Frequent Expression of the B-Cell–Specific Activator Protein in Reed-Sternberg Cells of Classical Hodgkin’s Disease Provides Further Evidence for Its B-Cell Origin

CLASSICAL HODGKIN’S DISEASE (cHD) is one of the most frequent malignant lymphomas.1 Despite intensive research, its cellular origin was a very controversial issue up to the middle of this decade. Phenotypic analysis of HD-derived cell lines or tumor cells in HD tissues was variously interpreted to suggest a derivation from lymphoid cells,2,3 T cells,4,5 B cells,6,7 interdigitating reticulum cells,8,9 follicular dendritic cells,10 or macrophages11 comprising most of the normal constituents of the lymph node. Although not completely specific, the consistent expression of the lymphoid activation antigen CD30,2,3 the presence of the Epstein-Barr virus (EBV) in a substantial proportion of cHD cases,12,13 and the cytokine profile with expression of lymphotoxin α,14,15 however, indicated a derivation from lymphoid cells. More recently, two groups showed the presence of clonally rearranged Ig genes in single tumor cells isolated from cHD tissues, which is strong evidence for a B-cell derivation of this disease.16,17 For a B-cell derivation, the immunophenotype of the Hodgkin and Reed-Sternberg cells (HRS cells), however, is highly unusual: unlike most peripheral B-cell neoplasms, expression of B-cell antigens is only found in a minority of cHD cases and usually only in a small number of HRS cells.7,18 An origin of HRS cells from T cells, in some cases, has been repeatedly suggested, and this hypothesis has more recently gained support from the demonstration that HRS cells of 10% to 20% of cHD cases express cytotoxic molecules that are considered to be specific for cytotoxic lymphocytes.19 20 However, the final proof for a T-cell derivation of some cHD cases, ie, the demonstration of rearranged T-cell receptor genes in HRS cells, is still to be provided.

B-cell–specific activator protein (BSAP) has recently been found to be specific for B-cell lineage,21 and its expression in HRS cells has been reported in a proportion of cHD cases in a small series.22 This molecule is a transcription factor encoded by the Pax-5 gene and is expressed in all B cells except terminally differentiated plasma cells.21 BSAP has been shown to be essential for the development of murine B cells, because mice with disrupted Pax-5 genes do not possess any peripheral B cells.23 It exerts many functions in B cells, including enhancement of B-cell antigen expression and of proliferation.24,25 Expression of BSAP in late B cells is sufficient to inhibit plasmacellular differentiation, indicating that alteration of BSAP levels is associated with plasmacellular differentiation.26 Some cases of lymphoplasmacytic lymphoma (immunocytoma) and marginal zone B-cell lymphoma harbor translocations, which result in upregulated BSAP expression.27,28 These genetic lesions are considered to contribute to lymphomagenesis in these cases.27 28 

In this study we investigated the expression of BSAP transcripts and protein in cHD cases by isotopic in situ hybridization and immunohistochemistry with a monoclonal antibody. We were interested in the expression of this factor in HRS cells for several reasons: (1) The expression of BSAP would provide further evidence for a B-cell derivation of this disease; (2) it could help distinguish this disease from anaplastic large cell lymphoma (ALCL), which can pose differential diagnostic problems with cHD; and (3) given the numerous functions of BSAP in B-cell physiology, deregulated BSAP expression in HRS cells could explain some of the unusual phenotypic features of these neoplastic B cells. B-cell lymphoma and T-cell lymphoma cases, as well as reactive lymphoid tissues, were also examined to confirm the specificity of BSAP expression for the B-cell lineage.

Formol-fixed/paraffin-embedded biopsy specimens of the following diseases were taken from the archives of the Institute of Pathology, Klinikum Benjamin Franklin: slightly hyperplastic tonsils (2 cases), ulcerating tonsillitis (1 case, human immunodeficiency virus [HIV]⁺ patient), hyperplastic adenoids (1 case, HIV⁺ patient), Piringer’s lymphadenitis (3 cases), different types of B-cell lymphomas (Table1) and peripheral T-cell lymphomas (Table 1), 10 cases of nodular lymphocyte predominance HD, and 33 cases of cHD (Table 2). The plasmacytoma cases were all from extramedullary sites. All the diagnosis were established according to the Revised European-American Classification of Lymphoid Neoplasms.29 Because the existence and classification of anaplastic large cell lymphoma Hodgkin’s-like is controversial, such cases were excluded from the study.

Four-micrometer sections of paraffin-embedded tissue blocks were stained applying the immunoalkaline phosphatase (APAAP) method.30 The primary antibodies used in this study were L26 (CD20), BER-H2 (CD30), and Cs1-4, a cocktail of 4 antibodies specific for the EBV-encoded latent membrane protein 1 (DAKO, Glostrup, Denmark). For the detection of CD3, a polyclonal antibody was used. All antibodies were used after high-pressure cooking to optimally visualize the antigens in paraffin sections. The monoclonal anti-BSAP antibody was purchased from Transduction Laboratories (Lexington, KY) and used after high-pressure cooking in citrate buffer (10 mmol, pH 6.0, 2 minutes) in a dilution of 1:10. For the detection of the bound anti-BSAP antibody, the APAAP method was applied. This resulted in nuclear positivity, which in a few cells was accompanied by very weak cytoplasmic staining. Simultaneous detection of 2 antigens was performed using the peroxidase technique for membraneous/cytoplasmic antigens (CD3, CD20, CD30) and the APAAP method for BSAP.

The Pax-5 probe was prepared from cDNA that was generated by reverse-transcriptase polymerase chain reaction (RT-PCR) from a B-cell line. The nucleic acid sequence of this probe was determined on the DNA sequencer 377 (Applied Biosystems, Foster City, CA) and proved to conform to published data.21 The EBER-1 and -2 probes were kindly provided by Dr G. Niedobitek (Erlangen, Germany). After linearization of the plasmid construct with appropriate restriction enzymes, anti-sense and sense (control) RNA probes were generated by run-off transcription as described,31 but with incorporation of 2 [³⁵S]-labeled nucleotides.

ISH was performed as previously described,31 but with microwave irradiation (5 minutes in citrate buffer) before the prehybridization steps. The slides were exposed for 3 and 6 weeks. Simultaneous double labeling for EBV-encoded small nuclear transcripts (EBER) and BSAP was performed as reported,32 also with the inclusion of microwave irradiation. Sections hybridized with sense probes showed only weak nonspecific background (not shown).

In cHD cases the whole slide was screened for BSAP-positive tumor cells, and only unequivocally labeled neoplastic cells containing huge nuclei with inclusion-like nucleoli were accepted as positive. Cases showing labeling of only 1 or 2 presumably neoplastic cells were considered negative. The slides of both the ISH experiments and of immunohistology were evaluated by 2 independent investigators (H.-D.F. and R.R. or H.-D.F. and I.A.). Cases with discrepant evaluations were discussed at the microscope.

In the lymphoid tissue of tonsils, strong BSAP-specific signals were observed in lymphoid cells within germinal centers and in the mantle zone, as well as in intraepithelial lymphocytes (Fig1A). In contrast, only rare lymphoid cells were labeled in the interfollicular zone. Signals were not observed in mesenchymal cells; however, focal weak labeling was found in the basal and parabasal parts of the squamous epithelium.

In tonsils, analysis of BSAP protein expression yielded almost identical results as ISH. However, the squamous epithelium did not display any labeling. The simultaneous demonstration of CD3 or CD20 and BSAP showed expression of BSAP in CD20⁺ cells and absence of BSAP in CD3⁺ T cells (Fig 1B). In addition to the findings described in tonsils, Piringer’s lymphadenitis cases demonstrated labeling of monocytoid B cells and some interfollicular blasts (Fig 1C through E).

All the B-cell lymphomas investigated, except plasmacytoma cases, consistently expressed BSAP transcripts (Table 1). In most cases, the majority of the neoplastic cells proved to contain BSAP transcripts; however, in immunocytoma cases, the number of labeled neoplastic cells was more variable than in the other B-cell neoplasms. Two of five plamocytoma cases focally expressed BSAP transcripts (<25% of tumor cells labeled), whereas the other cases were negative. In 10 of 11 cases of peripheral T-cell lymphoma, the tumor cells did not contain BSAP transcripts (Table 1); in 1 case of anaplastic large cell lymphoma (null cell type), there was a slight focal increase in silver grains over the neoplastic cells which, however, was not sufficient to meet the criteria for definitive positivity.

Again IH provided similar results as ISH with almost all B-cell neoplasms expressing BSAP protein (Table 1A and B, Fig 1F). A notable exception were plasmacytomas, which were consistently negative, including the 2 cases that were focally positive by ISH. Similarly, the neoplastic cells of all peripheral T-cell lymphomas did not contain BSAP protein, including the case of ALCL with equivocal labeling in ISH. Combined immunohistology for BSAP and CD20 or CD30 was performed in 2 ALCL cases containing some BSAP-positive medium-sized lymphoid cells. These experiments confirmed that BSAP expression is absent from the CD30⁺ neoplastic cells and is restricted to CD20⁺ reactive B cells (Fig 1G and H).

BSAP transcripts were found in the tumor cells of 22 of 25 cases (Table2, Fig 1I). This was also true for EBV-infected HRS cells as evidenced by simultaneous detection of EBER and BSAP. The number of labeled cells varied considerably from case to case. In 7 cases, more than 75% of the tumor cells expressed BSAP whereas in 4 cases less than 10% of the neoplastic cells were labeled. In all cases, scattered small- or medium-sized reactive cells and nodular accumulations of mantle zone cells expressed BSAP transcripts. Surprisingly, BSAP protein expression was detected in some of the cases in more neoplastic cells compared with BSAP transcript expression (see below and Table 2). This difference may be due to the fact that weak BSAP transcript expression may be difficult to distinguish from background signal and that rather short exposure times were used in the ISH experiments.

BSAP protein was detectable in HRS cells of 28 of 31 cHD cases (Fig 1J and K). In contrast to most of the B-cell lymphomas, the staining intensity was often very variable ranging from week to moderate to rarely strong (compared with small reactive lymphocytes). There was no obvious association between BSAP protein positivity and histological subtype (nodular sclerosis versus mixed cellularity), the expression of CD20, or the presence of EBV in the neoplastic cells. Two cases expressed cytotoxic molecules in the neoplastic cells; one of these (case 18) also displayed BSAP in HRS cells, whereas the other (case 33) did not.

The neoplastic cells of all 10 cases of nodular LPHD expressed BSAP protein (not shown). This occurred in more than 75% of the tumor cells in 8 cases and in 50% to 75% of the cells in an additional case. In 1 of the cases, the percentage of labeled tumor cells could not be determined because of staining inhomogeneity. The staining intensity, while being variable, appeared to be stronger than in cHD cases, with 6 of 10 LPHD cases containing at least some neoplastic cells displaying the same staining intensity as that seen in small lymphocytes.

In this study we investigated the expression of a B-cell–specific transcription factor, BSAP, in cHD and controls. In agreement with an earlier study using 2 polyclonal antibodies,22 analysis of the control tissues confirmed that BSAP protein detection with a monoclonal antibody is sufficiently specific for B cells: expression of this molecule was found in B-cell regions of nonneoplastic lymphoid tissues but was mostly absent from interfollicular areas and not found in CD3⁺ lymphocytes. Similarly, among non-Hodgkin lymphomas BSAP expression was observed in all B-cell lymphomas (with the exception of plasmacytoma cases confirming earlier findings33) but not detected in the tumor cells of any peripheral T-cell neoplasm. ISH yielded very similar results as did IH, with minor exceptions: Weak/questionable signals were observed in 1 case of anaplastic large cell lymphoma of null-cell type, in the tonsillar squamous epithelium, and in a minority of cells of 2 of 5 plasmacytoma cases. The lack of protein expression in these instances may indicate that BSAP transcripts are not always translated into a protein product.

In cHD, BSAP was found in a variable number of HRS cells in more than 80% of the cases both by IH and ISH. These findings strongly support a frequent B-cell derivation of cHD, which is in line with the molecular characteristics of this disease, ie, the presence of clonally rearranged Ig receptor genes in most cases.16,17 Similarly, detection of BSAP in all LPHD cases further confirms the well-established B-cell nature of this neoplasm. Our findings for cHD are only in partial agreement with another study22describing BSAP expression in 5 of 14 cHD cases. The polyclonal antibodies used in the quoted study produced in our hands a markedly weaker staining than the monoclonal antibody applied in this study, suggesting that this discrepancy may be caused by different sensitivities of the immunohistochemical procedures. Our results differ from those obtained by Hsu et al,34 who did not find functional BSAP in HD-derived cell lines. Our preliminary investigations indicate that BSAP protein expression is less frequently found in these lines than in primary tissues.

In addition to the confirmation of the B-cell nature of cHD, our results may have important diagnostic implications. In particular, BSAP protein expression may help to delineate ALCL of T-/null cell type from cHD. Both of these tumors contain large CD30-positive tumor cells that are usually numerous in ALCL and scarce in cHD.4 However, tumor-cell–rich cases of cHD may be difficult to differentiate from ALCL and vice versa, and these cases may actually be differently diagnosed by different pathologists.35 In addition to the detection of the ALK-protein or EBV-encoded molecules, BSAP protein expression may be of value in this setting because it has not been observed in any ALCL case investigated so far, but is frequently detectable in HRS cells. In this regard, the detection of BSAP expression is particularly satisfying because it bases the differential diagnosis on lineage derivation.

BSAP is known to influence many different B-cell functions, including the enhancement of the expression of B-cell antigens such as CD19 and, possibly, CD20.24,25 Therefore, it is conceivable that deregulated BSAP expression may be involved in the generation of the abnormal phenotype of HRS cells. Our results, however, show that BSAP expression in these cells is much more frequent than the detection of B-cell antigens.7 18 For these reasons, the complete absence of BSAP expression cannot account for the frequent absence of B-cell antigens in this disease. In contrast to most B-cell lymphomas, however, BSAP expression in cHD was only of weak to moderate intensity in the majority of cHD cases. Whether these low levels of BSAP expression in HRS cells do not suffice to induce B-cell antigens in HRS cells remains a matter of speculation.

In summary, our results and those of Krenacs et al22 show that BSAP is a reliable B-cell marker that may be useful for diagnostic immunohistology. BSAP is frequently expressed in the neoplastic cells of cHD, and this expression both confirms the B-cell nature of this neoplasm and helps to separate cHD from ALCL of T/null-cell type. Whether weak expression of BSAP contributes to the frequent absence of B-cell antigens in cHD remains to be determined.

This work contains part of the doctoral theses of R.R. and G.L. We are indebted to E. Berg and L. Öhring for excellent technical assistance.