Gut 1992;33: 1307-1311

Gastric B-cell mucosa associated lymphoid tissuelymphoma: a clinicopathological study in 56 patients

J M Castrillo, C Montalban, G Obeso, M A Piris, M C Rivas

Departments of InternalMedicine and Pathology,Fundacion Jimenez Diaz,Universidad Autonoma,MadridJ M CastrilloG ObesoM C Rivas

Department of InternalMedicine, HospitalRamon y Cajal,Universidad de Alcala,MadridC Montalban

Department ofPathology, HospitalVirgen de la Salud,Toledo, SpainM A PirisCorrespondence to:DrM C Rivas, Department ofPathology, FundacionJimenez Diaz, Av ReyesCatolicos 2, Madrid 28040,Spain.Accepted for publication2 March 1992

AbstractClinico-pathological features of 56 patientswith primary gastric lymphoma were evaluatedretrospectively. All cases were regradedaccording to a classification of Isaacson et alinto high grade and low grade B-cell mucosaassociated lymphoid tissue lymphoma. A thirdgroup of mixed grade was recognised in 11patients with low grade who also had occasionalareas ofhigh grade. Low grade and mixed gradepatients had a 100% actuarial survival at 156months, whichwas significantly better(p<0 01)than that of 52% for patients with high gradedisease. Different treatment methods -surgery, chemotherapy, or a combination ofboth - did not significantly affect survival. Lowgrade tumours occurred mainly in men with ahistory of several years, and who presentedwith non-specific gastric symptoms withoutremarkable exploratory or laboratory findings:most patients were in stage IE-IIE andachieved remission and cure. High grade canhave a shorter history, systemic symptoms,abnormal exploratory and laboratory findings,gastric tumour masses, stage IV disease, and aworse outcome. The only significant .prog-nostic factors for survival were the type oflymphoma and stage IV disease. These find-ings support the Isaacson classification systemwhich separates two extreme groups of gastriclymphomas with different morphology,behaviour, and outcome. The presence oflimited areas of high grade in a specimenshowing low grade does not change theoutcome but suggests that primary gastriclymphoma forms a continuum between theseextreme types.(Gut 1992; 33: 1307-1311)

An important advance in the understanding ofextranodal lymphomas has been the descriptionof a specific mucosa associated lymphoid tissue(MALT)'2 that is normally present mainly inintestinal and bronchial mucosa. 5 MALT is alsopresent in other organs, such as the stomach, butonly after chronic inflammation. Lymphomasarising in these specific areas share many histo-logical, phenotypical, and functional propertieswith normal MALT and exhibit a biologicalbehaviour that differs from that of the nodallymphomas.6 7

Gastrointestinal lymphoma is uncommon,8'2but gastric lymphoma is the most frequentform'3-'7 and could be a good model for theevaluation of MALT tumour behaviour. 3 16Thus, the aim of this study was to evaluateretrospectively the characteristics and outcomeof 56 cases of primary gastric lymphoma thathave been morphologically reclassified according

to recent knowledgelymphomas.

of the MALT derived

MethodsIn this series, 56 patients with primary gastriclymphoma were evaluated retrospectively. Onlypatients who fulfilled the criteria described byIsaacson et al2'` were included. Patients withsecondary involvement of the stomach by non-Hodgkin's lymphoma were therefore excluded.Thirty four patients were studied at theFundacion Jimenez Diaz (Madrid), 14 at theHospital Virgen de la Salud (Toledo) and eight atHospital Ramon y Cajal (Madrid) between 1963and 1990. Clinical data were obtained from theirmedical histories. Upper gastrointestinal radio-graphy was performed in 42 patients and endo-scopy with biopsy specimens in 48 patients.Laparotomy was performed for diagnosis ortreatment, or both, in 50 cases. Five of the sixpatients who had not undergone operation werestaged by computed tomography and bonemarrow biopsy. Patients were staged accordingto the Ann Arbor system.'8 All specimens fromgastrectomy, endoscopy, and bone marrow wereroutinely processed. A panel of monoclonalantibodies was used to characterise the gastriclymphoma in all 50 patients who had undergonesurgery. All the cases were classified as B-celllymphomas with a complete immunopheno-type.'9 All endoscopic biopsy specimens andgastrectomy materials were reviewed and classi-fied according to the proposals of Isaacson et al.20A new group, mixed grade B-cell7 16 19 20 21 was alsoincluded, when an otherwise typical low gradeB-cell MALT lymphoma showed focal areas ofhigh grade B-cell MALT lymphoma.

Gastric surgery was the only treatment in 10patients with low grade, in three patients withmixed grade, and in nine patients with highgrade lymphomas. Adjuvant chemotherapy withCHOP or CHOP-variants was used in threepatients with low grade, in four patients withmixed grade, and in 10 patients with high gradedisease. Two patients with low grade, two withmixed grade, and two with high grade weretreated exclusively with chemotherapy. Twopatients refused treatment. A complete remis-sion was defined as the resolution of clinical,radiological, and endoscopic evidence of disease.A partial remission was considered as a reductionin tumour mass of 50%.

Survival curves of the three histologicalgroups were analysed according to the Kaplanand Meier method22 and the differences betweenthese curves were calculated with the Mantletest. Analysis of survival was calculated in only15 of the 20 patients in the low grade group, in9 of 11 in the mixed grade group, and in 21 of 25

1307

on 9 January 2019 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.33.10.1307 on 1 October 1992. D

ownloaded from

Castrillo, Montalban, Obeso, Pins, Rivas

with high grade lymphomas as 11 patients hadeither inadequate follow up or no treatment.The same methods were also used for the analysisof survival in patients treated with surgery,surgery plus chemotherapy, or chemotherapyalone.

Statistical comparisons of means or percent-ages were performed between sexes, ages,duration of symptoms before diagnosis, symp-toms, data obtained from physical examination,laboratory data abnormalities, radiological andendoscopic findings, and the macroscopicappearances in the gastrectomy specimen in thejoint low/mixed grade group v those in the highgrade group. In patients who had undergoneoperation a X2 test was performed for theanalysis ofthe possible association with mortalityof the following variables: stage I-II v stage IVdisease, histological subtype low/mixed grade orhigh grade, presence or absence of gastric serosalinvasion, presence or absence of lymphoma inthe margin of resection or contiguous extensionof the lymphoma to nearby abdominal organs, orboth, and tumour size over or under 10 cm. Inthe high grade group, a Cox's multivariateanalysis23 was performed to identify which of thefollowing variables such as age, extension toadjacent organs or serosal invasion, stage, andmode of treatment (surgery, chemotherapy, orthe combination of both) might be of indepen-dent significance in predicting mortality.

ResultsThe general characteristics of the patients areshown in Table I. Twenty were classified as lowgrade, 11 as mixed grade, and 25 as high grade.

TABLEI Primary gastric lymphoma - clinical and laboratoryfindings

Low/mixedgrade High grade p

No of patients 20/11 25Sex: male/female 21/10=2.1 11/14=0.78Mean (SD) age (years) (range) 58(13) (24-75) 58(15) (23-79)Mean (SD) duration ofsymptoms (months) (range) 30 (73) (0.2-360) 7 (9) (0 2-36) p=0-08Symptoms (no (%)):

Epigastric pain 24(77) 18 (72)Nausea and vomiting 15 (48) 3 (12) p<001Gastric bleeding 12(38) 9(36)Dyspepsia 13(41) 10(40)Weight loss 11(35) 18(72) p<005Anorexia 7(22) 19(76) p<0 0002Malaise 7(22) 12(48)

Physical examination (no (%)):Palpable mass 2(6) 8(32) p<005Peripheral lymphadenopathy 2(6) 5(20)Hepatomegaly 2(6) 6(24)Splenomegaly 0 1(4)Hyponutrition 1(3) 4(16)Tenderness 9(29) 2(8)Ascites 1(3) 0Pleural effusion 1(3) 0Normal 15(48) 7(28)

Laboratory abnormalities (no (%)):Anaemia (Hb < 10 g/dl) 3(9) 4(16)ESR over 40mm 8(25) 12(48)Serum LDH level >500 U/l 2(6) 7(28)Alkaline phophatase >280 U/1 6(19) 6 (24)Monoclonal globulin 1(3) 0Raised ALT levels 4(13) 1(4)Low protein level 6(19) 6(24)Normal 10(32) 1(4) p<005

Low grade Mixed grade High gradePathological stage (%)*

IE 16/20 (80)** 6/10 (60) 9/25 (36)IIE 4/20 (20) 3/10 (30) 11/25 (44)IV 0/20 1/10 (10) 5/25 (20)

*6 patients not laparotomised. Clinical stage only (1 low grade, 3 mixed grade, 2 high grade).**1 patient was not staged.

Men outnumbered women in the low/mixedgrade group by a ratio of 2. 1. This ratio was0.78 in the high grade group. In the low/mixedgrade group the mean age was 58 (13) years(range 24 to 75) and in the high grade group itwas 58 (15) years (range 23 to 79). Seventy fourper cent of patients were over 51 years old. Themean duration ofsymptoms before diagnosis was30 (73) months in the first group and 7 (9.5) inthe second (p=0 08). In both groups the present-ing symptoms that required hospital consulta-tion were mostly ulcer related pain and a changein longstanding chronic symptoms. Main symp-toms at diagnosis were abdominal pain (77% inthe low/mixed grade group and 72% in the highgrade group), dyspepsia (41% in the low/mixedgrade and 40% in the high grade group), andupper gastrointestinal bleeding (38% and 36%respectively). Patients in the low/mixed gradegroup referred to frequent vomiting (48%) andpatients in the high grade group had weight loss(72%), anorexia (76%), and general deterioration(48%). In the physical examination of the low/mixed grade group, abdominal tenderness wasthe only relatively common finding, occurring in29%, and in 48% physical examination disclosedno abnormality. In the high grade group, apalpable mass was found in 32%, hepatomegalyin 24%, undernourishment in 16%, and peri-pheral lymphadenopathy in 20%: no abnor-mality was found in 28% of these patients.Routine laboratory investigations were notsignificantly abnormal in the low/mixed gradegroup and were, in fact, completely normal in32% of these patients. In the high grade group,48% had raised erythrocyte sedimentation rates,28% serum lactate dehydrogenase values over500 U/l, 24% raised alkaline phosphataseactivities, 24% low total protein concentrations,and 16% anaemia; only 4% of patients in thisgroup had all determinations within normalranges.

In the low grade group, 16 of 20 (80%) patientswere staged as IE and 4 of 20 (20%) as IIE. In themixed grade group 6 of 10 (60%) were staged asIE, 3 of 10 (30%) as IIE, and another case as IV.One case was not clearly staged. In the highgrade group 9 of 25 (36%) patients were stage IE,11 of 25 (44%) stage IIE, and 5 of 25 (20%) werestage IV.

Radiological, endoscopic, and macroscopicfindings in the gastrectomy specimens are shownin Table II. The macroscopic pattern found insurgical specimens correlated with the radio-logical pattern in only 7 of 18 (38%) of the low/mixed grade group and in 5 of 16 (3 1%) of thepatients in the high grade group; the patternscorrelated with the endoscopic findings in only8 of 19 (42%) of the low/mixed grade group andin 7 of 17 (41%) of the high grade group. In thegastrectomy specimen an infiltrative/ulcerativepattern was found in a significant percentage ofcases with low/mixed grade lymphoma (p<005)and a large tumour mass in patients with highgrade (p<OO1).

In the low grade group, all 15 patientsachieved complete remission and in the mixedgrade group 8 (88 8%) patients achieved com-

plete and 1 (11%) partial remission. In the highgrade group, complete remission was achieved in

1308

on 9 January 2019 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.33.10.1307 on 1 October 1992. D

ownloaded from

Gastric B-cell mucosa associated lymphoid tissue lymphoma: a clinicopathological study in 56 patients

TABLE II Primary gastric lymphomas - upper gastrointestinal radiology, endoscopy, andgastrectomy specimen patterns. (Values, no (%))

Radiology Endoscopy Gastrectomy

LGIMG HG LG/MG HG LGIMG HGPattern (n=23) (n= 19) (n=27) (n=21) (n=27) (n=23)

Infiltrative/ulcers 9 (39) 10(52) 10(37) 9 (42) 12 (44) 1(4)p<O.05

Infiltrative/erosions - - 5 (27) 2 (9) 4(15)Infiltrative 9 (39) 3 (16) - - -Ulcers 5 (22) 3 (16) 12 (44) 8 (38) 7 (26) -

Tumour/mass - 3 (16) - 2 (9) 4 (15) 22 (%)p<0-01

LG/MG=low/mixed grade; HG=high grade.

12 patients (57%), partial remission in 3 (14%),and there was no response in 6 (28%). The meanduration of complete remission in the low gradegroup was 65 (48) months (range 6 to 156) andonly 1 of 15 (6.6%) patients relapsed after 48months. In the mixed grade group, the mean

duration of complete remission was 46 (31)months (range 5 to 88) and no relapse was

recorded. In the high grade group, the mean

duration of complete remission was 66 (101)months (range 4-354); 2 of 12 (16-7%) of thepatients in complete remission relapsed after 48months.

Actuarial survival in the low and mixed gradegroups was 100% at 156 months, significantlybetter (p<001) than the 52% survival in the highgrade group (Figure). Only two patients withlow grade or mixed grade disease died, but inneither case was the cause related to lymphoma.However, nine patients in the high grade group

died. Six ofthese died between 10 to 60 days aftersurgery. All these patients had bulky gastrictumours extending to nearby or distant organsand death was due to progressive and uncon-

trollable disease and was not directly related tosurgery. The two patients who had a late relapsedied, and in another patient death was notrelated to lymphoma. In the whole series,actuarial survival was not significantly differentin patients treated with surgery, surgery pluschemotherapy, or chemotherapy alone. Highergrade of histological malignancy (p<0-01) andstage IV disease (p<0 05) were the only variablesthat influenced mortality. In the high gradegroup, none of the variables or different modesof treatment influenced mortality.

1*2

._

to.00.C-

CL

,a

1.0

0-8

0-6

L ga vigep**** *

MiLow grade v high gradeMixed grade vhigh grade <

0-4 _-

0-2

0

0 100 200 300

Time (months)

- High grade o Mixed grade * Low grade(n = 21) (n=9) (n = 15)

Survival probability in the three groups ofpatients with pnrmary gastric lymphomas.Mixed grade=low grade with occasional areas of high grade.

In four patients, a second neoplasm wasassociated with the gastric lymphoma - one hadcoexisting gastric carcinoma, another a coexist-ing colonic carcinoma, and a third developedurothelial carcinoma 11 years after treatmentwith a chemotherapeutic regimen that includedcyclophosphamide. A gastric lymphomadeveloped in the fourth patient seven years afterradiotherapy treatment for Hodgkin's disease.

DiscussionIn most series patients with primary gastriclymphoma have been classified according toRappaport,2>27 Kiel,' 4 15 28 Lukes-Collins, 129or Working Formulation systems." 24 25 29130These classifications are able to separate groupsof nodal lymphomas with a clear correlationbetween morphological features and clinicalbehaviour. However, these classifications cannotsuccessfully be extrapolated to gastrointestinallymphomas as MALT lymphocytes and theirderived lymphomas seem to have a physiologythat differs from their nodal counterparts.27 Inthis study, cases have been classified according tothe system proposed by Isaacson et a121 in anattempt to evaluate its reliability in separatingdifferent types of gastric lymphoma. Cases wereclassified into two groups ofMALT lymphomas:low grade and high grade with or without a lowgrade component.20 However, in a previousreport, using morphological, microscopic andultrastructural criteria, and immunologicalmarkers, we found that 46% of the patients withlow grade malignancy had areas with a popula-tion of large cells reflecting the presence of a highgrade malignant component.2' Thus, we havealso included a third group, the mixed gradeMALT lymphoma already discussed7 1619 20 21 butnot included in the initial classification ofIsaacson et al.20 In this series, 11 cases have beenclassified as mixed grade, representing a con-siderable percentage of all cases of B-cell MALTlymphoma (19-6%). This finding suggests thatprimary gastric MALT lymphoma forms a con-tinuum with well delineated entities at eitherend, which have a graded sequence ofan increas-ing amount oflarge cells. It is also possible that inthis continuum a tumour increases the amount oflarge cells and develops higher grades of malig-nancy transforming low grade into high grade asoccurs in nodal counterparts.

Actuarial survival in both low grade and mixedgrade is excellent, with a 100% probability ofsurvival after 156 months of treatment. Thissurvival is significantly better (p<001) than thatof the patients with high grade lymphoma whoseprobability of survival is limited to 52%. Thisdifferent outcome confirms that the groupingproposed by the Isaacson classification7 16 20 iSreliable in terms of prognosis and that thepresence of a limited component of high grade ina low grade case has no ominous prognosticsignificance.

In the whole series there were no differencesin survival in the groups treated with chemo-therapy, surgery, or the combination of both.These results and the good overall survival of thelow grade and mixed grade groups underline thefacts that most of these patients had localised

1309

on 9 January 2019 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.33.10.1307 on 1 October 1992. D

ownloaded from

Castrillo, Montalban, Obeso, Piris, Rivas

disease, that after achieving complete remissionmost patients are cured,9 151 7 and that surgery canbe a definitive treatment.242829 However, in thehigh grade group the disease can present with abulky gastric tumour extending to nearby ordistant organs and can display much moreaggressive behaviour which cannot be controlledby surgery. This presentation implies a very poorprognosis despite treatment, as was the case insix patients in this series. Relapses in high gradepatients also seem to be ominous events as bothpatients who relapsed have died. Patients whodid not present with bulky or extensive initialdisease and who did not relapse after completeremission had a good prognosis, comparablewith that of low or mixed grade patients.

In the high grade group in this series, surgery,chemotherapy, or the combination of both, didnot influence survival and the effectiveness ofthe different modes of treatment needs furtherassessment. Some authors maintain that effec-tive gastrectomy is essential for cure,'0 14 24 29while others suggest that surgery alone is insuf-ficient and a combination of chemotherapy andsurgery is better.253' Others have shown thatsurgery does not improve the results of chemo-therapy or radiotherapy.262830 Moreover,only two thirds of tumours are resectable atsurgery.0 14 26 Gastrectomy is not devoid ofcomplications, mortality has been reported asbeing as high as 16%' 428 and there is a highincidence of short and long term complicationsafter surgery.2521 The use of endoscopic biopsyspecimens in histopathological diagnosis26 27 andof imaging techniques in staging disease extentavoids the need for laparotomy and gastrectomywhich consequently can be reserved for aminority of cases.27 28131 A recent prospectivestudy of high grade gastric lymphoma30 foundthat survival was similar in patients who under-went complete, incomplete, or no surgicaltherapy before chemotherapy and that the prog-nostic factors predicting survival were similar tothose in patients with nodal lymphoma treatedwith the same chemotherapy regimen. Thisfinding supports the view that chemotherapyshould be the elective treatment in this group.Although bleeding and perforation have beenfeared complications in non-operated patientsthey have not occurred in this series or inothers.273032 If these results can be confirmed,chemotherapy without gastrectomy should bethe treatment of choice for aggressive gastriclymphomas.When the characteristics and the extent of the

gastric MALT lymphoma are considered, onlythe histological grade and stage IV disease (as inall major series'5 24 29 33) influenced mortality.Tumour size over 10 cm, presence of disease inthe margins of resection, or extension to adjacentorgans had no influence. Neither the classicalvariables nor the different treatments influencedsurvival in high grade lymphoma. These datasupport the validity of classifying the two differ-ent grades of gastric MALT lymphoma.20

Clinical findings also support the view that lowand mixed grade form a specific group that canbe separated from the high grade group. In bothgroups, patients were middle aged to elderly, butthere is a striking incidence of low/mixed grade

in men that is not found in the high grade group.Although in both groups the disease can be eitherindolent or slowly progressive and most caseshad a non-aggressive course, in patients groupedas low/mixed grade the disease can last for yearsand in those grouped as high grade for monthsonly, a difference that is almost significant. Thesymptoms are non-specific and not very differentin either group, except for the incidence ofanorexia, weight loss, and a general malaise inhigh grade patients. Reflecting this scarcity ofsymptoms, physical examination is normal inalmost 50% of patients with low/mixed grade andin 28% of those with high grade. In the highgrade group it is common to find a palpable mass,hepatomegaly or splenomegaly, and palpablelymphadenopathy. Anaemia, raised erythrocytesedimentation rates, and serum lactate dehydro-genase activities are common findings in highgrade, but the analytical findings are normal inup to one third of the patients with low/mixedgrade. Although both groups tend to appear as alimited disease, most cases being in stage I-IIE,the morphohistological subtype seems to be adecisive factor in dissemination, as only one case(5%) of low/mixed grade compared with 20% ofthe patients with high grade were classified asstage IV. Upper gastrointestinal radiographyand endoscopy are very sensitive in showing thepresence of gastric disease, with infiltrative andand ulcerative patterns in both groups and atumour pattern only in high grade. These pro-cedures can detect the disease, but they are not soaccurate as the macroscopic study of a surgicalspecimen in determining the pattern and exten-sion of the tumour. The correlations betweenradiology and endoscopy and the surgical speci-men were very poor: there was agreement withthe surgical specimen in only 31 to 38% and in 41to 42% respectively.A final point is the occurrence of non-

lymphoid neoplasias in four patients. Thisassociation means that primary gastric MALTlymphoma can develop as a second neoplasm orthat other neoplasms can occur in both treatedand untreated patients with gastric lymphoma.Despite the slow non-aggressive course of gastriclymphoma, these patients behave as immuno-compromised subjects who may eventuallydevelop immunodeficiency-related secondaryneoplasias. Moreover, a high incidence of gastricadenocarcinomas coexisting (over 60 cases) orfollowing (17 cases) gastric lymphomas havebeen reported.34 This increased incidence ofgastric carcinomas seems to be related to localfactors. An increased proliferative capacity ofthegastric epithelium next to the areas of MALTderived lymphomas has been postulated (P GIsaacson, personal communication) as a possiblemechanism. Helicobacter pylori infection isassociated with an increased risk of gastricadenocarcinoma35 36 and is also found in 92% ofpatients with gastric MALT lymphoma,37supporting its role as cofactor in the pathogenesisof both diseases. This finding may explain theeventual synchronous or metachronous occur-rence of these two diseases in the same patient.

In summary, the two types low/mixed gradeand high grade recognised in the Isaacson classi-fication20 delineate two extreme types of gastric

1310

on 9 January 2019 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.33.10.1307 on 1 October 1992. D

ownloaded from

Gastric B-cell mucosa associated lymphoid tissue lymphoma: a clinicopathological study in 56 patients 1311

MALT lymphomas that differ in their presenta-tion, clinical behaviour, and response to treat-ment.'3 Low/mixed grade occur in elderly menwith a history extending over several years, non-specific local symptoms, no abnormal explora-tory or laboratory findings, a gastric infiltrative/ulcerative pattern, limited stage I-II disease, andwith a very good response to treatment andprolonged survival. On the other hand, althoughhigh grade may follow a similar course, in somepatients it is associated with systemic symptoms,laboratory and exploratory abnormalities, largetumour masses in the stomach, extensive anddisseminated disease, and failure of treatmentleading to a poor prognosis.We are indebted to Professor P G Isaacson for his critical review ofthe manuscript and Dr J Mufiiz and Dr R Gabriel for their advicein the statistical study.

1 Freeman C, Berg JW, Cutler Sj. Occurrence and prognosis ofextranodal lymphomas. Cancer 1972; 29: 252-60.

2 Isaacson PG, Wright DH. Extranodal lymphoma. In:Anthony P, Mac Sween RNM, eds. Recent advances inhistopathology. London: Churchill-Livingstone, 1987:159-84.

3 Spencer J, Finn T, Isaacson PG. Human gut-associatedlymphoid tissue: a morphological and immunocytochemicalstudy of the human appendix. Gut 1985; 26: 672-9.

4 Spencer J, Finn T, Isaacson PG. A comparative study of thegut-associated lymphoid tissue of primates and rodents.VirchowArch [BJ (cell Pathol) 1986; 51: 509-19.

5 Bienenstock J, Befus D. Gut and bronchus associatedlymphoid tissue. AmJ Anat 1984; 170: 437-45.

6 Isaacson PG, Wright DH. Malignant lymphoma of mucosa-associated lymphoid tissue. A distinctive type of B-celllymphoma. Cancer 1983; 52: 1410-6.

7 Isaacson PG, Spencer J. Malignant lymphoma of mucosa-associated lymphoid tissue. Histopathology 1987; 11: 445-62.

8 Lewin K, Ranchod M, Dorfman RF. Lymphomas of thegastrointestinal tract. Cancer 1978; 42: 693-707.

9 Filippa DA, Lieberman PH, Weingrad DN, Decosse JJ,Bretsky SS. Primary lymphomas of the gastrointestinaltract. Analysis of prognostic factors with emphasis onhistological type. AmJ Surg Pathol 1983; 7: 363-72.

10 Dragosic B, Bauer P, Radaszkiewiczt T. Primary gastro-intestinal non-Hodgkin's lymphomas. A retrospective studyof 150 cases. Cancer 1985; 55: 1060-73.

11 Papadimitriou CS, Papacharalampous NX, Kittas C.Primary gastrointestinal malignant lymphoma. Cancer1985; 55: 870-9.

12 Burke JS, Sheibani K, Nathwani BN, Winberg CD,Rappaport H. Monoclonal small (well-differentiated)lymphocytic proliferation of the gastrointestinal tractresembling lymphoid hyperplasia. Hum Pathol 1987; 18:1238-45.

13 Isaacson PG, Spencer J, Finn T. Primary B-cell gastriclymphoma. Hum Pathol 1986; 17: 72-82.

14 Hockey MS, Powell J, Crocker J, Fielding JW. Primarygastric lymphoma. BrJ Surg 1987; 74: 483-7.

15 Ravaioli A, Amadori M, Faedi M, Rosti G, Folli S, Barbieri C,et al. Primary gastric lymphoma: a review of 45 cases. EurjCancer Clin Oncol 1986; 22: 1461-5.

16 Moore I, Wright DH. Primary gastric lymphoma. A tumour ofmucosa-associated lymphoid tissue. A histological andimmunohistochemical study of 36 cases. Histopathology1984; 8:1025-39.

17 Barzilay J, Rakowsky E. Gastric lymphoma: a clinical study of28 cases and evaluation of prognostic factors. Clin Oncol1984; 10: 233-40.

18 Carbone PP, Kaplan HS, Musshoff K, Smithers DW,Tubiana M. Report of the Committee on Hodgkin's diseasestaging procedures. Cancer Res 1971; 31: 1860-1.

19 Piris MA. Linfomas gastricos B del teiido linfoide asociado amucosas (MALT). Estudio humano y experimental. Con-sideraciones morfoinmunofenotipicas y ultraestructurales.Madrid: Facultad de Medicina, Universidad Autonoma,1990. Tesis Doctoral.

20 Isaacson PG, Spencer J, Wright DH. Classifying primary gutlymphomas. Lancet 1988; ii: 1148-9.

21 Piris MA, Rivas C, Morente M, Orradre JL, Rodriguez R,Cruz MA, et al. Gastric B-cell low grade lymphomaoriginated in mucosa-associated lymphoid tissue. Relation-ship of tumoral cells with marginal zone and monocytoidB-lymphocytes. Immunohistochemical and ultrastructuralstudies. Med Clin (Barc) 1990; 94: 601-6.

22 Kaplan EL, Meier P. Non parametric estimation fromincomplete observations. JAm Stat Assoc 1958; 53: 457-81.

23 Cox DR. Regression models and life tables. J Stat Soc 1972;34:187-220.

24 Rosen CB, Van Heerden JA, Martin JK, Wold LE, IlstrupDM. Is an aggressive surgical approach to the patient withgastric lymphoma warranted? Ann Surg 1987; 205: 634-40.

25 Shiu MH, Nisce LZ, Pinna A, Straus DJ, Tome M, FilippaDA, et al. Recent results of multimodal therapy of gastriclymphoma. Cancer 1986; 58: 1389-99.

26 Maor MH, Maddux B, Osborne BM, Fuller LM, Sullivan JA,Nelson SR, et al. Stages IA and IIE non-Hodgkin'slymphomas of the stomach. Comparison of treatmentmodalities. Cancer 1984; 54: 2330-7.

27 Maor MH, Velasquez WS, Fuller LM, Silvermintz KB.Stomach conservation in stages IE and IIE gastric non-Hodgkin's lymphoma.J Clin Oncol 1990; 8: 266-71.

28 Gobbi PG, Dionigi P, Barbieri F, Corbella F, Bertoloni D,Grignani G, et al. The role of surgery in the multimodaltreatment of primary gastric non-Hodgkin's lymphomas.Cancer 1990; 65: 2528-36.

29 List AF, Greer JP, Cousar JC, Stein RS, Johnson DH,Reynolds VH, et al. Non-Hodgkin's lymphoma of thegastrointestinal tract: an analysis of clinical and pathologicalfeatures affecting outcome. J Clin Oncol 1988; 6: 1125-33.

30 Salles G, Herbrecht R, Tilly H, Berger F, Brousse N,Gisselbrecht C, et al. Aggressive primary gastrointestinallymphomas: review of 91 patients treated with the LNH-84regimen. A study of the Groupe d'Etude des LymphomesAgressifs. AmJ Med 1991; 90: 77-84.

31 Burgers JM, Taal BG, Van Heerde P, Somers R, den HartogJager FC, Hart AA. Treatment results ofprimary stage I andII non-Hodgkin's lymphoma of the stomach. RadiotherOncol 1989; 11: 319-26.

32 Hermann R, Panahon A, Barcos MP, Walsh D, Stuztman L.Gastrointestinal involvement in non-Hodgkin's lymphoma.Cancer 1980; 46: 215-22.

33 Aozasa K, Ueda T, Kurata A, KimCW, Inoue M, Matsuura N,et al. Prognostic value of histologic and clinical factors in 56patients with gastrointestinal lymphomas. Cancer 1988; 61:309-15.

34 Garcia-Giron C, Feliu J, Ordoniez A, Zamora P, Blanco S,Sanchez MJ, etal. Gastric adenocarcinoma after treatment ofgastric non-Hodgkin's lymphoma. Med Clin (Barc) 1990;95:698-701.

35 Parsonnet J, Friedman GD, Vandersteen DP, Chang Y,Vogelman JH, Orentreich N, et al. Helicobacter pyloriinfection and the risk of gastric carcinoma. N Engl J Med1991; 325: 1127-31.

36 Nomura A, Stemmermann GN, Chyou PH, Kato I, Perez-Perez G, Blaser MJ. Helicobacter pylori infection and gastriccarcinoma among Japanese Americans in Hawaii. N EnglJ Med 1991; 325: 1132-6.

37 Wotherspoon AC, Ortiz-Hidalgo C, Falzon MR, Isaacson PG.Helicobacter pylori associated gastritis and primary B-cellgastric lymphoma. Lancet 1991; 338: 1175-6.

on 9 January 2019 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.33.10.1307 on 1 October 1992. D

ownloaded from