Main

Most primary non-Hodgkin lymphomas of the stomach can be classified into the indolent mucosa-associated-lymphoid-tissue lymphomas derived from marginal zone B cells and the aggressive diffuse large B-cell lymphomas.1

In 20% of gastric diffuse large B-cell lymphomas, a concurrent component of mucosa-associated-lymphoid-tissue lymphoma can be identified. About 65% of these diffuse large B-cell lymphoma/mucosa-associated-lymphoid-tissue lymphomas are associated with Helicobacter pylori (H. pylori) infection and can be cured by H. pylori eradication therapy. These clinical features of diffuse large B-cell lymphoma/mucosa-associated-lymphoid-tissue lymphomas are similar to those of mucosa-associated-lymphoid-tissue lymphomas, suggesting a common marginal zone B-cell origin with mucosa-associated-lymphoid-tissue lymphomas, and that the large cells are transformed from the mucosa-associated-lymphoid-tissue lymphomas.2, 3

In contrast, most gastric diffuse large B-cell lymphomas do not have a component of mucosa-associated-lymphoid-tissue lymphoma. Data on the clinical behavior, the cellular origin, and molecular features of these diffuse large B-cell lymphomas are still limited. Among these diffuse large B-cell lymphoma diffuse large B-cell lymphomas, the incidence of H. pylori infection varies from 15 to 50%.2, 3 Although higher expression of BCL6, the master transcription factor for the formation of germinal center B cells, is a predictor of a better prognosis in gastric diffuse large B-cell lymphomas,4 it is unclear how H. pylori infection may affect directly this molecular event and secondarily the clinical manifestations.

Epithelial–mesenchymal transition is a process occurring during embryonic development. Through epithelial–mesenchymal transition, epithelial cells lose polarity and intercellular contact, and become mesenchyme-like with an increased migratory ability.5 ZEB1 (zinc-finger E-box-binding homeobox 1), which is regulated by miR-200, is a transcription factor for the induction of epithelial–mesenchymal transition in epithelial cells. The process of epithelial–mesenchymal transition is also found in cancer cells of the epithelial origin. Increased expression of ZEB1 or decreased expression of miR-200 is associated with epithelial–mesenchymal transition and metastasis of cancer cells.6, 7

Although studies on ZEB1 focus mostly on its roles in epithelial–mesenchymal transition and metastasis, it is becoming evident that ZEB1 is also critical in the development of hematopoietic cells, including both B cells and T cells. For the B-lineage, ZEB1 may inhibit BCL6, the master transcription factor for the formation of germinal center B cells,8 and is required for the normal development of marginal zone B cells and the production of immunoglobulin M.9, 10 Consistent with these data, overexpression of ZEB1 induces a more aggressive behavior of diffuse large B-cell lymphoma of the lymph nodes.11 For the T-lineage, ZEB1 regulates T-cell development.12, 13, 14 Loss of ZEB1 in knockout mice accelerates the development of adult T-cell leukemia.15

In this study, we found that H. pylori-positive gastric diffuse large B-cell lymphomas had less lymph node metastasis and a better response to chemotherapy. Genome-wide expression profiles of both miRNAs and mRNAs identified inhibition of ZEB1 by miR-200, with secondary induction of BCL6, as one of the molecular events that characterized H. pylori-positive gastric diffuse large B-cell lymphomas, and partially accounted for the less aggressive behavior of H. pylori-positive gastric diffuse large B-cell lymphomas.

Materials and methods

Tissue Samples

From the archives at the Pathology Department of the National Taiwan University Hospital, cases with a diagnosis of gastric diffuse large B-cell lymphoma after 2000 were selected. Cases with incomplete induction therapy, with additional H. pylori eradication therapy, or with a component of mucosa-associated-lymphoid-tissue lymphoma at the time of diagnosis or during follow-up were excluded. A total of 102 cases were finally selected. The ethics committee of the National Taiwan University Hospital approved this study. The diagnosis of diffuse large B-cell lymphoma was based on the WHO criteria, characterized by diffuse infiltrates of large lymphoid cells, without a detectible component of mucosa-associated-lymphoid-tissue lymphoma. All the specimens were first examined on serial H&E sections. A case with dense clusters of curved spiral bacilli was declared H pylori-positive. A Diff-Quik stain and a slide urease test were performed if the results of H&E stains were equivocal or negative. Clinical staging was performed with endoscopic ultrasonography, abdominal CT, or MRI. Stage-IE diffuse large B-cell lymphomas were limited to the stomach. Stage-IIE1 diffuse large B-cell lymphomas involved both the stomach and peri-gastric lymph nodes. Stage-IIE2 diffuse large B-cell lymphomas involved both the stomach and more distant abdominal lymph nodes, such as mesenteric and para-aortic lymph nodes. Stage-IIIE diffuse large B-cell lymphomas involved extra-abdominal lymph nodes. Stage-IVE diffuse large B-cell lymphomas involved the bone marrow. Clinical data, treatments, and outcomes were extracted from the medical records.

Genome-Wide miRNA Profiles in 18 Gastric Diffuse Large B-cell Lymphomas

A series of 18 gastric diffuse large B-cell lymphomas including nine H. pylori-positive and nine H. pylori-negative gastric diffuse large B-cell lymphomas were used. The nCounter miRNA Expression Analysis System (NanoString Technologies, Seattle, WA, USA) was used to obtain the expression profiles of 633 human miRNAs.16, 17, 18 A panel of six synthetic miRNAs was used as internal positive controls. Approximately 0.1 μg of total RNAs were extracted from formalin-fixed paraffin-embedded tissue blocks. Mature miRNAs were ligated to sequence-specific miRtags, and the miRtagged-mature miRNAs were hybridized to fluorescent reporter probes and to capture probes. After hybridization, the samples were immobilized to a solid-phase cartridge, and a CCD digital analyzer was used for data acquisition. The data were normalized to the internal positive controls. The normalized data were submitted as Supplementary Table S1.

Genome-Wide mRNA Expression Profiles in 18 Gastric Diffuse Large B-cell Lymphomas

The Agilent SurePrint G3 Human GE 8 × 60 K Microarrays were used according to the manufacturer’s recommendations (Agilent Technologies, Santa Clara, CA, USA). Approximately 0.2 μg of total RNAs were reverse-transcribed, amplified, and Cy3-labeled with a Low Input Quick-Amp Labeling kit. The Cy3-labeled cRNAs were fragmented to an average size of 50–100 nucleotides and hybridized to the microarrays at 65 °C for 40 h. After washing, the microarrays were scanned with an Agilent microarray scanner at 535 nm for Cy3. The scanned images were analyzed with the Feature Extraction 10.5.1.1 software. The normalized data are submitted in Supplementary Table S2.

Bioinformatics

Two-sample comparisons between nine H. pylori-positive and nine H. pylori-negative gastric diffuse large B-cell lymphomas were performed. Hierarchical clustering was performed with the Genesis software.19

Quantitative RT-PCR for miRNA-200

Total RNAs were extracted from formalin-fixed, paraffin-embedded tissue blocks with the Trizol method. A stem-loop RT-PCR for miRNAs was performed according to a published protocol with modification.20, 21

For RT, the reverse primers were: 5′-N44-ACATCG-3′ for miR-200a, 5′-N44-TCATCA-3′ for miR-200b, and 5′-N44-TCCATC-3′ for miR-200c, where N44 is 5′-GTCGTATCCATGGCAGGGTCCGAGGTATTCGCCATGGATACGAC-3′. The RT mixture included 0.1 μM miRNA-specific stem–loop RT primer, 10 mM dNTPs, 50 mM Tris-HCl at pH 8.3, 75 mM KCl, 3 mM MgCl2, 0.1 M DTT, 100 units of reverse transcriptase, and 1 μg RNA in a final volume of 20 μl. The mixture was incubated at 42 °C for 80 min.

For PCR, an universal reverse PCR primer, 5′-TGGCAGGGTCCGAGGT-3′, and miRNA-specific forward primers were used. The specific forward primers were: 5′-GGGTAACACTGTCTGGTAA-3′ for miR-200a, 5′-GGGTAATACTGCCTGGTAA-3′ for miR-200b, and 5′-GGGTAATACTGCCGGGTAAT-3′ for miR-200 c. Real-time PCRs were carried out in a 20-μl reaction mixture that included 5 μl diluted cDNA, 0.1 μM miRNA-specific forward primer, 0.1 μM universal reverse primer, and 10 μl of SYBR Green PCR master mix (TOYOBO, Tokyo, Japan). A PCR cycle consisted of denaturation at 95 °C for 15 s and annealing at 60 °C for 1 min. The PCR products, 54∼58 bp for miRNAs and 140 bp for U6, were confirmed by 8% acrylamide gel and sequencing. The difference between the threshold cycles of a specific miRNA and U6 snRNA, dCt (Ct–Ct), was calculated from the mean values of duplicate measurements. This definition of dCt implies an inverse correlation between dCts and miRNA levels.

Immunohistochemistry for ZEB1 and BCL6

Immunoperoxidase stains on formalin-fixed, paraffin-embedded tissue sections with a ZEB1 antibody (1:100, rabbit polyclonal, Novus, Littleton, CO, USA) and a BCL6 antibody (1:100, PG-B6P, mouse monoclonal, Abcam, Cambridge, MA, USA). Antigen retrieval was performed in Tris buffer at pH 6 for ZEB1 and at pH8 for BCL6. The primary antibody was applied to the slides at 37 °C for 45 min. Biotin-conjugated secondary antibodies, peroxidase-conjugated streptavidin, and DAB were used sequentially for completing the reactions. The percentages of positive cells, as averages of 10 randomly selected high-power fields, were determined.

Results

H. pylori-Positive Gastric Diffuse Large B-cell Lymphomas have Less Lymph Node Metastasis and a Better Response to Chemotherapy than H. pylori-Negative Gastric Diffuse Large B-cell Lymphomas

A series of 102 gastric diffuse large B-cell lymphomas without a concurrent component of mucosa-associated-lymphoid-tissue lymphoma was used to investigate the differences in the clinical manifestations between 48 H. pylori-positive and 54 H. pylori-negative groups (Table 1).

Table 1 Clinical features of H. pylori-positive and H. pylori-negative gastric diffuse large B-cell lymphomas
Full size table

Regardless of the H. pylori status, gastric diffuse large B-cell lymphomas do not show a gender bias. At the time of presentation, the patients have a median age of about 60 years. However, the H. pylori-positive group is more likely to present at an early stage than the H. pylori-negative group (73% vs 52% at stage I/II, P=0.03). Consistent with the earlier staging, the H. pylori-positive group has a better Eastern Cooperative Oncology Group (ECOG) performance status (P=0.023), is more likely to have a normal LDH level (P=0.008), and belongs to a lower International Prognostic Index (IPI) risk group (P=0.009).

The H. pylori-positive and H. pylori-negative groups received similar chemotherapy regimens. Significantly, 75% of the H. pylori-positive group but only 43% of the H. pylori-negative group achieved complete remission (P=0.001). Consistent with the better complete remission rate, the H. pylori-positive group had a higher 5-year event-free survival rate (73% vs 29%, P<0.001) and a higher 5-year overall survival rate (74% vs 35%, P<0.001) than the H. pylori-negative group. Note that the outcome parameters were all better in the H. pylori-positive group than in the H. pylori-negative group by a margin of at least 30%, which cannot be totally attributed to the 13% difference in the percentages of patients treated with Rituximab therapy (31% in the H. pylori-positive group vs 18% in the H. pylori-negative group).

To summarize the clinical comparisons, we found that H. pylori-positive gastric diffuse large B-cell lymphomas have less lymph node metastasis and a better response to chemotherapy than H. pylori-negative gastric diffuse large B-cell lymphomas. H. pylori-positive gastric diffuse large B-cell lymphoma is, therefore, a less aggressive lymphoma than H. pylori-negative gastric diffuse large B-cell lymphoma.

H. pylori-Positive Gastric Diffuse Large B-cell Lymphomas have Higher Levels of miR-200 than H. pylori-Negative Gastric Diffuse Large B-cell Lymphomas

To identify a molecular mechanism underlying the less aggressive behavior of H. pylori-positive gastric diffuse large B-cell lymphomas, we obtained genome-wide miRNA expression profiles in nine H. pylori-positive and nine H. pylori-negative gastric diffuse large B-cell lymphomas. This series is independent of the clinical series of 102 cases.

Out of 633 miRNAs, eight miRNAs have a difference in the expression levels between H. pylori-positive and H. pylori-negative groups (Table 2). Most significantly, miR-200a, miR-200b, and miR-200c are the three miRNAs with the most significant differential expressions (Table 2; Figure 1a). The levels for miR-200 a, b, or c are all increased by 2.5- to 3-fold in the H. pylori-positive group at a P-value <0.01.

Table 2 miRNAs differentially expressed between H. pylori-negative and H. pylori-positive gastric diffuse large B-cell lymphomas
Full size table
Figure 1
figure 1

Inhibition of ZEB1 by miRNA-200 characterized H. pylori-positive gastric diffuse large B-cell lymphomas. (a) H. pylori-positive gastric diffuse large B-cell lymphomas have higher levels of miR-200 a, b, and c. y axis: miR-200 expression levels measured by the NanoString technology; x axis: miR-200, a, b, or c in nine HP-negative (white) or nine HP-positive (gray) gastric diffuse large B-cell lymphomas. (b) H. pylori-positive gastric diffuse large B-cell lymphomas have lower levels of ZEB1. y axis: ZEB1 expression levels measured by microarrays; x axis: ZEB1 in nine H. pylori-negative (white) or nine H. pylori-positive (gray) gastric diffuse large B-cell lymphomas. (c) Inverse correlations between ZEB1 and miR-200 in gastric diffuse large B-cell lymphomas. y axis: ZEB1 levels; x axis: miR-200 a, b, or c. Gray: nine H. pylori-negative gastric diffuse large B-cell lymphomas; black: nine H. pylori-positive gastric diffuse large B-cell lymphomas. Correlation coefficients at −0.60 at P=0.008 for miR-200a, −0.71 at P<0.001 for miR-200b, and −0.63 at P=0.005 for miR-200c, respectively.

Full size image

H. pylori-Positive Gastric Diffuse Large B-cell Lymphomas have Lower Levels of ZEB1 than H. pylori-Negative Gastric Diffuse Large B-cell Lymphomas

We also obtained genome-wide mRNA expression profiles in nine H. pylori-positive and nine H. pylori-negative gastric diffuse large B-cell lymphomas.

Out of 42 405 probes, 30 mRNAs were found to have different expression levels between H. pylori-positive and H. pylori-negative groups (Table 3). Owing to our interests in identifying a miRNA-mediated mechanism that distinguishes H. pylori-positive from H. pylori-negative gastric diffuse large B-cell lymphomas, we used the miR.org program to identify potential miR-200 targets out of the list of 30 mRNAs.22, 23 Two mRNAs FUBP3 (far upstream element binding protein 3)24 and ZEB1 were thus identified.

Table 3 mRNAs differentially expressed between H. pylori-negative and H. pylori-positive gastric diffuse large B-cell lymphomas
Full size table

ZEB1 is a well-characterized target of miR-200,6 whereas—to our knowledge—no experimental data are available on FUSBP3 as a target of miR-200. Inhibition of ZEB1 by miR-200 is, therefore, a molecular mechanism that may separate H. pylori-positive from H. pylori-negative gastric diffuse large B-cell lymphomas. The mean level of ZEB1 was reduced from 0.37 in H. pylori-negative to 0.26 in H. pylori-positive gastric diffuse large B-cell lymphomas, P=0.025 (Table 3; Figure 1b).

Figure 1c further demonstrates a significant inverse correlation between ZEB1 and miR-200 a, b, or c, with a correlation coefficient at −0.60 at P=0.008, −0.71 at P<0.001, and −0.63 at P=0.005, respectively.

Inhibition of ZEB1 by miR-200 Distinguishes H. pylori-Positive from H. pylori-Negative Gastric Diffuse Large B-cell Lymphomas

Using the miR.org program, we found that miR-200 inhibits ZEB1 and FUBP3, and miR-544 inhibits CYFIP2. No other interactions between the eight miRNAs and 30 mRNAs were identified. These findings further support a critical role of mIR-200 and ZEB1 in H. pylori-positive gastric diffuse large B-cell lymphomas (Table 3).

We performed a hierarchical clustering, using all eight miRNAs from Table 2 and 30 mRNAs from Table 3. As shown in Figure 2a, cluster I included eight H. pylori-positive gastric diffuse large B-cell lymphomas, whereas cluster II included one H. pylori-positive and nine H. pylori-negative gastric diffuse large B-cell lymphomas (8:0 vs 1:9, P<0.001).

Figure 2
figure 2

Hierarchical clustering of nine H. pylori-positive and nine H. pylori-negative gastric diffuse large B-cell lymphomas. (a) Hierarchical clustering with the whole panel of eight miRNAs and 30 mRNAs generated cluster I with eight H. pylori-positive gastric diffuse large B-cell lymphomas, and cluster II with nine H. pylori-negative gastric diffuse large B-cell lymphomas plus one H. pylori-positive gastric diffuse large B-cell lymphomas (8:0 vs 1:9, P<0.001). (b) Hierarchical clustering with miR-200 and ZEB1 generated cluster I with eight H. pylori-positive gastric diffuse large B-cell lymphomas and two H. pylori-negative gastric diffuse large B-cell lymphomas, and cluster II with seven H. pylori-negative gastric diffuse large B-cell lymphomas and one H. pylori-positive gastric diffuse large B-cell lymphomas (8:2 vs 1:7, P=0.007).

Full size image

To estimate the significance of the miR-200-ZEB1 pathway, we used only miR-200 and ZEB1 for a hierarchical clustering. Cluster I included eight H. pylori-positive and two H. pylori-negative gastric diffuse large B-cell lymphomas, whereas cluster II included one H. pylori-positive and seven H. pylori-negative gastric diffuse large B-cell lymphomas (8:2 vs 1:7, P=0.007) (Figure 2b).

The miR-200-ZEB1 pathway is almost as effective as the combined eight miRNAs and 30 mRNAs in distinguishing H. pylori-positive from H. pylori-negative gastric diffuse large B-cell lymphomas. Therefore, the miR-200-ZEB1 is a critical pathway in separating H. pylori-positive form H. pylori-negative gastric diffuse large B-cell lymphomas. Of course, the roles of the other miRNAs in Table 2 and mRNAs in Table 3 remain to be explored.

Histopathology, Immunohistochemistry for ZEB1 and BCL6, and qRT-PCR for miR-200 in H. pylori-Positive and H. pylori-Negative Gastric Diffuse Large B-cell Lymphomas

Owing to the small sample size used in genome-wide expression profiles, we used a second series of 30 H. pylori-positive and 30 H. pylori-negative gastric diffuse large B-cell lymphomas to confirm inhibition of ZEB1 by miR-200 in H. pylori-positive gastric diffuse large B-cell lymphomas. This series is part of the clinical series of 102 cases.

Figure 3 shows that H. pylori-positive gastric diffuse large B-cell lymphomas are characterized by the presence of Helicobacter organisms, frequently surrounded by increased eosinophils, whereas H. pylori-negative gastric diffuse large B-cell lymphomas do not have Helicobacter organisms and demonstrate minimal eosinophilic clusters (40% vs 4%, P=0.02, Figure 3; Table 4).

Figure 3
figure 3

Histopathology of gastric diffuse large B-cell lymphomas. (a) Presence of H. pylori can be identified on H&E sections of H. pylori-positive gastric diffuse large B-cell lymphomas. (b) H. pylori-positive gastric diffuse large B-cell lymphomas have more inflammatory infiltrates, such as clusters of eosinophils. (c) H. pylori-negative gastric diffuse large B-cell lymphomas have no evidence of H. pylori even on a Diff-Quik stain. (d) H. pylori-negative gastric diffuse large B-cell lymphomas have less inflammatory infiltrates, without clusters of eosinophils. (a, c, d) H&E stains at × 1000; (b) Diff-Quik stain for H. pylori at × 1000. All stains were performed on formalin-fixed paraffin-embedded sections. An Olympus BX60 microscope with × 10/0.22 ocular lens and × 100/1.35 objective lens was used (Olympus, Tokyo, Japan). Arrows indicate the presence of the bacteria, Helicobacter pylori.

Full size image
Table 4 Pathologic features of H. pylori-negative and H. pylori-positive gastric diffuse large B-cell lymphomas
Full size table

Immunohistochemistry for ZEB1 was performed and confirmed weaker expression of ZEB1 in H. pylori-positive gastric diffuse large B-cell lymphomas (42% vs 25%, P<0.001, Figures 4a and b; Table 4). Immunohistochemistry for BCL6 was also performed and showed stronger expression of BCL6 in H. pylori-positive gastric diffuse large B-cell lymphomas (23% vs 33%, P=0.002, Figures 4a and c; Table 4). Finally, qRT-PCR for miR-200 was performed and confirmed higher levels of miR-200 in H. pylori-positive gastric diffuse large B-cell lymphomas (dCt: 3.97 vs 0.26 for miR-200a at P=0.013, 3.64 vs 0.41 for miR-200b at P=0.025, 1.21 vs −1.65 for miR-200c at P=0.018, Figure 4d; Table 4).

Figure 4
figure 4

Immunohistochemistry for ZEB1 and BCL6 and qRT-PCRs for miR-200 in gastric diffuse large B-cell lymphomas. (a) Immunoperoxidase stains for ZEB1 and BCL6 in H. pylori-positive and H. pylori-negative gastric diffuse large B-cell lymphomas ( × 1000). An Olympus BX60 microscope with × 10/0.22 ocular lens and × 100/1.35 objective lens was used (Olympus). (b, c) Distributions for the percentages of ZEB1-positive (b) or BCL6-positive (c) cells in gastric diffuse large B-cell lymphomas. y axis: percentages of positive cells; x axis: H. pylori-negative vs H. pylori-positive gastric diffuse large B-cell lymphomas. (d) qRT-PCRs for miR-200 a, b, or c in gastric diffuse large B-cell lymphomas. y axis: dCt; x axis: H. pylori-negative vs H. pylori-positive gastric diffuse large B-cell lymphomas.

Full size image

As ZEB1 suppresses BCL6, and BCL6 is a known predictor of a better prognosis in gastric diffuse large B-cell lymphomas, the immunohistochemistry and qRT-PCR data are compatible with the model, in which miR-200 inhibits ZEB1 with secondary induction of BCL6 in H. pylori-positive gastric diffuse large B-cell lymphomas with a better prognosis.

Discussion

Although gastric diffuse large B-cell lymphomas are often associated with H. pylori infection, the biologic and clinical significance of the association is still unclear. In this study, we demonstrated that H. pylori-positive gastric diffuse large B-cell lymphomas have less lymph node metastasis and a better response to chemotherapy, and that this behavior could be attributed, at least partially, to a higher level of miR-200, which inhibits ZEB1 with a secondary increase in BCL6.

ZEB1 is known to increase the potential for metastasis in cancers of epithelial origin through induction of epithelial–mesenchymal transition. ZEB1 inhibits the cell adhesion molecule E-cadherin, and, as a result of the loss of cell adhesion, increases migratory ability of the cancer cells.5, 6, 7

In this study, we found that expression of ZEB1 in H. pylori-negative gastric diffuse large B-cell lymphomas is also associated with increased lymph node metastasis. However, in gastric diffuse large B-cell lymphomas, genome-wide mRNA expression profiles showed that the levels of E-cadherin are very low, regardless of the level of ZEB1 (Supplementary Table S2). Therefore, ZEB1 must have increased metastatic potential of gastric diffuse large B-cell lymphomas through another mechanism. Interestingly, loss of BCL6 is associated with disease progression in breast cancer and gastric cancer,25, 26 suggesting that the loss of BCL6 might be responsible for the increased lymph node metastasis in gastric diffuse large B-cell lymphomas too.

Diffuse large B-cell lymphomas of the lymph nodes can be classified into a germinal center B-cell type derived from germinal center B cells and an activated B-cell type derived from activated B cells. The germinal center B-cell type has a better prognosis than the activated B-cell type.27 As BCL6 is the master transcription factor for germinal center B cells,28 the association of higher expression of BCL6 with a better prognosis in gastric diffuse large B-cell lymphomas is consistent with the finding that nodal diffuse large B-cell lymphomas of the germinal center B-cell type have a better prognosis.4, 27 However, BCL6 may contribute to a better prognosis through additional pathways other than promoting germinal center B-cell differentiation. The detailed molecular pathways of how BCL6 contributes to a better prognosis remain to be clarified.

We have set a threshold for expression level at 0.1 to identify a list of 30 mRNAs differentially expressed in H. pylori-positive and H. pylori-negative gastric diffuse large B-cell lymphomas. Owing to its low expression level below the threshold of 0.1, BCL6 is not one of the 30 mRNAs, despite its association with H. pylori-positive gastric diffuse large B-cell lymphomas.

H. pylori infection is also found in gastric mucosa-associated-lymphoid-tissue lymphoma, and translocation of the H. pylori-encoded oncoprotein cagA into the nuclei of gastric mucosa-associated-lymphoid-tissue lymphoma cells is associated with a better prognosis.29, 30 It is possible that cagA can likewise translocate into the nuclei of gastric diffuse large B-cell lymphomas to produce a better prognosis. The potential mechanisms that might have connected cagA to the activation of miR-200 remain to be clarified.

In summary, we found that H. pylori-positive gastric diffuse large B-cell lymphomas have less lymph node metastasis and a better response to chemotherapy, both of which could be attributed partially to the inhibition of ZEB1 by miR-200, with secondary induction of BCL6. The model is consistent with previous reports on the associations between ZEB1 and a more aggressive behavior of nodal diffuse large B-cell lymphomas11 and between BCL6 and a better prognosis in gastric diffuse large B-cell lymphomas.4, 31 Further studies would be needed to clarify the details of the underlying molecular mechanisms how BCL6 reduces lymph node metastasis and/or induces sensitivity to chemotherapy. Finally, the data may be clinically important, as we have recently found that some H. pylori-positive gastric diffuse large B-cell lymphomas can be cured by H. pylori eradication therapy.32 Further studies on the miR-200-ZEB1-BCL6 pathway might identify a molecular marker that can be used to guide clinical managements of gastric diffuse large B-cell lymphomas.