981J Clin Pathol 1990;43:981-986

Gastric epithelium in the duodenum: Itsassociation with Helicobacter pylori andinflammation

J I Wyatt, B J Rathbone, G M Sobala, T Shallcross, R V Heatley, A T R Axon,M F Dixon

AbstractDuodenal biopsy specimens from 471adults and 47 children were examined todetermine the prevalence and distribu-tion of gastric epithelium in theduodenal bulb in relation to age, gender,gastroduodenal inflammation, smoking,alcohol and consumption of non-steroidal anti-inflammatory drugs(NSAID). Gastric metaplasia waspresent in the anterior wall duodenalbiopsy specimen in 31%, was significan-tly less common in patients under 17than in adults, and was more common inmales than females. In sixty two adultswho underwent multiple radial duodenalbiopsy gastric metaplasia was randomlydistributed around the duodenal cir-cumference; sixty three per cent of thepatients with gastric metaplasia foundon multiple biopsy were detected by justthe anterior biopsy. Gastric metaplasiawas not obviously associated with alco-hol, cigarette, or NSAID consumption.While the presence of gastric metaplasiawas associated with adulthood, male sex,and low fasting gastric juice pH, itsextent was associated with activeduodenitis and Helicobacter-associatedgastritis. On logistic regression, gastricmetaplasia in the duodenum and gastricHelicobacter pylori were independentpredictors of active duodenitis, but werenot significantly associated with inactiveduodenal inflammation. H pylori wasobserved in duodenal biopsy specimensfrom 32 patients, all with active duoden-itis; bacteria were present only on foci ofgastric metaplasia, and were more likelyto be seen when the metaplasia wasextensive.

It is proposed that inflammatoryinjury to the duodenal mucosa by Hpylori may stimulate the development offurther gastric metaplasia, and that thearea of duodenum susceptible to colon-isation with H pylori may thereforeincrease progressively until mucosalintegrity is compromised and ulcerationsupervenes.

Helicobacter-associated chronic gastritis isvery strongly associated with duodenalulceration, being present in over 90% ofpatients with duodenal ulcer.' 2 We have

previously reported that active chronicduodenitis (the pattern of inflammation com-monly present at the margin of duodenalulcers3) occurs specifically in the group ofpatients who have both Helicobacter-associated gastritis and more than a minimaldegree of metaplastic gastric epithelium in theduodenum.4 Gastric metaplasia is common inthe proximal duodenum, where it is believedto be induced by acid injury.3 Helicobacterpylori colonises only gastric type epith-elium'245 and is therefore able to infect theduodenum only when gastric metaplasia ispresent. We have proposed that the combina-tion of H pylori and gastric metaplasiainduced by acid are the prerequisites for thedevelopment of active duodenitis, and thatthis inflammation impairs mucosal defences,leading ultimately to duodenal ulceration.6Duodenal ulcer is a multifactorial con-

dition.7 The aetiology is closely related toenhanced acid secretion, but has been foundto be influenced by many other factors includ-ing gender, genetic predisposition, alcoholconsumption, drug intake, smoking, and mostrecently H pylori infection. Some of thesefactors may also affect the occurrence of meta-plastic gastric epithelium in the duodenumand consequently influence susceptibility toduodenitis and ulceration.The aim of this study was to investigate the

prevalence and distribution of gastric epi-thelium in the duodenum in relation to age,gender, alcohol consumption, smoking, fast-ing gastric juice pH, ingestion of non-steroidal anti-inflammatory drugs (NSAID)and gastroduodenal inflammation.

MethodsFour hundred and seventy one patients withdyspepsia who had upper gastrointestinalendoscopy with biopsy performed at definedsites as part of three dyspepsia studies wereinvestigated. The studies had been approvedby the hospital ethical committees and thepatients had given informed consent. (Thesepatients include 252 of the patients in aprevious study.4) All of the patients (aged 18-84, 209 of them male) had a biopsy specimentaken from the anterior wall of the first part ofthe duodenal mucosa as well as specimensfrom constant sites in the antral and bodygastric mucosa.To determine the distribution of gastric

epithelium in the duodenum and to estimate

Department ofHistopathology,St James's UniversityHospital, Leeds, YorksLS9 7TFJ I WyattDepartment ofMedicineT ShallcrossR V HeatleyA T R AxonDepartment ofGastroenterology,Leeds GeneralInfirmaryB J RathboneGM SobalaDepartment ofPathologyM F DixonCorrespondence to:Dr J I WyattAccepted for publication19 July 1990

on February 8, 2020 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.43.12.981 on 1 Decem

ber 1990. Dow

nloaded from

Wyatt, Rathbone, Sobala, Shallcross, Heatley, Axon, Dixon

Figure IA Gastric metaplasia present in a duodenal biopsy specimen which is otherwise normal. The metaplasticepithelium (arrows) is most often present on the tips of villi (alcian blue/PAS).Figure IB Duodenal biopsy specimen showing active duodenitis with surface erosion occurring in an area ofgastricmetaplasia (alcian bluelPAS).Figure 1C The samefield as seen in Fig lB; H pylori can be seen colonising the metaplastic gastric epithelium oneither side of the eroded area (arrows) (modified Giemsa stain).Figure ID High magnification to show H pylor closely applied to the surface of thegastric type cells (modifiedGiemsa stain).

the sampling error of a single biopsy specimenadditional directed duodenal biopsy speci-mens from the posterior wall, roof, and floorof the first part of the duodenum werecollected prospectively from 62 of the patients(aged 18-74, 28 of the male), and 38 patientsin this group also had anterior and posteriorbiopsy specimens taken from the second partof the duodenum.

Details of endoscopic findings, ulcer his-tory, smoking habit, and alcohol and NSAIDconsumption were recorded on pro-formarecord sheets for all of the adult patients. ThepH of fasting gastric juice aspirated at thebeginning of endoscopy was measured in 94.

Forty seven children aged under 17 fromwhom one to five (median two) gastric andone to two (median one) duodenal biopsyspecimens were available were included in thestudy. This material was examined retro-spectively and represented all of the paediatricgastroduodenal biopsy specimens obtained atSt James's Hospital between 1981 and 1988.

HISTOLOGICAL ANALYSISAll biopsy specimens were stained withhaematoxylin and eosin and examined by onepathologist for gastritis and duodenitis, usingstandard criteria as described previously.48

"Chronic duodenitis" was characterised by adefinite increase in lamina propria inflam-matory cells together with flattening of villiand evidence of epithelial cell damage. Ifneutrophils had infiltrated the lamina propriaand epithelium in addition to evidence ofchanges indicative of chronic duodenitis theappearances were classified as "active duo-denitis". The presence of gastric epitheliumwas not a diagnostic criterion for active duo-denitis. Mild changes in villous architectureand minor increases in lamina propria cellu-larity were designated "no significant duo-denitis".

Sections of all biopsy specimens were alsostained by the modified Giemsa technique forH pylori,9 and duodenal biopsy specimenswere stained by alcian blue/periodic acidSchiff for the detection of gastric epitheliumwhere this was not apparent on haematoxylinand eosin staining. Gastric epithelium wasdefined by the presence of adjacent surfaceepithelial cells containing periodic acid Schiffpositive neutral mucin (fig 1). The extent ofthe gastric epithelium was graded 1-4 depen-ding on whether one pathologist estimatedthat it covered less than 5%, 5-20%, 20-50%or greater than 50% of the surface in thebiopsy specimen respectively, and this grad-

982

on February 8, 2020 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.43.12.981 on 1 Decem

ber 1990. Dow

nloaded from

Prevalence and distribution ofgastric metaplasia

Table 1 Prevalence and extent ofgastric metaplasia (GM) in 62 patients withmultiple duodenal biopsy specimens

Maximum gastric Median gastric Mean gastricNumber of metaplasia grade metaplasia grade metaplasia gradebiopsy specimens for each patient of biopsy specitnen of biopsy specimenwith gastric Number (%) with gastric with gastricmetaplasia ofpatients 1 2 3 4 metaplasia metaplasia

o 27 (0)* - - - - - -

I 11 (36) 8 2 - 1 1 1-42 5(20) 4 - - 1 1 1-33 12 (100) 1 2 3 2 2 2-04 12 (100) 1 2 3 6 3 2-4

*Percent of these patients with gastric metaplasia in the anterior duodenal biopsy specimen.

ing was used to relate gastric metaplasia to theother variables in the results section. Thedegree of gastric metaplasia in 100 consecutiveduodenal biopsy specimens was also gradedindependently by a second histopathologistto determine the degree of interobservervariation.

Statistical analysis was performed using thex2 test, Fisher's exact test, the Wilcoxon andKruskall Wallis methods for ranked, non-parametric variables and K statistics for theinterobserver variation. Multiple regressionand logistic regression analyses were per-formed using the SAS package.10

ResultsA total of 780 duodenal biopsy specimens from518 patients were studied. Biopsy specimensfrom 174 patients (33-6%, 95% confidencelimit 29-5-37-7%) contained gastric epi-thelium, the extent of which varied from smallfoci of at least three adjacent epithelial cells, toareas comprising the whole of the biopsyspecimen surface. Gastric epithelium was oftenbut not always associated with a degree ofvillous blunting.

In five patients gastric epithelium was seento overlie fully differentiated fundic-type gas-tric mucosal glands without any inflammation;this was considered to represent congenitalgastric heterotopia in the duodenum, asdescribed previously.4 In all other cases thegastric epithelium occurred without this typeof mucosa beneath, fulfilling the criteria for"gastric metaplasia".3" This change tended tobe present more frequently on villous tips thanon the sides or in mucosal crypts. The periodicacid Schiff positive epithelium varied fromcells with abundant cytoplasmic mucin to cellswhere the periodic acid Schiff positive brushborder of normal enterocytes was replaced by anarrow zone of cytoplasmic positivity, oftenwith underlying periodic acid Schiff positivegranules within the cytoplasm. Both were in-cluded in the definition of gastric metaplasia inthis study, although it was apparent that inactive inflammation the niucin content of thecells tended to be reduced.

Table 2 Association between gastric metaplasia and age

Grade ofgastric metaplasia Total numberTotal number with gastric

Age None 1 2 3 4 ofpatients metaplasia (%)

< 20 54 3 1 - - 58 4 (7)20-39 97 20 15 9 4 145 48 (33)40-59 119 33 25 3 8 188 69 (37)> 60 87 18 14 4 4 127 40 (31)Total 357 74 55 16 16 518 161 (31)

DISTRIBUTION OF GASTRIC METAPLASIA IN THEDUODENUMOf the 62 patients from whom multipleduodenal biopsy specimens were taken, 35(56%) (95% confidence limit 43-69%) hadgastric metaplasia in at least one. In 22 (35-5%)(95% confidence limit 45-79%) gastric meta-plasia was present in the anterior biopsyspecimen, a prevalence similar to the 33% inthe group of 409 adult patients from whomsingle anterior biopsy specimens had beentaken. Therefore, in this group the singleanterior duodenal biopsy detected 22 out of 35(63%) patients with gastric metaplasia foundby multiple biopsy. Gastric metaplasia wasdistributed randomly around the circum-ference of the duodenum; the numbers ofpatients with gastric metaplasia in the anterior,posterior, roofand floor biopsy specimens were22, 17, 27, and 20, respectively. Gastric meta-plasia was present in the biopsy specimens ofthe second part of the duodenum in only two ofthe 38 patients studied; both of these hadgastric metaplasia in all four of the first partbiopsy specimens.The extent ofgastric metaplasia in the biopsy

specimens from the first part of the duodenumis shown in table 1. In general, gastric epi-thelium was either found as small foci in one totwo specimens, or was present in three to four,in which case it tended to be more extensive.Both the median and the mean grade of gastricmetaplasia for each biopsy specimen increasedsignificantly with increasing prevalence or gas-tric metaplasia among the radial biopsyspecimens (p = 0-0051).

RELATION OF GASTRIC METAPLASIA TO AGE ANDGENDERThe prevalence of gastric epithelium inanterior duodenal biopsy specimens in patientsaccording to age is shown in table 2. It waspresent significantly less frequently in children(p < 0-0001) but did not vary with age amongthe adult patients. Gastric metaplasia waspresent in 41% of male patients and in 23% offemale patients (p = 0-0005).

RELATION OF GASTRIC METAPLASIA TOGASTRODUODENAL INFLAMMATIONGastric metaplasia was not seen in duodenalbiopsy specimens from any of the 17 patientswho had chronic atrophic gastritis and intest-inal metaplasia affecting body-type gastricmucosa, but was present in 12 out of 57 (21%)patients with intestinal metaplasia in theantrum. Otherwise the prevalence of duodenalgastric metaplasia did not vary with gastricinflammation. Gastric metaplasia was presentin the anterior biopsy specimen in 34% of Hpylori positive patients and 32% of patientswithoutH pylori. The extent of the metaplasia,however, was signficantly greater in thosepatients with Helicobacter-associated gastritisthan in the others, with 20-8% of the Hpyloripositive patients having more than 5% gastricmetaplasia compared with 14-9% of the Hpylori negative patients (p = 0-002) (table 3).

There was a clear, quantitative relation bet-ween gastric metaplasia and active duodenal

983

on February 8, 2020 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.43.12.981 on 1 Decem

ber 1990. Dow

nloaded from

Wyatt, Rathbone, Sobala, Shallcross, Heatley, Axon, Dixon

Table 3 Extent ofgastric metaplasia in anterior duodenal biopsy specimen in relationto H pylori state

Grade ofgastric metaplasia Total (%)Hpylori with gastricstate None 1 2 3 4 Total metaplasia

Positive 176 36 28 1 1 16 267 91(34)Negative 139 35 25 5 0 204 65 (32)

inflammation. Active duodenitis was rarelyseen in the absence of gastric metaplasia andshowed an increasing prevalence in biopsyspecimens with increasing extent of gastricmetaplasia (fig 2). This applied only to activeduodenitis; in patients with chronic duodenitisand those without clinically important duoden-itis the duodenal histology showed no sig-nificant association with gastric metaplasia.

Active duodenitis was present in 56 anteriorbiopsy specimens and Hpylori was observed in32 of these. H pylori were patchily distributedand seen only on areas ofgastric metaplasia; thebacteria were more likely to be seen in biopsyspecimens with larger areas of gastric epi-thelium (fig 2). Overall, 43 out of the 512patients had a duodenal ulcer, and 40 (93%) ofthese had Helicobacter-associated gastritis; theother three had all recently taken NSAID. In29 (67%) of the patients with duodenal ulcergastric metaplasia was identified in the anteriorbiopsy specimen, compared with 30 7% ofadults without duodenal ulcer (p < 0-0001).Twenty nine of the patients with duodenalulcer had duodenitis, which was active in 19,and H pylori was seen in the duodenal biopsyspecimen in 12. Thirty seven patients hadactive duodenitis not associated with duodenalulceration, and H pylori was observed in 20 ofthese biopsy specimens (p > 0-05).

REPRODUCIBILITY OF GRADING OF GASTRICMETAPLASIAThe two pathologists agreed over the presenceor absence ofgastric metaplasia in 94 out of 100biopsy specimens (K value = 0 88); in the sixcases of disagreement five were graded as 1( < 5°% metaplasia) by one of the pathologists.

Figure 2 Histogramshowing the relationbetween duodenalinflammation and grade ofgastric metaplasia.

EJ Normal duodenal histology=I No clinically important duodenitis3 Chronic duodenitism Active duodenitis H pylor negative

Active duodenitis H pylori positive1001 r- lrE

80 1

_ l60-e

.1 40-

20 -

0

8-

7-

i6-El

3.

& 4-

us 3-

1.

A Male H pori negtiVe*Male Hpoi positiveo Femae H pyori ngtive* Female H pylori positive

0A

A*A

A&O*AA

None .5 5-20 20-50 >50Extent of gastric metaplasia in anterior duodenal biopsy (1/.)

Figure 3 The relation between gastric metaplasia in theduodenum andfasting gastric juice pH, gender, andH pylori state in 94 patients.

When the agreement over the grading of theextent of metaplasia (graded 0-4) was tested,there was concurrence in 79 out of 100 of thebiopsy specimens, and in only five cases wasthere more than one grade difference in theallocations. This gave a K value of 0-66, whichindicates a very high level of agreement overthe five categories. The biopsy specimens inwhich there was more than one grade of dis-agreement tended to be those where the gastricmetaplasia was characterised by only a narrowzone of periodic acid Schiff positivity beneaththe epithelial cell surface (so-called"immature" gastric metaplasia)."2

RELATION OF GASTRIC METAPLASIA TO OTHERFACTORSThere was a significant association betweengastric metaplasia and a low fasting gastric juicepH (p = 0-0001) (fig 3). Gastric metaplasiaoccurred in patients with low fasting pHregardless of their gender or Hpylori state. Nosignificant association was found between thepresence or extent of gastric metaplasia and ahistory of smoking, alcohol consumption, orNSAID.

Multiple regression analysis showed that forthe 94 patients with fasting gastric juice pHmeasurements, male gender was not anindependent predictor of gastric metaplasiaafter pH had been taken into account. Logisticregression showed that Helicobacter-associated gastritis and the degree of gastricmetaplasia were independent predictors ofactive duodenitis (p = 0-0001 for gastric meta-plasia, p = 0-0321 for H pylori state), butneither showed any significant association withduodenal inflammation which was not active.If gastric juicepH was substituted in the modelfor the extent of gastric metaplasia, then Hpylori state and gastric juice pH remainedsignificant independent predictors of activeduodenitis (p = 0-01 for pH, p = 0-027 for Hpylori state).

(n416) DiscussionIt is now widely accepted that H pylori infec-

I I I I

984

*:**-0

.1

u

6A

*A

0t.

AAfi&

A*%AO

on February 8, 2020 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.43.12.981 on 1 Decem

ber 1990. Dow

nloaded from

Prevalence and distribution ofgastric metaplasia

tion of the gastric mucosa produces an activechronic inflammatory response.'2 Variouspathogenic mechanisms have been proposedfor H pylori"; all require a close proximity ofthe organism to the epithelium for epithelialdamage to occur. Hpylori only colonises gastrictype epithelium; therefore, the presence ofgastric epithelium in the duodenum is essentialfor duodenal infection by the organism. Wehave proposed that active duodenitis developswhen H pylori colonises foci of gastric meta-plasia in the duodenum, and that becausegastric metaplasia is considered to be aresponse to acid injury, this sequence of eventscan account for the independent role of acidandH pylori in duodenal inflammation.The previously suggested quantitative asso-

ciation between active duodenitis and gastricmetaplasia" 1 has been confirmed in this studyin which the more extensive the gastric meta-plasia, the more likely there was to be activeduodenitis. This observation can be explainedby the development of more extensive gastricmetaplasia in patients with duodenitis in re-sponse to the injury caused by the inflamma-tion itself. If gastric metaplasia provides the"foothold" in the duodenum for H pyloricolonisation, then a vicious circle may developby which H pylori infects foci of gastric meta-plasia and the resulting duodenitis leads toincreasing areas of gastric metaplasia suscept-ible to further H pylori infection. Duodenalulcer may supervene when the duodenitisbecomes severe, and the inflamed mucosa canno longer maintain its integrity against "acidattack", perhaps as a result of "high outputfailure" of epithelial regeneration.'5We have found that there is a lower pre-

valence of gastric epithelium in the duodenumin children than in adults. This is important forproving that this is acquired rather than con-genital in origin, as had been suggested byMarshall.'6 Parietal cells are usually present insmall numbers in the antral mucosa'7 and theirpresence in the duodenum of patients withulcers, as recently highlighted by Carrick,'8might therefore be expected in foci of gastricmetaplasia. We restrict the term "heterotopia"to fully organised gastric fundic tissue, which isa rare cause of duodenal nodularity and ispresumed to be congenital, and in our ex-perience and that of Shoushall not associatedwith duodenal ulcer disease.

Like Patrick et al we rarely found gastricmetaplasia beyond the first part of the duo-denum'9 and only in patients with extensivegastric metaplasia in the first part. There is awide variation in the extent of gastric meta-plasia within the first part ofthe duodenum andit seems to be randomly distributed around thecircumference, at least in patients withoutduodenal ulcer. Investigators studying bothduodenitis20 and gastric metaplasia2' in patientswith duodenal ulcer have found the abnor-mality to be more common at the margin of theulcers or at the site of a healed ulcer thanelsewhere. The findings from our study sug-gest that in the absence of duodenal ulcer thedistribution of gastric metaplasia depends onproximity to the pylorus and that it does not

occur preferentially in sites characteristic ofduodenal ulcer. Therefore, factors determiningsite of a chronic duodenal ulcer-for example,possible localised mucosal ischaemia-differfrom those which affect the distribution ofgastric metaplasia.The conventional view of gastric metaplasia

in the proximal duodenum is that it is anacquired change in response to acid injury.'Our results with respect to fasting gastric juicepH support this, although total acid secretorycapacity was not measured. The greater pre-valence of gastric metaplasia in men may resultfrom their greater acid secretory capacity,which would be consistent with the correlationbetween male gender and pH in our multi-variate analysis. There is additional support foran acid induced mechanism, from observationsof extensive gastric metaplasia in patients withZollinger-Ellison syndrome,22"2 the correlationof gastric epithelium with maximal acid out-put,'9 and from animal studies.2"26 These havebeen reviewed previously.27 Recent studies inrats by Tatsuta et al28 have clearly shown thatthe gastric metaplasia, which developed inresponse to chemically induced duodenalinjury in rats, was sigrlificantly more extensiveand prolonged in the animals rendered hyper-chlorhydric with tetragastrin, and did notoccur in a group of rats which had undergonevagotomy.

Metaplasia is a non-specific response tomucosal injury, and factors other than acidpeptic damage to the mucosa may result ingastric metaplasia in the duodenum. Twoprevious studies have shown an associationbetween heavy alcohol consumption andgastric metaplasia.29 0 This was not the case inour study, but our series included few heavydrinkers. Similarly, like Carrick et al 8 we havenot found support for the view that mucosalinjury as a result of cigarette smoking orNSAID consumption leads to increased gastricmetaplasia in the duodenum.

In this paper we have directed our attentionto gastric metaplasia in relation to its role induodenal ulcer pathogenesis. Its detection isalso of relevance in routine histopathology,because it is a characteristic of the pattern ofduodenal inflammation associated with pepticulcer disease. It also identifies those areas ofthebiopsy specimen surface where Hpylori can besought. Gastric metaplasia develops in a patchydistribution, and its detection thereforedepends on sampling error. In 62 of ourpatients this problem was addressed by takingmultiple biopsy specimens; we found that theprevalence of gastric metaplasia decreased withdistance from the pylorus, but was notdependent on the biopsy site radially in the firstpart of the duodenum, although it tended to bemore common in the anterior and roof biopsyspecimens. One biopsy specimen from theanterior wall (the position most readily biop-sied) would detect 22 out of 35 (63%) patientswith gastric metaplasia on multiple biopsy; twobiopsy specimens (anterior and roof) detected33 out of 35 (94%) patients with gastric meta-plasia. We agree with Frierson et al" that thealcian blue/periodic acid Schiffstain was useful

985

on February 8, 2020 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.43.12.981 on 1 Decem

ber 1990. Dow

nloaded from

Wyatt, Rathbone, Sobala, Shallcross, Heatley, Axon, Dixon

in highlighting the areas of gastric metaplasia,and we found good agreement between dif-ferent histopathologists in detecting thisfeature.

In summary, this study has examined theassociation between gastric metaplasia in theduodenum and other factors related to duo-denal ulcer disease. We found that foci ofgastric metaplasia in the duodenal epitheliumwere an acquired change and were more com-mon in men, perhaps because of their greateracid output. Gastric metaplasia was not relatedto alcohol, smoking, or NSAID consumption.Gastric metaplasia was not more common in Hpylori positive patients, but if present wasgreater in extent; this suggested that themucosal injury related to active duodenitis inthese patients may act as a further stimulus tometaplasia. Thus a positive feedback loop maybe set up whereby the area of duodenal mucosasusceptible to H pylori colonisation progres-sively increases, resulting in increasing inflam-mation until mucosal integrity is compromisedand ulceration occurs.

1 Blaser MJ. Gastric Campylobacter-like organisms, gastritis,and peptic ulcer disease. Gastroenterol 1987;93:371-83.

2 Graham DY. Campylobacter pylori and peptic ulcer disease.Gastroenterol 1989;96:615-25.

3 Joffe SN, Lee FD, Blumgart LH. Duodenitis. Clin Gastro-enterol 1978;7:635-50.

4 Wyatt JI, Rathbone BJ, Dixon MF, et al. Carnpylobacterpyloridis and acid-induced gastric metaplasia in thepathogenesis of duodenitis. J Clin Pathol 1987;40:841-8.

5 Thomas JM, Poynter D, Gooding C, et al. Gastric spiralbacteria. Lancet 1984;ii:100.

6 Wyatt JI, Rathbone BJ, Dixon MF, et al. Campylobacterpylori and development of duodenal ulcer. Lancet 1988;i:118-9.

7 Misiewicz JJ, Pounder RE. Peptic ulcer. In: Weatherall DJ,Ledingham JGG, Warrell DA, eds. Oxford textbook ofmedicine. 2nd ed. Oxford: Oxford University Press, 1987:12-64.

8 Whitehead R. Mucosal biopsy of thegastrointestinal tract. 3rded. Philadelphia: WB Saunders, 1984.

9 Grey SF, Wyatt JI, Rathbone BJ. Simplified techniques foridentifying gastric Campylobacter pyloridis on tissuesections. J Clin Pathol 1986;39:1279-80.

10 SAS/STAT User's Guide. Release 6.03 Edition. SASInstitute Inc, Cary, NC, USA, 1988.

11 Shousha S, Spiller RC, Parkins RA. The endoscopicallyabnormal duodenum in patients with dyspepsia: biopsyfindings in 60 cases. Histopathology 1983;7:23-34.

12 Gregory MA, Spitaels JM. Variations in the morphology ofvillous epithelial cells within 8 mm of untreated duodenalulcers. J Pathol 1987;53:109-19.

13 Rathbone BJ, Wyatt JI, Heatley RV. Possible pathogeneticpathways of Campylobacter pylori in gastro-duodenaldisease. Scand J Gastroenterol 1987;23(suppl 142):40-3.

14 Jenkins D, Goodall A, Gillett FR, et al. Defining duodenitis;quantitative histological study of mucosal responses andtheir correlations. J Clin Pathol 1985;38: 1119-26.

15 Bransom CJ, Boxer ME, Palmer KR, et al. Mucosal cellproliferation in duodenal ulcer and duodenitis. Gut 1981;22:277-82.

16 Marshall BJ, McGechie DB, Rogers PA, Glancy RJ. Pyloriccampylobacter infection and gastroduodenal disease. MedJAust 1985;142:439-44.

17 Hara M, Harasawa S, Tani N, Miwa T, Tsutsumi Y. Gastricmetaplasia in duodenal ulcer; histochemical considera-tions of its pathophysiological significance. Acta PatholJapan 1988;38:1011-8.

18 Carrick J, Lee A, Hazell S, Ralston M, Daskalopoulos G.Campylobacter pylori, duodenal ulcer, and gastric meta-plasia: possible role of functional heterotopic tissue inulcerogenesis. Gut 1989;30:790-7.

19 Patrick WJA, Denham D, Forrest APM. Mucous change inthe human duodenum; a light and electron microscopicstudy and correlation. Gut 1974;15:767-76.

20 Paoluzi P, Pallone F, Palazzesi P, et al. Frequency and extentof bulbar duodenitis in duodenal ulcer. Endoscopic andhistological study. Endoscopy 1982;14:193-5.

21 Marshall BJ, Warren JR, Blincow ED, et al. Prospectivedouble blind trial after eradication of Campylobacterpylori. Lancet 1988;ii:1437-41.

22 James AH. Gastric epithelium in the duodenum of a patientwith gastric hyperacidity. 2nd World Congress ofGastroenterology, Munich, 1962;H:540-3;(1963).

23 Parrish JA, Rawlins DC. Intestinal mucosa in Zollinger-Ellison syndrome. Gut 1965;6:286-9.

24 Gaskin RJ, Gad A, Barros AAJ, et al. Natural history andmorphology of secretagogue-induced ulcers in rats.Gastroenterol 1975;69:903-10.

25 Rhodes J. Experimental production of gastric epithelium inthe duodenum. Gut 1964;5:454-8.

26 Florey HW, Jennings MA, Jennings DA, et al. The reactionsofthe intestine ofthe pig to gastric juice. JPathol Bacteriol1939;49: 105-23.

27 Wyatt JI. Relationship of Campylobacter pylori to duodenalulcer disease. In: Blaser MJ, ed. Campylobacter pylori ingastritis and peptic ulcer disease. New York: Igaku-Shoin,1989.

28 Tatsuta M, Ishii H, Yamamura H, Yamamoto R, TaniguchiH. Enhancement by tetragastrin of experimental induc-tion of gastric epithelium in the duodenum. Gut 1989;30:311-5.

29 Lev R, Thomas E, Paul FF. Pathological and histo-morphometric study ofthe effects ofalcohol on the humanduodenum. Digestion 1980;20:207-13.

30 Shousha S, Barrison IG, El-Sayeed W, et al. A study ofincidence and relationship of intestinal metaplasia ongastric antrum and gastric metaplasia of duodenum inpatients with non-ulcer dyspepsia. Dig Dis Sci 1984;29:311-6.

31 Frierson HF, Caldwell SH, Marshall BJ. Duodenal bulbbiopsy findings for patients with non-ulcer dyspepsia withor without Campylobacter pylori gastritis. Modern Pathol1990;3:271-7.

986

on February 8, 2020 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.43.12.981 on 1 Decem

ber 1990. Dow

nloaded from