48uep6bbphidcol2|ID

48uep6bbphidvals|1886

48uep6bbph|2000F98CTab_Articles|Fulltext

Stomach is the commonest site of involvement in extra-nodal non-Hodgkin’s lymphoma with incidence ranging between 15 to 20%.^1-3 Nevertheless, primary gastric lymphoma (PGL) is uncommon; accounting only for 2-8% of all tumors of the stomach.⁴ The recent reports suggest increase in incidence of primary gastric lymphoma.^5,6

Most of the gastric lymphomas are of non-Hodgkin type histologically and virtually all gastric lymphomas arise from B-cell. T-cell non-Hodgkin’s lymphoma of stomach is extremely rare accounting for approximately 6% of total cases.⁸ PGL is defined according to Dawson’s criteria that include (1) absence of peripheral lymphadenopathy at the time of presentation; (2) lack of enlarged mediastinal lymph nodes; (3) normal total and differential white blood cell count; (4) predominance of bowel lesion at the time of laparotomy with only lymph nodes obviously affected in the immediate vicinity; and (5) no lymphomatous involvement of liver and spleen.⁹

Gastric lymphomas are heterogeneous disease in terms of site of involvement, histological subtypes, and treatments offered. Bone marrow (BM) involvement by gastric MALT (mucosa-associated lymphoid tissue) lymphoma has been reported to be around 1-8% cases.^10,11 BM examination is recommended as initial staging work-up according to European Society for Medical Oncology (ESMO)¹² guidelines whereas National Comprehensive Cancer Network (NCCN)¹³ guidelines suggest BM examination in selected cases. In the European Gastro-Intestinal Lymphoma Study (EGILS) consensus report (2011)¹⁴, it was suggested that BM examination should only be recommended when tumor regression is not achieved with H. pylori eradication after an adequate time period. The incidence of BM involvement in gastric diffuse large B-cell lymphoma (DLBCL) is not well characterized.

Primary gastric lymphomas have been scarcely reported in India.^15,16 In this study we present a retrospective study of 120 cases of PGL from a single centre in northern India to evaluate the following - 1) diagnostic yield in endoscopic mucosal biopsy for PGL, 2) histological subtypes and associated histological features in patients with PGL, and 3) frequency of BM involvement by PGL.

All the gastric biopsies and resected specimens during last 14 years (2001-2014) were retrieved from the hospital records and those with a diagnosis of gastric lymphoma were reviewed. Immunostaining with lymphoma panel (LCA, CK, CD20, CD3, cyclin D1, CD5, CD23, and BCL2) was performed routinely depending on availability of tissue. All the cases were reclassified according to the World Health Organization (WHO) 2016 classification17 based on morphological and immunophenotypic criteria.

Baseline demographic profile, endoscopic findings and rapid urease test (RUT) results were obtained from case files. Bone marrow aspiration and biopsy slides were evaluated to look for marrow involvement. The endoscopic biopsy finding was compared with resection specimens wherever available.

A total of 7458 gastric specimens (biopsies / resections) were received in a 14-year period. There were 120 (1.6%) cases of primary gastric lymphomas (PGL), 400 (5.4%) cases of gastric carcinoma and 21 (0.3%) cases of primary gastric neuroendocrine tumor diagnosed during this period. There was increasing trend of PGL every year. Thirty-nine of 3006 biopsies (1.3%) were diagnosed from 2001 to 2007 whereas 81 of 4452 biopsies (1.8%) were diagnosed from 2008 to 2014. In 120 cases of gastric specimens diagnosed as lymphoma, 77 cases (64%) had only biopsy, 11 cases had only resection specimen whereas 32 cases had both biopsy and resection specimens. 

The age range in 120 cases of gastric lymphoma was 17-87 years (mean 52.5 years, median 51 years). Seventy-seven patients (65%) were more than 50 years of age with a peak incidence in 6th decade. Only one case was diagnosed in a child aged 17 years with histological diagnosis of Burkit lymphoma. There were 90 males and 30 females with M: F ratio of 3:1. The clinical presentation was nonspecific and variable (table 1). The anatomic site of lesions is summarized in table 2. Endoscopy data was available in 97 cases. The endoscopic appearance was proliferative in 30 cases, ulcerative in 25 (figure 1A), ulceroproliferative in 30 cases, diffuse wall thickening in 5 cases and polypoidal in 6 cases. Complications like gastric diverticulum were seen in 3 cases and hiatus hernia was seen in one case.

The frequencies of different histological subtypes of PGL are listed in table 3. Immunohistochemistry (IHC) was available in 84 patients. Among these 84 cases all cases except one were B cell lymphoma (figure 1B to 1F). Diffuse large B-cell lymphoma (DLBCL) was the commonest PGL subtype accounting for 62.5% followed by maltoma (18.3%). Ten cases (13.3%) of DLBCL had low-grade maltoma component along with high-grade areas. Helicobacter pylori were seen in 22 (18.3%) cases of PGL as documented by biopsy. RUT was conducted in 54 cases of which 23 patients (42.6%) were tested positive and 31 cases (57.4%) were negative. H. pylori were seen on biopsy in 12/65 cases of DLBCL, 5/12 cases of maltoma and in 5/29 cases of lymphoma NOS. 

BM examination (both aspiration and biopsy) was done in 79 cases (including 50 cases of DLBCL). Marrow involvement by lymphoma was not observed in any of the cases. 

A total of 43 resection specimens were received which included 22 cases of distal partial gastrectomy, 1 case of proximal gastrectomy, 5 cases of subtotal (distal) gastrectomy, 12 cases of total gastrectomy and 4 cases of radical gastrectomy. Grossly ulceroproliferative growth was identified in 24 (55.8%) cases, ulcerative in 14 (32.5%) cases and nodular or diffuse thickening in 5 (11.6%) cases with ulceration. Multifocal lesions were identified in 3 of the specimens. Proximal esophageal involvement was seen in one specimen and distal duodenal infiltration was noted in six specimens. Lymph nodes ranging from 0 to 30 (mean 10.2) were dissected in all resected cases of which nodal involvement was seen in 36 (83.7%) cases. Gross infiltration into the liver parenchyma was seen in two cases and one case showed pancreatic as well as splenic involvement.

On microscopic examination of resection specimens, 42 show a diffusely infiltrating growth. No residual tumor was identified in one case (post-chemotherapy). All most all cases showed focal / complete surface ulceration. Lymphoepithelial lesions were identified in 19 (45.2%) cases. In all resected cases tumor was infiltrating the muscularis propria of gastric wall. In 14 patients tumor was confined within the muscularis whereas in 13 cases sub-serosal involvement was also seen and in another 14 cases serosa was also involved. In 20 cases one or more perigastric nodes were involved (with focal / complete effacement of architecture) with perinodal extension of atypical lymphoid cells in three cases.  

Both biopsy and resection specimens were available in 32 patients. Endoscopic biopsies disclosed malignant lesions in 27 cases (84%) and were diagnostic of lymphoma in 13 (48%). High suspicion of gastric lymphoma was raised in ten cases (table 4). In 5 cases biopsy was either inadequate or non-representative or showed extensive crushing artifact making it unsuitable for final diagnosis. We also reviewed the biopsy and resection findings of 63 cases of gastric adenocarcinoma received in 5 years duration, of which two cases were misdiagnosed as lymphoma whereas 61 cases were truly reported as adenocarcinoma on endoscopic biopsy. Thus the sensitivity and specificity of endoscopic biopsy for diagnosing lymphoma in this study was 85% and 97% respectively.

Some of the cases showed other associated findings. One resection specimen showed presence of gastric carcinoid distal to the lymphoma lesion. In another specimen, nodular congophilic deposits were identified within the lesion suggesting gastric amyloidosis. Multiple submucosal granulomas were identified in another one case and candidial colonization was observed in one case.  Foreign body type giant cell reaction was observed in three cases due to previous biopsy. In another one case multiple granulomas with central caseous necrosis (s/o tuberculosis) were identified in perigastric nodes. Three patients had ascites and were subjected to cytologic fluid analysis. All three patients had lymphomatous infiltration of ascitic fluid. 

The gastrointestinal tract is the most frequent site of extranodal lymphoma, and the stomach is involved in up to two-thirds of this cases.¹⁸ Although a rare disease, the frequency of PGL is rising in the last decades.⁶ Only a few studies on primary gastrointestinal lymphoma have been published from India. In this large retrospective study, 1.6% of the cases were PGL of the 7400 biopsies whereas the incidence of gastric carcinoma was 5.4%. 

Nonspecific endoscopy patterns have been described for gastric lymphoma. The endoscopic diagnosis of gastric lymphoma at first sight may be difficult, because of the various macroscopic patterns; ranging from benign gastritis to carcinoma.¹⁹ In this study, we analyzed different endoscopic patterns of gastric lymphoma. Although large stellate ulcers are characteristic for NHL, pure proliferative and ulceroproliferative pattern of involvement was seen in more than 50% cases in the present study. Six cases in our series had polypoid mass lesion and five other cases had linitis plastic like diffuse thickening on endoscopy suggestive of carcinoma. 

Diffuse large B cell lymphoma (with or without a MALT component) is the most common histological subtype in our series. Sixty percent of cases in the present study were also DLBCL, which is comparable to other Indian large studies¹⁵ and contrasting to western studies.7 Such high number of cases of DLBCL in Indian population can be explained by late stage of presentation. 

A correct histological diagnosis after the first endoscopic biopsy was made in 76% (67/88 biopsies) of the cases which was similar to other studies.^20,21 Lymphoma was misdiagnosed in 4 cases as adenocarcinoma and 5 cases had non-representative biopsy, however a histological suspicion of lymphoma was made in another 10 cases and a repeat biopsy was advised. In a study by Xu et al²² repeat biopsy could diagnose >90% of cases. In the study by Arista-Nasr et al²³ a review of the histological sections indicated that the changes, either diagnostic of or compatible with lymphoma, were present in 75% and 18% of the sections, respectively. They also suggested few histological features indicative of gastric lymphoma such as (1) marked increase in density of the lymphoid infiltrate in the gastric mucosa, (2) replacement of gastric glands by lymphoid infiltration, and (3) lymphoepithelial lesions. We obtained a high sensitivity and specificity of endoscopic biopsy for the diagnosis of PGL. 

A total of 106 resection specimens (including both NHL and adenocarcinoma) were reviewed in this study. False positive reporting of lymphoma was done in two patients which turned out to be poorly differentiated adenocarcinoma on resection. Therefore a high index of suspicion for lymphoma should be raised on endoscopic biopsies with diffuse pattern of tumor cells and especially if there is crushing artifact on small biopsies. Immunohistochemistry should always be used for confirmation in such type of cases. 

Currently BM examination is mandatory for initial staging of gastric lymphomas. A few previous studies on gastric MALT lymphoma showed rare involvement of BM.  In a recent study by Park et al suggested that BM involvement in gastric MALTOMA is only 1%.²⁴ This incidence rate is as low as 0.5% in another study.²⁵ None of the cases (including the cases with stage IV disease), in our study had secondary BM involvement. There have been different opinions about the requirement of BM examination in initial staging work-up of gastric MALT lymphoma. According to the European Gastro-Intestinal Lymphoma Study (EGILS) consensus report (2011), BM examination should only be recommended when tumor regression is not achieved with H. pylori eradication after an adequate time period. 

The role of H. pylori in the pathogenesis of PGL has been established. Corresponding to this we had nearly half of the cases showing positivity for RUT or presence of H.pylori on histology. DLBCL was the commonest histological subtype (n=65; 62.5%) in our study, followed by low-grade marginal zone lymphoma. This is in accordance to previous studies from India as well as outside India. A low-grade component of MALT lymphoma was seen in 6.76% (n=15) of the DLBCL, which is much lower than 33% reported by Koch et al²⁶ and higher than 6.7% reported by Arora et al.¹⁵

In a study by Roukos et al^{21, 28} surgically resected specimens of PGL, 9 cases had limited disease (within submucosa) whereas 19 patients had advanced disease. Forty-three resection specimens in the present study showed surface ulceration as a consistent finding and infiltration of muscularis (corresponding to Paris stage T2) in all cases. Interestingly one case had gastric carcinoid and another had gastric amyloidosis in addition to PGL. Therefore, the rate of advanced disease was higher in the present study. A high rate of diagnostic discordance (65.6%) between endoscopic biopsy and resection specimen (available in 32 cases) could be attributed to multiple factors like a non-representative biopsy, poorly preserved morphology and poorly differentiated carcinoma as mimicker of DLBCL. 

However, few limitations were encountered in the study such as the management and follow up data was available for only 40% of cases. Immunohistochemistry was available in 70% of cases which posed difficulty in subcategorizing the cases. None of the patients in present study had BM involvement. Therefore, identifying risk factors for BM involvement could not be studied. 

In conclusion, this large retrospective study illustrates the increasing incidence of PGL in current years as compared to gastric carcinoma. A high diagnostic yield of endoscopic biopsy was established in our study in diagnosing PGL. Endoscopic biopsy along with immunohistochemistry could diagnose majority of cases of PGL. Hence gastric resection can be avoided in few selected cases. Moreover, our study showed the BM involvement is extremely rare in PGL, even with advanced disease. Therefore, BM examination as part of initial diagnostic work-up in PGL should be avoided in PGL.