Clinical and endoscopic presentation of primary gastric lymphoma: a multicentre study


Dr A. Zullo, Ospedale Nuovo Regina Margherita, Gastroenterologia ed Endoscopia Digestiva, Via Morosini 30, 00153 Roma, Italy.




Although the stomach is the most frequent site of intestinal lymphomas, few data are available on both clinical endoscopic presentation of gastric lymphoma and possible differences between low-grade and high-grade lymphomas.


Clinical, histological and endoscopic records of consecutive patients with primary low-grade or high-grade lymphoma diagnosed were retrieved. Symptoms were categorized as ‘alarm’ or ‘not alarm’. The endoscopic findings were classified as ‘normal’ or ‘abnormal’.


Overall, 144 patients with primary gastric lymphoma were detected, including 74 low-grade and 70 high-grade lymphoma. Alarm symptoms, particularly persistent vomiting and weight loss, were more frequently present in patients with high-grade lymphoma than in those with low-grade lymphoma (54% vs. 28%; P = 0.002). Low-grade lymphomas presented as ‘normal’ appearing mucosa (20% vs. 0%; P = 0.0004) or petechial haemorrhage in the fundus (9% vs. 0%; P = 0.02) more frequently than high-grade lymphomas, being also more often confined to the antrum (47% vs. 27%, P = 0.03) and associated with Helicobacter pylori infection (88% vs. 52%, P < 0.0001). On the contrary, high-grade lymphomas presented more commonly as ulcerative type (70% vs. 52%; P = 0.03), being also more frequently diagnosed in stage >I when compared with low-grade lymphomas (70% vs. 21%, P < 0.0001).


The overall prevalence of alarm symptoms is quite low and may be absent in more than 70% of patients with low-grade lymphoma.


The gastrointestinal tract is the most frequent site of extranodal non-Hodgkin's lymphomas and the stomach is the most common location accounting for more than 75% of such tumours.1–3 Although primary gastric lymphoma remains a rare disease, its frequency has been rising in the last decades.4 An incidence as high as 13.2 cases per 100 000 per year has been estimated in north-eastern Italy, which is significantly higher when compared with that of other European countries.5

The clinical presentation of gastric lymphoma is often vague, with dyspeptic symptoms such as epigastric pain or discomfort centred in the upper abdomen1–10 as the only clinical manifestation. Similarly, a variety of non-specific endoscopic patterns of gastric lymphoma have been described.10–13, 13–19 Although such neoplasia may appear at endoscopy as a clear malignancy (multiple or giant ulcers, vegetant mass, etc.), it tends to be characterized in a large number of cases by erosions, small nodules, thickening of gastric folds, or simple redness of gastric mucosa, generally suggesting a benign condition.7

In the last decade, novel therapeutic approaches for the management of patients with gastric lymphoma have been pioneered. The natural history of low-grade, B-cell, mucosa-associated lymphoid tissue (MALT) lymphoma (LG lymphoma) has been changed by the discovery of its association with Helicobacter pylori infection. A total regression of this lymphoma occurs following H. pylori eradication in a large number of patients in which the neoplasia has been diagnosed in an early stage.20 More effective therapeutic regimens have also been introduced for both LG lymphomas resistant to anti-H. pylori therapy and high-grade lymphomas (HG lymphoma).1, 2, 21 Nonetheless, the diagnosis of gastric lymphoma is too often performed in an advanced stage, undermining the possibility of successful management.

Based on these observations, we designed this retrospective, multicentre study in order to further evaluate both clinical and endoscopic features of patients with primary gastric lymphoma. Moreover, we attempted to detect the differences in the clinical endoscopic presentation between LG and HG lymphoma.

Materials and methods

Study design

This was a retrospective, multicentre, Italian study involving five Hospitals (one north, two centre, two southern). Clinical, histological and endoscopic records of consecutive patients diagnosed with LG or HG lymphoma between 1993 and 2004 were retrieved and accurately evaluated.

The main symptom for which the upper endoscopy was performed has been recorded for statistical analysis. Symptoms were categorized as ‘alarm’ (anaemia/melaena/haemorrhage, persistent vomiting, weight loss) or ‘not alarm’ (epigastric/abdominal pain, dyspepsia/bloating, heartburn), according to the current European guidelines.22Anaemia was considered present when haemoglobin values <13.4 g/dL in male and <12.3 g/dL in female patients were found.23

For histological assessment, the De Jong's classification was used, and all the cases diagnosed before 1997 were reclassified in each centre according to such a classification.24 For statistical analysis, A and B classes were considered as LG lymphoma, and C and D as HG lymphoma. Each biopsy was immunohistochemically investigated by staining for CD20 and CD3. Neoplasia staging was performed according to Lugano classification.25 A complete physical examination including Waldeyer's ring, routine laboratory tests, chests radiographs, computerized tomography of the abdomen and pelvis was performed in all patients as well as a bone marrow biopsy. The presence of H. pylori on gastric mucosa at histological assessment was registered.

The main endoscopic finding and the site of the lesions were recorded for each patient. The macroscopic pattern was classified as: (i) ulcerative type (single or multiple ulcerations or multiple erosions); (ii) exophytic type (tumour-like appearance with an irregular or polypoid mass); and (iii) hypertrophic type (large or giant folds, nodular pattern), as suggested elsewhere.7 In addition, we considered a gastric mucosal petechial haemorrhage pattern characterized by the presence of a few or several petechial haemorrhages.26

Statistical analysis

Statistical comparisons were performed by using Student's t-test for unpaired data, chi-square test, as appropriate. A P-value of <0.05 were considered significant.


A total of 157 cases of gastric lymphoma were identified during the study period. In 10 patients, the stomach was a secondary localization of a diffuse lymphoproliferative disease, and these cases were not included. Three further patients were excluded because the clinical records were incomplete and/or the histological material was unavailable. Consequently, the final study population consisted of 144 patients with primary gastric lymphoma. Of them, 29 cases were diagnosed in northern, 32 in central and 82 in southern Italy. The male to female ratio was 1.4:1, with a value of 1.6:1 and 1.3:1 in LG and HG lymphomas respectively. The neoplasia was classified as stage I in 79 (54%) patients, as stage II in 59 (41%), and as stage IV in three (2%) cases, whilst staging was not available in the remaining three patients (all with HG lymphoma).

Overall, 59 (41%) patients complained of alarm symptoms at diagnosis, whilst ‘not alarm’ symptoms were present in the remaining 85 (59%) cases. Among the alarm symptoms, haemorrhage (haematemesis/melaena), vomiting and weight loss had a similar occurrence, being present in 19 (32%), 16 (27%), and 16 (27%) patients respectively. Anaemia was the main symptom in seven (11%) patients, whilst lymphoma presented with an ulcer perforation in one (1%) case.

The main endoscopic feature was an ulcerative type in 87 (60%) cases (single ulcer: 63; multiple ulcer: 19; multiple erosions: five), a hypertrophic type in 19 (13%) patients (nodules: 12; thickening of gastric folds: seven), an exophitic type in 15 (10%) cases (irregular mass: 14; polyp: one), whilst a gastric petechial haemorrhage on the fundic mucosa was found in seven (5%) patients. No macroscopic lesion was detected in the remaining 15 (10%) cases. At endoscopic examination, lymphoma involved the antrum in 54 (37%) patients, gastric body/fundus in 63 (44%) and a diffuse involvement was detected in the remaining 26 (18%) cases.

At histological examination, 74 cases of LG and 70 of HG lymphoma were identified. As shown in Table 1, the overall prevalence of alarm symptoms was significantly higher in HG when compared with LG lymphoma patients (54% vs. 28%; P = 0.002). Both weight loss and persistent vomiting were more frequent in HG than in LG lymphomas (P = 0.046; Table 2). No correlation between endoscopic findings and symptoms was found. As shown in Table 3, alarm symptoms did not correlate with any endoscopic features in both LG and HG lymphoma cases. When considering only patients with single or multiple gastric ulcers, the prevalence of overt gastrointestinal bleeding (haematemesis/melaena) was similar between LG and HG lymphoma patients (15% vs. 14%; P = 0.8). At diagnosis, HG lymphomas were detected in a stage >I more frequently than LG lymphomas (70% vs. 21%, P < 0.0001). At bone marrow biopsy, a neoplastic involvement was present in one (1%) case with LG and in two (2%) cases with HG lymphoma.

Table 1.  Comparison between LG and HG lymphoma patients
 LG lymphoma (N = 74)HG lymphoma (N = 70)P-value
  1. * In three cases of HG lymphoma data were unavailable.

  2. † In two cases with LG and in 19 cases with HG lymphoma data were missing.

  3. LG, low grade; HG, high grade.

Age (mean ±  s.d.; years)59.4 ± 13.359.7 ± 15.1N.S.
 Not alarm5432
Endoscopic finding
 Ulcerative type39490.03
 Hypertrophic type910N.S.
 Exophitic type411N.S.
 Petechial haemorrhage70.02
 Normal mucosa150.0004
Gastric site
 Gastric body/fundus2935N.S.
Helicobacter pylori
Table 2.  The main symptom of presentation in both LG and HG lymphomas
 LG lymphoma, n (%; N = 74)HG lymphoma, n (%; N = 70)P-value
  1. LG, low grade; HG, high grade.

Alarm symptoms21 (28.4)38 (54.3)0.002
 Weight loss4 (5.4)12 (17.1)0.046
 Vomiting4 (5.4)12 (17.1)0.046
 Haematemesis/ melaena10 (13.6)9 (12.9)N.S.
 Anaemia3 (4)4 (5.8)N.S.
 Perforation1 (1.4)
Not alarm symptoms53 (71.6)32 (45.7)0.002
 Epigastric/ abdominal pain39 (52.7)26 (37.1)N.S.
 Dyspepsia/bloating9 (12.2)4 (5.8)N.S.
 Heartburn5 (6.7)2 (2.8)N.S.
Table 3.  Correlation between endoscopic findings and symptoms
Endoscopy finding Alarm symptoms (N = 59)Not alarms symptoms (N = 85)P-value
  1. LG, low grade; HG, high grade.

Ulcerative type4048N.S.
Hypertrophic type911N.S.
Exophitic type77N.S.
Petechial haemorrhage25N.S.
Normal mucosa150.002

Low-grade lymphoma was detected in seven (9%) patients with only gastric petechial haemorrhage on the fundic mucosa and in 15 (20%) with normal appearing gastric mucosa, whilst these findings were never encountered in patients with HG lymphoma (P = 0.02 and P = 0.0004; respectively), in which, on the contrary, ulcerative lesions were more frequently observed (70% vs. 52%; P = 0.03). The main endoscopic lesion of LG lymphomas was more frequently confined to the antrum (47% vs. 27%, P = 0.03), and the prevalence of H. pylori infection was significantly higher when compared with HG lymphomas (88% vs. 52%, P < 0.0001).


The stomach is the most frequent site of extranodal non-Hodgkin's lymphomas1–3 and a higher prevalence of such neoplasia has been observed in north-eastern Italy when compared with other countries.5 In the present study, we retrospectively collected data of all the patients referred to five participating centres scattered throughout Italy in the last 12 years which had been diagnosed with a primary gastric lymphoma. Our study confirms that alarm symptoms may be absent in more than half of patients with primary gastric lymphoma, in agreement with previous studies. Such symptoms were absent in >45% of patients with HG lymphoma, whilst their prevalence was even lower in LG lymphoma (27%), confirming similar data reported in other studies.14 As it is of paramount importance to detect gastric lymphoma at an early stage when treatment may be more effective and less toxic, all these observations strongly suggest the need to improve the diagnostic strategies. For instance, the absence of alarm symptoms in a large number of these patients further supports the necessity to promptly investigate with an endoscopic examination patients aged >45 years with dyspeptic symptoms, as suggested in the current European guidelines.22

As far as staging is concerned, our data found that HG lymphoma is diagnosed in an advanced stage (>I) in as many as two in every three cases, differently from the results of a previous study performed in the Netherlands, in which HG lymphoma was diagnosed in an advanced stage in <30% of cases.14 Of note, even LG lymphoma was disappointingly diagnosed in an advanced stage in more than 20% of cases. The bone marrow involvement in patients with LG lymphoma has been reported to widely ranging from 0% to 15%.1 In the present study, we observed a rate of 1%, which is in agreement with the results of a large study performed in Germany (five of 116; 4%).3 As bone marrow involvement corresponds to stage IV, which requires a more aggressive therapeutic approach, our data further strengthen the need of bone biopsy in all LG lymphoma patients.

As far as the endoscopic presentation is concerned, in agreement with another study,27 we observed that in our series LG lymphoma was confined to the antrum in nearly half of the cases. This observation seems to be different from previous studies performed in South Korea and in Japan, where only 9–20% of MALT lymphomas were confined to the antrum.16, 18 Interestingly, we found that nearly 9% of LG lymphomas (all stage I) may appear at endoscopy as gastric petechial haemorrhages on the fundic mucosa. In a previous study performed in Japan, such an endoscopic finding was observed in 3% of primary gastric lymphomas.12 This endoscopic picture has also been previously reported in non-steroidal anti-inflammatory drugs (NSAIDs) users26 and in Henoch-Schönlein purpura,28 but none of these conditions occurred in our patients. Therefore, we suggest that adequate gastric biopsies should be performed in all patients with petechial haemorrhages on gastric fundus, particularly if patients were not currently taking NSAIDs. Unlike the HG lymphoma, we also observed that LG lymphoma might be present in patients without alarm symptom in the absence of macroscopic alterations of gastric mucosa. Indeed, 20% of LG lymphoma patients (all stage I) showed an apparently normal mucosa, and the neoplasia was revealed by the routine histological examination. This relevant finding is in agreement with the results of a previous Italian study and others in which 9–30% of gastric lymphomas were detected on macroscopically normal mucosa.2, 11, 16, 29 Therefore, as previously suggested,30 it is likely that LG lymphoma in such cases has been diagnosed in a very early phase, before an endoscopic lesion appeared. It could be speculated that this finding depends on the increased propensity to perform gastric biopsies during diagnostic upper endoscopy on normal appearing mucosa, in order to search for H. pylori infection in dyspeptic patients, according to current guidelines.22 Such an observation further strengthens the value of performing routine gastric biopsies during endoscopic examinations, even on normal appearing mucosa. Finally, our data clearly found that LG lymphoma are strongly associated with H. pylori, and the 89% infection rate observed in our series is in deep agreement with previously reported observations.20

In conclusion, the overall prevalence of alarm symptoms in primary gastric lymphoma is quite low, and they may be absent in nearly 75% of LG lymphoma patients. Moreover, contrary to HG, LG lymphoma may be characterized by a normal endoscopic picture and it is more frequently associated with H. pylori infection. At diagnosis, HG lymphoma is more often detected in an advanced stage when compared with LG lymphoma.


No external funding was received for this study.