MMML), funded by the Deutsche Krebshilfe.