Gcet1 (centerin), a highly restricted marker for a subset of germinal center-derived lymphomas

Germinal center (GC) B cell–expressed transcript 1 (Gcet1, centerin, serpin A9) is a recently described molecule that belongs to the family of serine-protease inhibitors or serpins.^1,–3  The gene encoding Gcet1 is located on chromosome 14q32 and has been identified independently by 2 groups each using a different approach. First, mRNA was subtracted from highly purified GC B cells and demonstrated to be highly restricted to GC B cells and a selection of Burkitt lymphoma (BL) cell lines. GCET1 was also demonstrated to be induced when naive B cells were stimulated in vitro via CD40 signaling, whereas Staphylococcus aureus Cowan strain activation failed to do so.¹ 

The second approach was based on the results from cDNA microarray analysis. A group of genes was identified as being highly expressed in normal GC B cells and GC B cell–derived malignancies, but not in resting or activated peripheral blood B cells.^4,5  Specifically, bcl6 and 2 ESTs (IMAGE clones 1334260 and 814622, which correspond to GCET1 and GCET2-HGAL, respectively) were included in the GC set of genes, which was found to be associated with a more favorable prognosis in diffuse large B-cell lymphoma (DLBCL), and had the highest discriminatory capacity for differentiating GC B-DLBCL from other subtypes.⁵  This was confirmed by quantitative real-time polymerase chain reaction, which also revealed that Gcet1 expression was restricted to GC B cells and GC B cell–derived lymphomas.² 

Further support for the possible diagnostic relevance of GCET1 expression was provided by the comparison of follicular lymphoma (FL) gene expression data with those of follicular hyperplasia (“Follicular lymphoma signature”), which revealed a nearly 2-fold overexpression of GCET1. These findings prompted us to develop a monoclonal antibody that acts on paraffin-embedded tissue, to investigate further the role of Gcet1 and its potential diagnostic value as a GC marker.

This study therefore aimed to characterize Gcet1 expression at the protein level in normal and reactive human tissues, cell lines, and a large series of human B- and T-cell lymphoid neoplasms. To this end, we performed immunohistochemical and Western blot analyses using a newly generated monoclonal antibody that is reactive in paraffin-embedded tissues.

Spanish National Cancer Centre (CNIO) Institutional Review Board approval was obtained for these studies. Informed consent was obtained from all patients in accordance with the Declaration of Helsinki. A series of 33 patients diagnosed with FL in lymph nodes was investigated. Total RNA was extracted from frozen tumor samples using the Trizol reagent (Invitrogen) and purified and treated with RNase-free DNase I using the RNeasy kit (Qiagen, Valencia, CA); 1 to 5 μg of target RNA was amplified using T-7 in vitro transcription⁶  and 2.5 μg of amplified RNA (aRNA) was directly labeled with cyanine 5-conjugated dUTP (Cy5) or cyanine 3-conjugated dUTP (Cy3) (Amersham, Uppsala, Sweden). The reference sample used was 2.5 μg of aRNA from the Universal Human Reference RNA (Stratagene, La Jolla, CA).

Microarray studies were carried out using the CNIO OncoChip. Labeling and hybridization were performed as previously described.^6,–8  The clone sequences of all the genes included in the OncoChip and the reproducibility of the expression data (measured by quantitative polymerase chain reaction) of multiple genes have been verified.^6,7,9,–11  Scanning and image analysis were performed using a Scanarray 5000 XL (GSI Lumonics, Kanata, ON) and GenePix Pro Software (Axon Instruments, Union City, CA), respectively.

Raw microarray data were processed as previously described.^7,8,10  Data from 33 FL cases were normalized against the average expression for each gene from 5 reactive lymph node samples. Only the data from genes for which at least 50% of the data were available in the control samples were normalized; all other genes were excluded from the analysis.

A cDNA encoding the full-length human Gcet1 protein was derived from the laboratory of Dr W. C. Chan. The human GCET1 gene was amplified by polymerase chain reaction and introduced into the pCOLDII expression vector (Takara, Otsu, Japan). The 6xHis-Gcet1 fusion protein was then expressed in Escherichia coli strain Rosetta (Novagen, Nottingham, United Kingdom) at 15°C for 48 hours. The recombinant protein was solubilized from its insoluble inclusion bodies in 20 mM Tris-HCl containing 0.5 M NaCl, 5 mM imidazole, 1 mM 2-mercaptoethanol, and 6 M guanidine–HCl (pH 8.0), and purified by affinity chromatography on a HiTrap Chelating column (GE Healthcare, Little Chalfont, United Kingdom) under denaturing conditions. Renaturation of the protein was induced on-column by decreasing the urea concentration in a linear gradient from 8 to 0 M more than 3 hours. After refolding, the renatured protein was eluted by increasing gradually the concentration of imidazole up to 0.5 M in 20 mM Tris (pH 8.0), 0.5 M NaCl, and 1 mM 2-mercaptoethanol. The fractions containing Gcet1 were collected and concentrated using Vivaspin centrifugal filter devices (Vivascience, Hanover, Germany). Three BALB/c mice were injected intraperitoneally (3 times at 14-day intervals) with 100 μg 6xHis-Gcet1 fusion protein and Freund's adjuvant. A 150-μg booster of the recombinant Gcet1 protein was injected intraperitoneally, and fused 3 days later, as described previously.¹²  Hybridoma supernatants were screened by enzyme-linked immunosorbent assay (ELISA). The mouse mAb that was raised against Gcet1 (RAM) was cloned by the limiting dilution technique. Animal experiments were performed under the experimental protocol approved by the Institutional Committee for Care and Use of Animals (CEUCA 001/02).

All tissues were obtained from the tissue archives of the CNIO Tumor Bank. We used a Tissue Arrayer device (Beecher Instruments, Sun Prairie, WI) to construct several tissue microarray (TMA) blocks, according to conventional protocols.¹³  All cases were histologically reviewed and representative areas were selected. Also standard tissues sections were analyzed in some cases.

TMAs were also performed with normal tissues containing tonsil, spleen, bone marrow, brain, larynx, parotid gland, thyroid, gallbladder, liver, lung, skin, skeletal muscle, kidney, pancreas, stomach, duodenum, appendix, small and large intestine, bladder, ovary, uterus, breast, placenta, prostate, testis, fetal liver, fetal thymus, thymus, and lymph node with follicular hyperplasia.

A large series (369 cases) of human lymphoma samples was analyzed in the study. The diagnosis of the TMA tissue was confirmed histologically in all cases by central review using standard tissue sections. All cases corresponded to initial diagnostic biopsies before treatment. Histologic criteria used for diagnoses and classification of cases were those of the World Health Organization Classification.¹⁴  Some of these cases have been included in previous analyses.^{9,11,12,15,16} 

DLBCL (SUDHL4, SUDHL6, OCILY19, MHHPREB1, DOHH2, KARPAS422, DB, HT), cHD (L428, KMH2, L540, L1236, HDMYZ), BL (DAUDI, RAMOS, RAJI), T-cell lymphoma (TCL; KARPAS299, JURKAT, HUT78), myeloma (L-363, U226, KARPAS-620, LP-1), acute myeloid leukemia (AML; HL60, KG1), chronic myeloid leukemia (CML; K562,EM2), and FL (WSUNHL) cell lines were obtained from the German Collection of Microorganisms and Cell Cultures (DSMZ, Braunschweig, Germany).

The human cell lines DEV (nodular lymphocyte predominant Hodgkin lymphoma [NLPHL]), OCILY3 (DLBCL) and AKATA (BL) were kindly provided by Prof Poppema (University Medical Center Groningen, Groningen, the Netherlands), Dr Pulford (Nuffield Department of Clinical Laboratory Sciences, Oxford, United Kingdom), and Dr Campanero (Instituto de Investigaciones Biomédicas, Madrid, Spain), respectively.

Cells were grown at 37°C in 5% CO₂ in RPMI 1640 (Invitrogen, Carlsbad, CA) supplemented with 10% fetal calf serum (Sigma Chemical, St Louis, MO). KARPAS-620 medium was also supplemented with interleukin-6 (20 ng/mL; PeproTech, London, United Kingdom).

Other antibodies used were BCL-6 (clone GI191E/A8, dilution 1:120),¹²  MUM-1/IRF4 (polyclonal Ab, dilution 1:200; Santa Cruz Biotechnology, Santa Cruz, CA), and CD10 (clone 56C6, dilution 1:10; Novocastra, Newcastle, United Kingdom).

To extract total protein, cells were lysed in a buffer containing 50 mM Tris (tris(hydroxymethyl)aminomethane)-HCl, pH 7.5, 150 mM NaCl, 1% Igepal (Sigma Chemical), and protease inhibitors (Roche, Mannheim, Germany). Lysates were incubated in a cool room on a rotary shaker for 1 hour and cleared by centrifugation.

The total lysates of each cell line were denatured by heating in Laemmli sample buffer, resolved in a 10% sodium dodecyl sulfate–polyacrylamide gel, and transferred onto nitrocellulose membranes for 2 hours. Membranes were incubated overnight with blocking solution (5% milk in phosphate-buffered saline) and immunoblotted for 1 hour at room temperature with anti-Gcet1 (RAM; neat supernatant) and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) monoclonal antibody (1:10; CNIO), followed by incubation with horseradish peroxidase-conjugated secondary antibody (DAKO, Glostrup, Denmark). Finally, the blots were visualized using the ECL detection system (Amersham Biosciences, Buckinghamshire, United Kingdom) in accordance with the supplier's instructions.

Gcet1 immunohistochemical staining was performed as follows: 2- to 4-μm–thick paraffin-embedded TMA and complete tissue sections were cut onto DAKO slides (DAKO) and subsequently dewaxed, rehydrated, and subjected to antigen retrieval by heating in 50 mM Tris [tris(hydroxymethyl)aminomethane] (Trizma base)-2 mM ethylenediaminetetraacetic acid (Sigma Chemical; pH 9) in a microwave pressure cooker (Menarini Diagnostics, Wokingham, United Kingdom) at 900 W for 2 minutes. The slides were cooled and treated with peroxidase-blocking solution (DAKO) for 5 minutes. Sections were then immunostained with Gcet1 mAb by the 2-stage peroxidase-based EnVision technique (DAKO), counterstained with hematoxylin, and mounted. Incubations either omitting the specific antibody or containing unrelated antibodies were used as a control of the technique. Cytocentrifuge preparations were immunostained as previously described.¹² 

Double immunoenzymatic labeling of paraffin sections was performed using a protocol previously described.¹⁷  In the first reaction, immunostaining was performed using the EnVision peroxidase kit (DAKO) and diaminobenzidine chromogen-substrate (DAKO K5507; DAKO). In the second reaction, immunostaining was performed using the alkaline phosphatase kit (DAKO K5355; DAKO) and chromogen provided with the kit.

The combined immunoperoxidase technique and single immunofluorescence labeling was performed on paraffin sections of normal tonsil following a previously described method.^17,18  Briefly, Gcet-1 was immunostained by the 2-stage peroxidase-based EnVision technique, then washed in phosphate-buffered saline for up to 5 minutes followed by an incubation in normal human serum for up to 10 minutes before immunofluorescence labeling of the antibodies: BCL6 (green) and Ki-67 (green) in normal tonsil. Slides were examined on a Nikon E800 Eclipse fluorescence microscope (Nikon, Kingston-upon-Thames, United Kingdom) equipped for epifluorescence. Fluorescence images were captured with an Axiocam charge-coupled device camera (Zeiss, Jena, Germany) and Axiovision version 4.6 software (Imaging Associates, Bicester, United Kingdom), and adjusted using Adobe Photoshop version 9.0 software (Adobe Systems, San Jose, CA). The immunoperoxidase image was viewed by transmitting light, and corresponding gray-scale images were inverted and pasted into the red channel of the red-green-blue Photoshop image. Using this technique, cells expressing Gcet1 protein were visualized as red, whereas the immunofluorescence labeling marker was detected as green.

Gcet1 protein expression was assessed by immunohistochemistry of normal and neoplastic lymphoid human TMAs, and of complete tissue sections wherever detailed morphologic analysis and comparison with other immunohistochemical markers were deemed necessary. Signal localization, intensity, and number of positive cells were assessed. Each case was assigned to a semiquantitative category based on the number of positive cells: negative (no positive neoplastic cells), low (1%-15% positive cells), intermediate (> 15%-50% positive cells), and high (> 50%-100% positive cells).

All hematoxylin and eosin and immunohistochemistry slides were studied using an Olympus BX41 microscope (Olympus Europe, Barcelona, Spain) equipped with 20×/0.40 and 40×/0.65 objective lenses. Images were photographed using an Olympus DP70 camera device (Olympus Europe) and Axiovision version 4.6 software (Imaging Associates), and were adjusted using Adobe Photoshop version 9.0 software (Adobe Systems).

Data from cDNA microarrays from 33 FL patients showed 2.21-fold greater expression of GCET1(Serpina9) compared with reactive lymph node samples (Figure 1). With the exception of histone subunits, GCET1 expression was, in absolute terms, the strongest of all the genes included in the FL signature (Table 1). Although an increase was also seen in other GC markers, such as CD10 (MME), the magnitude of the change was highest for GCET1 (Table 1). Data charts are available (https://redlinfomas.cnio.es/webcnio/home.htm).

To confirm that RAM mAb recognized the human Gcet1 protein, immunohistochemistry on frozen cytospin preparations of V5-tagged human Gcet-1 expressed in HEK293T was performed (Figure 2). Labeling with the anti-V5 mAb confirmed the efficiency of transfection. Cytospin preparation of V5-tagged human Gcet-2 protein was used as a negative control.

These results were confirmed by Western blot using cell lysate of GCET1 and GCET2-transfected cells (Figure 3).

As shown in Figure 2, a 46-kDa band specific to the Gcet1 protein was observed in GC-DLBCL (SUDHL4 and SUDHL6), other DLBCL (KARPAS 422, DB and HT), some BL (RAJI and DAUDI), and FL (WSUNHL) cell lines. The Gcet1 protein was not present in a third group of DLBCL (MHHPREB1, DOHH2), some BL (AKATA, RAMOS), PTL (KARPAS299, HUT78, JURKATT), myeloma (L-363, U226, KARPAS-620, LP-1), AML (HL60, KG1), or CML (K562, EM2) cell lines. The NLPHL-derived DEV cell line was strongly positive, while those derived from cHL (L428, KMHH2, L540, L1236, and DHMYZ) showed no specific band.

Characterization of Gcet1 protein expression in normal human tissues was performed on paraffin-embedded tissue sections using the anti-Gcet1 mAb. Gcet1 expression was restricted to lymphoid tissue, specifically to GC B cells. It was found in the germinal centers present in tonsil, lymph node, and spleen white pulp. No expression was found in the normal thymus or in the other normal fetal and adult tissues listed in “Tissue samples and preparation of tissue microarrays.” In GC B cells, Gcet1 protein was identified in the cytoplasm of the cell, with a granular pattern, frequently with a perinuclear halo or dot (Golgi staining). In the GC, it stains centroblasts (large noncleaved cells) and large centrocytes (large cleaved cells) but not small centrocytes (small cleaved cells) or other accessory cells. No clear preference was found between dark and light or outer zones of the GC. Only exceptionally scattered large B cells in the mantle zone were positively stained with Gcet1 mAb. Comparison of Gcet1 with Bcl6 and CD10 immunostaining in follicular hyperplasia cases revealed a distinctive staining pattern for the new marker, by which a subpopulation of Bcl6-CD10 double-positive cells was selected. Double immunoenzymatic staining with Gcet1 and MUM-1/IRF4 demonstrated a mutually exclusive staining in most of cells, but with a distinct small population of Gcet1-MUM1/IRF4 double-positive cells (Figure 2).

Using a double immunofluorescence technique in normal tonsil, Gcet1 was found to be expressed in a relatively large proportion of Bcl6-positive cells. However, some Bcl6-positive cells were negative with Gcet1. Double immunostaining for Ki67 and Gcet1 demonstrated that all Gcet1⁺ cells were double-positive with Ki-67, that is, in the proliferative phase, while some proliferating cells were negative with Gcet1 (Figure 2).

The immunostaining results on paraffin sections from 369 patients with lymphoma are summarized in Table 2. Gcet1 was expressed in a large number of B-cell neoplasms, especially in those thought to derive from GC B cells (82 of 89, or 92%, in FL and 7 of 19, or 37%, in BL). Some FLs exhibited only scattered Gcet1⁺ cells. These positive cells were found in GC and marginal zone areas, while interfollicular B cells were usually negative (Figure 4). In BL-positive cases, the staining was diffuse, with the characteristic cytoplasmic localization (Figure 4). However, some cases of FL and BL were negative. None of the other small B-cell lymphomas, that is, chronic lymphocytic leukemia (CLL; 20 cases), splenic marginal zone lymphoma (MZL; 19 cases), and mantle cell lymphoma (MCL; 19 cases), showed staining with Gcet1 outside entrapped nonneoplastic germinal centers (data not shown).

Roughly half of the DLBCL cases (34 of 72, or 47%) were positive for Gcet1. In 40% of the positive cases, the staining was diffuse and intense in more than 50% of the neoplastic population. However, in the other 60% of positive cases, the staining was heterogeneous, with positive cells accounting for less than 50% of the neoplastic population (Figure 4).

When DLBCLs were divided using the Hans algorithm into GC and non-GC types,¹⁹  cases of GC types showed increased expression of Gcet1, although the match was not perfect (GC: 21 of 31 +; non-GC: 3 of 21 +).

More than half (7 of 13 cases, or 54%) of T cell–rich B-cell lymphomas were Gcet1⁺. In such cases, the staining was intense and selective for large neoplastic B cells (Figure 4).

In Hodgkin lymphoma, Gcet1 recognized almost all the NLPHL cases (38 of 40 cases, or 95%), a subset of LRCHL (2 of 15 cases, or 13%) but none of the nodular sclerosis (NS) and mixed celluarity (MC) cases (56 cases). In NLPHL the staining was intense and specific for the LH cells (Figure 4). Only scattered Gcet1⁺ blast B cells were observed. These tissue protein expression results fit well with those from Western blot in DEV cell line a NLPHL-derived cell line (Figure 2).

None of the 18 cases of PTCL NOS evaluated was positive with Gcet1.

Gcet1 (centerin) is a recently described molecule belonging to the family of serine-protease inhibitors, or serpins.^1,–3  The gene is located on chromosome 14q32 and has 4 splicing isoforms, of which the longest (GCET1A) is 1787 bp in length and encodes a 435-amino acid protein. The Gcet1 primary amino acid sequence is homologous with the noninhibitory carrier protein TBG, with which it shares a 50% amino acid identity. However, it also possesses an arginine residue at position P1, suggesting that it may inhibit trypsinlike serine proteases, that is, endopeptidases, that cleave C-terminal to basic arginine and lysine residues.²⁰  Alignment of the deduced Gcet1 protein sequence with several other serpin family members reveals that the protein possesses all the motifs necessary for it to become an active protease inhibitor. Furthermore, it has recently been demonstrated that Gcet1 is likely to function in vivo in the GC as an efficient inhibitor of a trypsin-like protease.³  However, the nature of the Gcet1 target protease(s) is still unknown. Furthermore, it has been demonstrated that GCET mRNA is up-regulated in naive B cells stimulated in vitro via CD40 signaling.¹ 

As expected, our immunohistochemical study found expression of Gcet1 in GC B cells within secondary follicles, but not in primary follicles. In relation to the proliferative nature of the GC, other serpins have been found to serve as inhibitors of apoptosis, that is, PI-9 and CrmA.^21,22  They share an acidic reactive center loop residue with Gcet1. This is an aspartic acid P3′ residue in Gcet1. These similarities imply that Gcet1 could be an inhibitor of apoptosis in GCs. Considering also its CD40-dependent up-regulation, it could be hypothesized that Gcet1 may be part of some of the downstream events after triggering of the CD40 molecule, that is, rescue of GC B cells from apoptosis, isotype switching, somatic hypermutation, affinity maturation, and induction of memory B cells. However, the exact role of Gcet1 in the initiation and maintenance of the T cell–dependent GC reaction remains to be fully elucidated.

Previous reports demonstrated that the expression of GCET1 mRNA was restricted to GC B cells and their malignant counterparts. Northern blot studies revealed a restricted transcription to lymph node samples, the Burkitt-derived cell line Raji and HL-60 promyelocytic lines, and its absence from other normal tissues (including bone marrow, thymus, and spleen) and cancer cell lines.¹  We found no expression of Gcet1 in HL-60 and KG1 cell lines, which is consistent with the lack of expression of the gene in cells of myeloid lineage and with the findings of the TMA study of normal tissues, in which Gcet1 expression was limited to GC B cells.

Our results demonstrate that Gcet1 protein expression is restricted to a subset of GC B cells and derived neoplasms, thereby establishing a distinct heterogeneity among normal and neoplastic GC B cells.

Therefore, Gcet1 seems to be expressed exclusively by neoplasms hypothetically derived from GC B cells, including FL, NLPHL, and a subset of DLBCL, T-cell/histiocyte rich B-cell lymphoma (T/HRBCL), and BL. Thus, Gcet1 is expressed in the great majority of the FL cases (82 of 89, or 92%). None of the other small B-cell lymphomas, that is, CLL (20 cases), splenic marginal zone lymphoma (19 cases), and MCL (19 cases), was positive with Gcet1, which underlines the potential utility of Gcet1 expression in lymphoma diagnosis.

Specifically, this new marker may serve to distinguish follicular center cell lymphoma from other non-GC small B-cell lymphomas (ie, MZL and MCL) with a follicular pattern of growth or with extensive follicular colonization. In this regard, it has been established that approximately 10% of FLs show marked marginal zone differentiation. Sometimes this kind of differentiation is so pronounced that it may be almost impossible to differentiate this type of FL from a primary marginal zone lymphoma with follicular colonization on purely morphologic grounds.²³  Under these conditions, Gcet1 remains positive in the marginal zone when the neoplastic cells are of follicular center origin, in contrast with the negativity seen in the interfollicular B cells.

In addition, microarray and protein expression data reveal a stronger degree of expression of GCET1 mRNA and protein in FL cases than in reactive lymphadenitis showing follicular hyperplasia. The increase in the expression levels was more pronounced than that observed in other well-known GC markers, such as CD10 and BCL6. Taking into account the relatively low percentage of Gcet1⁺ GC B cells compared with CD10- and bcl6-positive cells in the GC, the difference seems likely to be of significance. Thus, although more accurate studies with GC microdissection in follicular hyperplasia cases may be required to avoid the dilutional effect of other non-GC cells, it seems that GCET1 is overexpressed in neoplastic FL cells. This should prompt further research into the survival advantage that increased GCET1 expression may confer on GC cells.

It is striking that several FL cases do not express this protein. This unexpected heterogeneity recalls that observed in reactive GCs and implies that Gcet1 expression could be used for a more refined division of normal and neoplastic GCs. The previous observation that the GC signature (including Gcet1) is associated with ongoing SHM suggests an interesting research hypothesis.²⁴ 

Greater heterogeneity is observed in BL and DLBCL cases. Our results in BL cell lines completes the picture drawn by Frazer et al.¹  Using Western blot and immunohistochemistry, we found heterogeneous expression of Gcet1 in BL, whereby some cases did not express the protein (including AKAYA and RAMOS cell lines). The significance of these results is unclear, although it could be explained if BL were derived from a cell in 2 possible stages of differentiation, an early centroblast or a post-GC memory cell, as proposed by Bellan et al.²⁵  More studies about the mutational status of the V_H genes in Gcet1⁺ GC B cells are needed to address the issue of the temporal modulation of GCET1 transcription and its possible relationship with the different stages of differentiation of the GC B cells.

In DLBCL, 47% of the cases were positive for Gcet1. Of these, GC-type cases according to the Hans algorithm showed preferential Gcet1 expression compared with the ABC type, although with imperfect matching. Further studies are needed to determine whether Gcet1 expression can identify DLBCL cases with different clinical courses or biologic features, as has been demonstrated for another new GC markers as Gcet2-HGAL.²⁶ 

More than half (7 of 13 cases, or 54%) of the T cell–rich B-cell lymphomas studied were positive with Gcet1. In such cases, the staining was similar to the pattern observed in NLPHL. Our results suggest that at least a proportion of our T/HRBCLs show follicular center differentiation consistent with the postulated GC origin of this type of lymphoma.²⁷  However, the negative cases could indicate a non-GC origin for some T/HRBCLs. Thus, Gcet1 may be able to divide T/HRBCL into 2 groups, one of GC B-cell origin and the other of non-GC B-cell origin, reflecting the scenario found in DLBCLs. These observations also demonstrate that Gcet1 serves to identify a common GC trait shared with other lymphomas at the GC differentiation stage, including NLPHLs. As stated before, accumulated evidence indicates that NLPHL is derived from highly mutated GC B cells.²⁸  The strong expression of Gcet1 in NLPHL, as observed in the majority of cases, is consistent with this assumption, and is evidence of the usefulness of Gcet1 as a marker of GC differentiation. Furthermore, we found a subgroup of LRCHL that seems to retain some traits of GC derivation. This subset could keep the expression of the B-cell lineage repertoire, which is lost in the majority of cases, thereby establishing a closer relation between it and NLPHL.²⁹  Similar results in CHL were obtained with HGAL, another marker of GC derivation.^30,31 

In conclusion, we found that Gcet1 is highly restricted to a subset of GC B cells and GC-derived lymphomas, which therefore establishes a distinctive heterogeneity among the BLs and DLBCLs studied here. In addition, Gcet1 expression could be used as a potential diagnostic marker in the recognition of T/HRBCL, NLPHL, and a subset of CHL, and in the differential diagnosis between FL and MZL or MCL. Further work is required to clarify other potential diagnostic uses of the antibody and the biologic significance of GCET1 expression.

We thank María Encarnación Castillo for technical assistance and Dr L. Colomo for his useful suggestions.

This study was supported by grants G03/179, PI051623, PI052800 from the Ministerio de Sanidad y Consumo and by grants SAF2005-00 221 and SAF2004-04 286 from the Ministerio de Ciencia y Tecnología, and supported by the Fundacion la Caixa, Spain.

Contribution: S.M.-M. designed and performed research, analyzed the data, and wrote the paper; G.R. designed and performed research, contributed vital new reagents, and wrote the paper; L.M. performed research and contributed vital new reagents; N.M. analyzed the data; L.S.-V. performed research; F.C and J.C. performed research and analyzed the data; J.L.M.-T. performed research and contributed vital new reagents; Y.S. contributed vital new reagents; W.C.C. contributed vital new reagents and drafted the manuscript; and M.A.P. designed and performed research and wrote the paper.

S.M.-M. and G.R. contributed equally to this study.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Santiago Montes-Moreno, Molecular Pathology Programme, Spanish National Cancer Centre, Melchor Fernández Almagro 3, 28029 Madrid, Spain; e-mail: smontes@cnio.es.