First-line antibiotic therapy in Helicobacter pylori-negative low-grade gastric mucosa-associated lymphoid tissue lymphoma

First-line antibiotic treatment for eradicating Helicobacter pylori (HP) infection is effective in HP-positive low-grade gastric mucosa-associated lymphoid tissue lymphoma (MALToma), but its role in HP-negative cases is uncertain. In this exploratory retrospective study, we assessed the outcome and potential predictive biomarkers for 25 patients with HP-negative localized gastric MALToma who received first-line HP eradication (HPE) therapy. An HP-negative status was defined as negative results on histology, rapid urease test, 13C urea breath test, and serology. We observed an antibiotic response (complete remission [CR], number = 8; partial remission, number = 1) in 9 (36.0%) out of 25 patients. A t(11;18)(q21;q21) translocation was detected in 7 (43.8%) of 16 antibiotic-unresponsive cases, but in none of the 9 antibiotic-responsive cases (P = 0.027). Nuclear BCL10 expression was significantly higher in antibiotic-unresponsive tumors than in antibiotic-responsive tumors (14/16 [87.5%] vs. 1/9 [11.1%]; P = 0.001). Nuclear NF-κB expression was also significantly higher in antibiotic-unresponsive tumors than in antibiotic-responsive tumors (12/16 [75.0%] vs. 1/9 [11.1%]; P = 0.004). A substantial portion of patients with HP-negative gastric MALToma responded to first-line HPE. In addition to t(11;18)(q21;q21), BCL10 and NF-κB are useful immunohistochemical biomarkers to predict antibiotic-unresponsive status in this group of tumors.

Epidemiologic studies have shown that the presence of cytotoxin-associated gene A (CagA) protein, the most important HP virulence factor, is associated with the formation of lymphoid follicles and MALT lymphoma of the stomach 28,29 . Previous studies reported that the CagA-seropositive rate in patients with gastric MALT lymphoma ranged from 89% to 96% 30,31 .
Lehours et al. reported detection of the CagA gene in 47.4% of the HP strains obtained from 90 cases of gastric MALT lymphoma 32 . Among t(11;18)(q21;q21)-negative gastric MALT lymphoma cases, Sumida et al. found that titers of anti-CagA were significantly higher in HP-dependent cases than in HP-independent cases 33 . We recently found that 11 HP strains isolated from patients with HP-dependent gastric lymphomas (5 gastric diffuse large B-cell lymphomas with histologic evidence of MALT and 6 gastric MALT lymphomas) were CagA positive 34 . We and other investigators demonstrated that CagA can promote cellular proliferation and attenuate apoptosis of B-cells through activation of CagA-signaling such as SRC homology-2 domain-containing phosphatase (SHP2) and extracellular signal-regulated kinase (ERK)-related signaling, or BAD phosphorylation and p53 accumulation [35][36][37][38] . Furthermore, we reported that HP CagA protein and its signaling pathway proteins, such as phospho (p)-SHP2, p-ERK, p-38 mitogen-activated protein kinase (MAPK), BCL-2, and BCL-XL, can be detected in tumors of gastric MALT lymphoma 39,40 . The expression of CagA and CagA-signaling molecules is closely associated with HP-dependence of these tumors 40 , indicating CagA may serve as a marker for the presence of HP for gastric MALT lymphoma.
In this study, we assessed the response rate and the long-term disease-free status of patients with localized HP-negative gastric MALT lymphoma (all negative for histology [including HP, atrophic gastritis, and intestinal metaplasia], rapid urease test, 13 C urea breath test, and serology as well as for CagA expression in tumor cells and gastric microenvironments) who received first-line HPE regimens consisting of PPIs plus clarithromycin and amoxicillin. We also investigated the association between potential biomarkers, including t(11;18)(q21;q21), nuclear BCL10 expression, and nuclear NF-κB expression, and antibiotic-unresponsive status of the same type of tumors.

Results
Clinicopathological features and tumor response to HP eradication therapy. Between January 1, 2005, and June 30, 2014, 25 patients with newly diagnosed stage IE/IIE1 primary HP-negative (results of the histology [including HP, atrophic gastritis, and intestinal metaplasia], rapid urease test, 13 C urea breath test, and serology were all negative) gastric MALT lymphoma who received HPE as first-line treatment were included. Among them, 18 cases were also negative for HP cultures. Furthermore, CagA expression was not detected in tumor cells of all patients, indicating that HP is not present in these 25 cases (Fig. 1). We also showed that there was no CagA gene detected in gastric tumor biopsies obtained from patients with antibiotic-responsive tumors (Supplementary method, data not shown). Among these, 22 (88.0%) were at stage IE and three (12.0%) were at stage IIE1 (Table 1). Regarding the underlying diseases, four patients had hepatitis B virus infection, one patient had a hepatitis C virus infection, and one patient had an autoimmune disease (Sicca syndrome). Twenty-one (84.0%) of the 25 patients were treated with amoxicillin, clarithromycin, and omeprazole, whereas 4 (16.0%) patients received amoxicillin, clarithromycin, and lansoprazole. The histological scoring system proposed by the Groupe d'Etude des Lymphomes de l' Adult (GELA) is currently recommended to improve the consistency between the findings of different studies regarding first-line HPE for gastric MALT lymphoma 41 . The European Gastro-Intestinal Lymphoma Study (EGILS) consensus and the International Extranodal Lymphoma Study Group (IELSG) study therefore recommend the routine use of the GELA criteria in evaluating the response to treatment, including HPE, of gastric MALT lymphoma 42,43 . A complete remission (CR) is defined by the GELA grading system as the total disappearance of gross lymphoma and a negative histologic finding (CR or probable minimal residual disease [pMRD]), whereas partial remission (PR) is defined as normalization or reduction of macroscopic findings, histologic signs of lymphoma regression, and no signs of progression 42,43 . Previously, Fishbach et al. reported that 32 (32%) of 101 patients with pMRD or PR of tumors (according to the GELA criteria) after successful HPE achieved a histologic CR and 62% of patients had stable disease during the second-year of follow-up 44 . Another international randomized LY03 trial showed that the addition of an alkylating agent, chlorambucil, did not result in a better recurrence/progression-free survival or overall survival for patients with gastric MALT lymphoma who responded to HPE (including a CR or PR) compared with those who received "watch and wait" 45 .
Therefore, based on the GELA criteria for evaluating responses of gastric MALT lymphomas to HPE, we categorized our patients into two subgroups, those having antibiotic-responsive tumors (including CR and PR) and those having antibiotic-unresponsive tumors (including stable disease [SD] and progressive disease [PD]). The clinicopathological features of 9 patients (CR, number = 8; PR, number = 1) with antibiotic-responsive tumors ( Fig. 2a-d) and 16 patients with antibiotic-unresponsive tumors, and the responses of their tumors to HPE are summarized in Table 1. We observed antibiotic responses in 9 (36.0%; 95% confidence interval [CI], 17.2-54.8%) out of 25 patients, and the median time to a CR (number = 8) was 7 months (95% CI, 0.1-13.9 months) (Fig. 3a). The antibiotic-responsive rate for 22 patients with stage IE was 40.9% (9/22), whereas the antibiotic-responsive rate for three patients with stage IIE1 was 0%. Of 20 patients receiving serum immunofixation electrophoresis (IFE) assessments, immunoglobulin M lambda monoclonal gammopathy was detected in 3 (25.0%) of the 12 antibiotic-unresponsive tumors, but not in the 8 antibiotic-responsive tumors (P = 0.242, Table 1).
Among the 16 cases without a tumor CR or PR, 14 patients who had persistent or increasing epigastric discomfort and were endoscopically or pathologically documented to have progressive tumors during the regular follow-up period were administered salvage treatments including oral alkylating agents (chlorambucil), rituximab-based regimens, and radiotherapy ( Table 1). The time to aforementioned rescue therapy following HPE failure in these 14 patients is listed in Table 1. Of the 14 patients receiving second-line treatments, 12 patients achieved a CR while two patients had SD. Notably, two patients with SD undergoing observation alone experienced no progression during the follow-up (18 to 51 months after treatment) ( Table 1).
At a median follow-up of 51.0 months (95% CI, 34.7-67.3 months), all patients with responsive tumors after HPE therapy were alive and free of lymphomas and progression, whereas 2 (16.7%) out of 12 patients with responsive tumors after second-line therapy experienced relapse (one in the parotid gland and the other in the stomach) ( Table 1).

Discussion
In this study, we demonstrated that nine (36.0%) out of 25 patients with HP-negative gastric MALT lymphoma were responsive to HPE, and remained lymphoma-free and progression-free at the longest follow-up. Our findings are consistent with a systematic review of published articles that demonstrated that a first-line HPE regimen resulted in a CR rate of 15.5% in 110 patients with HP-negative gastric MALT lymphoma 10 . As an addition to Zullo et al. 10 , who analyzed the CR rate after first-line antibiotic treatment, the diagnostic methods for HP, and the administration of HP regimens, we assessed the time to CR, the potential markers, including clinical stage, t(11;18)(q21;q21), and BCL10 expression, in 22 published results from 1999 through 2016 (summarized in Table 3) [5][6][7]9,15,16,25,26,[46][47][48][49][50][51][52][53][54][55][56][57][58] . Overall, including our report, the CR rate after completing HP eradication treatments was observed in 68 (27.9%) out of 244 patients. The most commonly used HPE regimens consisted of PPIs plus at least two antibiotics such as amoxicillin, clarithromycin, or metronidazole for 7 to 14 days. Examinations for the presence of HP were mostly based on positive results from histology, rapid urease tests, 13 C urea breath tests, and serology, and culture and stool antigen tests had been evaluated in five studies 7,16,25,47,53 , including our study.
Although the aforementioned published series (some series comprising less cases and some series comprising more cases than our present cases) demonstrating a CR rate of 27.4% (60/219) (Table 3), our current data underline the reality that a proportion of patients without evidence of HP infection can be cured by first-line HPE. First, we showed that none of our patients had histologic evidence of atrophic gastritis or intestinal metaplasia (the aforementioned histomorphological findings are clues of a previous HP infection) in their specimens before HPE 15,16 , even if they had undergone previous eradication therapy or antibiotics treatment. Second, we showed that none of our patients exhibited CagA expression in tumor cells or in the gastric microenvironment. We also showed that there was no CagA gene detected in gastric tumor biopsies obtained from patients with antibiotic-responsive tumors. These findings indicated that the CagA-negativity of the tumors of our cases is actually just another suggestion that it is a real HP-negative gastric MALT lymphoma. Third, we demonstrated that the time to response for patients with antibiotic-responsive tumors was 7 months (range: 1 to 24 months), and importantly, after the median long-term follow up of 51 months, all patients with responsive tumors were free of lymphoma or progression. Combining our results with that of five other investigators (Table 3) 9,16,55,57,58 , the median time to a CR for patients with HP-negative gastric MALT lymphoma who received first-line HPE was 6.1 months (95% CI, 4.7-7.5 months) (Fig. 3b).
In the current study, we reported that 29 (31.5%) of 93 patients with gastric MALT lymphoma were HP-negative. However, in a systematic review of gastric MALT lymphoma (including diffuse large B-cell In addition to PPIs, the most commonly used antibiotics in the first-line treatment of HP in gastric MALT lymphoma include amoxicillin, clarithromycin, and metronidazole. In the present study, all patients received PPI plus clarithromycin and amoxicillin. It should be noted that increasing antimicrobial resistance of HP, especially for clarithromycin, has been observed in patients with HP infections worldwide [62][63][64] . In Europe, the prevalence of clarithromycin resistance by HP ranges from 5.6% to 36.6% 65 . In a nationwide study of primary resistance to HP in Taiwan after implementation of a national policy to restrict antibiotic consumption since 2001, Liu et al. showed that the prevalence of primary resistance to clarithromycin was 11.2% (95% CI 9.6-13%) 66 . In a similar time period, in the Asia-Pacific region, the prevalence of primary resistance to clarithromycin in China (2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009) 67 , Japan (2000-2013) 68 , and Korea (2009-2012) 69 was 23.8% (69/290), 31.1% (334/1073), and 23.7% (27/114), respectively. Nevertheless, the prevalence of antimicrobial resistance of HP to clarithromycin in gastric MALT lymphoma is rarely studied because the sensitivity of HP culture is obviously lower than the sensitivity to histologic detection of HP in gastric MALT lymphoma 32 . Recently, Bilgilier et al. found that the rate of clarithromycin resistance in 13 cases of gastric MALT lymphoma was 15% through analyses of the HP 23S rRNA gene containing genotypic clarithromycin resistance 70 . Although the question of why a certain proportion of HP-negative gastric MALT lymphomas may respond to antibiotics remains unanswered, several crucial findings may support the following speculations: (1) HPE regimens may also eradicate other bacteria or Helicobacter-like bacteria, such as H. heilmannii that is associated with the development of gastric MALT lymphoma in humans [71][72][73]   Another macrolide, erythromycin, was found to have an inhibitory effect on proliferation of T-cells, and a possible mechanism is the down-regulation of NF-κB expression 78 . In CD4+ T-cells, azithromycin effectively inhibited cell proliferation and cytokine secretion through down-regulation of the activity of mammalian target of rapamycin 79 . The aforementioned immunosuppressive effect on CD4+ T-cells was also observed at a higher concentration of clarithromycin (40 mg/L) 79 55 . Our study also revealed a CR rate of 32% in patients receiving a clarithromycin (500 mg twice a day for 14 days)-based regimen. Furthermore, one recently published phase II trial reported that high-dose clarithromycin (2 g a day for 14 days for each course) resulted in a CR rate of 26.9% in patients with relapsed or refractory extranodal MALT lymphoma 82 .
In this study, we showed that nuclear expression of BCL10 or NF-κB is closely associated with an antibiotic-unresponsive status and that both molecules are associated with the status of t(11;18)(q21;q21). These findings are consistent with those of Ye et al. 7 and Sumida et al. 33 who reported on the relationship between an antibiotic-unresponsive status and BCL10 nuclear expression in an HP-negative gastric MALT lymphoma.
As shown in Table 3, notably, two cases of HP-negative gastric MALT lymphomas with t(11;18)(q21;q21) remained antibiotic-responsive 49,52 . In a large series of HP-positive gastric MALT lymphomas, Liu et al. also demonstrated that 2 (4.5%) out of 44 patients with t(11;18)(q21;q21) remained HP dependent 24 . In our series of HP-positive gastric MALT lymphomas (data not shown), the pivotal role of BCL10 or NF-κB in HP-independent growth 27,50,86,87 was demonstrated by the finding that two cases with t(11;18)(q21;q21) but lacking both nuclear expression of BCL10 and NF-κB responded well to HPE (Supplementary Fig. 1) 87,88 . Of these two cases with t(11;18)(q21;q21)-positive but no nuclear NF-κB expressing tumors, one case harbored a fusion transcript of t(11;18)(q21;q21) that contained 3 intact BIR domains in the amino terminal API2 region, and an intact caspase-like domain, but none of the immunoglobulin-like domains in the carboxyl terminal MALT1 region. Previous studies showed that the fusion product of t(11;18)(q21;q21) comprising an intact immunoglobulin-like domain had a greater ability to stimulate NF-κB signaling than the fusion product without an intact immunoglobulin-like domain 89,90 . Several studies have demonstrated that t(11;18)(q21;q21)-mediated NF-κB activation requires an interaction between API2-MALT1 and TRAF2 or TRAF6 [83][84][85] . The lack of an immunoglobulin-like domain and the disruption of the interaction with TRAF2 or TRAF6 of the API2-MALT1 fusion protein may be linked to the absence of nuclear NF-κB expression in some t(11;18)(q21;q21)-positive tumors that remain antibiotic-responsive.
In summary, the results of this study indicate that a substantial proportion of patients with early-stage HP-negative gastric MALT lymphoma remain antibiotic-responsive and can be cured using a first-line HPE regimen. In addition to t(11;18)(q21;q21), nuclear expression of BCL10 or NF-κB can help us predict antibiotics' unresponsiveness. Further investigations into microbiota associated with the lymphomagenesis of HP-negative gastric MALT lymphoma are warranted.

Patients and Methods
Ethics statement. All experimental protocols were approved by the Institutional Review Board (IRB) of the Research Ethical Committee of National Taiwan University Hospital (NTUH IRB number: 9361700774). All experiments were conducted in accordance with the approved guidelines and regulations. The patients' medical data were anonymized prior to access and analysis. All patients provided written informed consent to participate in and to provide tissue material for biological studies.
Patients, treatment, and tissue samples. We screened study subjects from the Cancer Registry, Medical Information Management Office, and the lymphoma database of the Department of Pathology of the National Taiwan University Hospital in Taipei, Taiwan between January 1, 2005 and June 30, 2014. We identified 93 patients with stage IE/IIE1 gastric MALT lymphoma from patients diagnosed with primary gastric lymphoma. We retrospectively reviewed the medical records and pathologic records of these patients to evaluate whether these gastric MALT lymphomas were HP-positive or HP-negative tumors. Evidence of HP infection was defined as positive results on biopsy, histology, a urease test, a 13 C urea breath test, or serology [91][92][93] . An HP-negative status was defined as total negative results on histology (included HP, atrophic gastritis, and intestinal metaplasia) 15,16 , a rapid urease test, a 13 C urea breath test, and serology.
There were 63 patients with HP-positive tumors and 29 patients with HP-negative tumors. Since gastric MALT lymphoma is relatively indolent and pseudo-negative HP tests may occur, most HP-negative patients in our institution were treated with first-line HPE regimens, particularly if their symptoms were insignificant.
From complete medical records, among 29 patients with HP-negative gastric MALT lymphoma, two patients received radiotherapy, 2 patients received alkylating agents-based chemotherapy, and 25 patients received antibiotics as a first-line treatment. The antibiotics regimens were the same as the HPE regimen, which consisted of 500 mg of amoxicillin administered four times a day (or 1000 mg of amoxicillin administered twice a day), 500 mg of clarithromycin administered twice a day, and 20 mg of omeprazole or 30 mg lansoprazole administered twice a day for 2 weeks as first-line treatment.
Diagnosis of gastric MALT lymphoma was made according to the histological criteria described by Isaacson et al. and the European Gastro-Intestinal Lymphoma Study consensus report on gastric extranodal marginal zone B-cell MALT lymphoma 42,94 . The tumors were staged and classified according to the Musshoff modification of the Ann Arbor staging system. The patients also received an examination for the presence of monoclonal gammopathy using serum IFE. The patients underwent their first follow-up after an upper gastrointestinal endoscopic examination or an ultrasonic endoscopic examination 4 to 8 weeks following HPE. This examination was repeated every 12 to 16 weeks until we observed histological evidence of remission.
The regression of the tumor following HPE was histologically evaluated according to the criteria of the (GELA) histological scoring system 42,43 . A CR is defined by the GELA grading system as the total disappearance of gross lymphoma and a negative histologic finding (CR or pMRD), whereas PR is defined as normalization or reduction of macroscopic findings, histologic signs of lymphoma regression, and no signs of progression. Tumors that resolved to a CR or PR after HPE were considered antibiotic-responsive 42,43 . Two subgroups of patients were considered antibiotic-unresponsive: (1) those who had SD but failed to show histologic regression 24 months following HPE, and (2) those with tumors exhibiting objective evidence of PD at any time during the follow-up 42,43 . Multiplex reverse transcription polymerase chain reaction for the API2-MALT1 fusion transcript of t(11;18)(q21;q21) in lymphoma cells. Total cellular RNA was extracted from formalin-fixed and paraffin-embedded tissues using an Ambion RNA isolation kit (RecoverAll ™ Total Nucleic Acid Isolation Kit, Ambion ® | Life Technologies) and was analyzed for the API2-MALT1 fusion transcripts of the t(11;18) (q21;q21) translocation using multiplex reverse transcription polymerase chain reaction, followed by sequencing as described previously 27,95,96 . Gastric MALT lymphoma samples with API2-MALT1 fusion transcripts served as positive controls.
Immunohistochemistry. Immunohistochemistry for BCL10 (sc-9560; Santa Cruz Biotechnology, Santa Cruz, CA, USA), NF-κB (p65; sc-109; Santa Cruz Biotechnology), and CagA (A10; sc-28368, Santa Cruz Biotechnology) was performed on paraffin-embedded sections of pre-HPE endoscopic biopsies, using an indirect immunoperoxidase method according to the manufacturer's instructions 27,39,96 . Paraffin sections with the first, second, or both primary antibodies omitted were used as negative controls to verify the specificity of the staining.
The percentages of positive cells were averaged to yield an immunohistological score of 0-100%. The staining was considered positive for BCL10 and NF-κB (p65) if the protein was detected in more than 10% of the tumor cells; nuclear staining was performed according to the criteria described by Ye et al. 97 and Oshima et al. 98 . For the CagA maker, positive expression was defined as ≥10% of cells with moderate or strong immunostaining (tumor cells with readily appreciable brown staining distinctly marking the tumor cell nucleus or cytoplasm), as previously described 39,40 . Statistical analysis. In this study, the Fisher exact test was used to compare the clinical characteristics, the presence of t(11;18)(p21;q21) as well as the expression levels of BCL10 and NF-κB (p65) between the antibiotic-responsive and antibiotic-unresponsive cases. The analyses were conducted using follow-up data that became available on December 31, 2015. Differences between the results of the comparative tests were considered statistically significant if the two-sided P-value was <0.05.