Gut, 1985, 26, 1226-1232

Gastric lesions in portal hypertension: inflammatorygastritis or congestive gastropathy?T T McCORMACK, J SIMS, I EYRE-BROOK, H KENNEDY, J GOEPEL,A G JOHNSON, AND D R TRIGER

From the University Departments of Surgery, Medicine and Pathology, Sheffield University, RoyalHallamshire Hospital, Sheffield

SUMMARY This paper reports the incidence and natural history of macroscopic gastritis in a seriesof 127 consecutive patients with portal hypertension of various aetiologies. Gastritis was

observed endoscopically in 65 patients (51%) and was of two main types. Twenty eight patientshad severe or persistent gastritis which caused clinically significant bleeding on 80 occasions andaccounted for 25% of the bleeds from all sources. The remainder had mild gastritis. The presence

of gastritis seemed to be independent of the severity of liver disease or the degree of rise ofwedged hepatic venous pressure and there was no difference in age, sex, or drugs prescribed inpatients with or without gastritis. The mean follow up period and the mean number ofsclerotherapy treatments was significantly greater (p<0.005) in patients with gastritis. Fullthickness gastric biopsies in seven surgical patients and 11 autopsy specimens showed dilated andtortuous submucosal veins. Endoscopic biopsies in 14 patients showed vascular ectasia in themucosal layer which was in excess of the degree of inflammatory infiltrate. Gastritis occurred inpatients with portal hypertension of all common aetiologies and the clinical and pathologicalevidence supports the contention that it reflects a congested gastric mucosa and should berenamed congestive gastropathy. As injection sclerotherapy improves survival from varicealbleeding congestive gastropathy may become more common. The response to conventional('anti-erosive') therapy is poor and measures aimed at reducing the gastric portal pressure may bethe only effective means of treating this condition.

Gastric mucosal lesions are common in portalhypertension. They are an important cause of bloodloss, which may be slow and insidious causinganaemia,1 or sudden and severe, causing massive

and occasionally fatal haemorrhage.2 The use ofsclerotherapy for bleeding oesophageal varices com-bined with regular endoscopic follow up has pro-vided a unique opportunity to study the progressionof changes occurring in the gastric mucosa. Wereport our clinical and endoscopic experience ofgastritis in portal hypertension over a four yearperiod.

Methods

PATIENTS

Definition of gastritisThe endoscopic changes which occur in the gastricAddress for correspondence: Dr D R Triger, Department of Medicine, RoyalHallamshire Hospital, Sheffield, S10 2JF.

Received for publication 24 January 1985

1226

mucosa were classified according to the description ofTaor et al.3 (i) a fine pink speckling or 'scarlatina'type rash, (ii) a superficial reddening, particularlyon the surface of the rugae giving a stripedappearance, (iii) a fine white reticular patternseparating areas of raised red oedematous mucosaresembling a 'snake skin'.These were included under the term 'mild gastri-

tis'.During the course of the study two additional

forms of gastric mucosal changes were noted: (i)discrete red spots analogous to the cherry red spotsdescribed in the oesophagus. These spots canbecome confluent giving a local area of severegastritis which may bleed, (ii) a diffuse haemorrha-gic gastritis.These were termed 'severe gastritis'.Gastritis has been arbitrarily defined as 'transient'

if present at one endoscopy and resolved by the nextendoscopy six to eight weeks later. If gastritis was

on 21 Septem

ber 2018 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.26.11.1226 on 1 Novem

ber 1985. Dow

nloaded from

Gastropathy in portal hypertension

present for more than eight weeks at two or moreconsecutive endoscopies it was said to be 'persis-tent'. Gastritis occurring within two weeks ofballoon tamponade was excluded as this could bebecause of contact irritation.Over the last four years (August 1979-August

1983) 127 patients with portal hypertension haveattended this unit. One hundred and fourteenpatients presented with bleeding and a further 13had oesophageal varices with no clinical evidence ofhaemorrhage. These were discovered during routineevaluation of their portal hypertension. The pre-sence or absence, and the degree of gastritis wasnoted at each endoscopy. All patients were endo-scoped at presentation by an experienced endoscop-ist and any patient with upper gastrointestinal tractbleeding underwent emergency endoscopy. Anypatients with oesophageal varices were treated byinjection sclerotherapy and followed up by regularcheck endoscopies.

PATHOLOGY TECHNIQUEBiopsies of the stomach were studied in 41 patientswith portal hypertension. These samples wereobtained (a) at endoscopy (23 patients - nine withmacroscopically normal mucosa and 14 with gastri-tis), (b) during surgical procedures (seven patients -four with bleeding gastritis and three with bleedinggastric varices) and (c) at necropsy (11 patients).Further operative samples were also taken duringrefashioning of an ileostomy and at resection of anoesophagojejunal anastomosis. Two patients withduodenitis also had endoscopic biopsies taken.Endoscopic biopsies and surgical specimens werefixed by immersion in formal saline. Post mortemspecimens were fixed by gently filling the unopenedstomach and oesophagus with formal saline andimmersing the whole specimen in formal saline.Paraffin sections were prepared and stained byhaematoxylin and eosin (H and E) or Miller's elasticstain counterstained by van Gieson's stain forcollagen (EVG).

Results

ENDOSCOPIC FINDINGS

Sixty five patients (51%) had macroscopic gastritisat some stage during follow up. The changes weremost commonly seen in the fundus and body of thestomach. There was no significant difference in ageor sex in patients with gastritis compared with therest (Table 1). Although there was a trend forgastritis to be more frequent in patients with lesssevere liver disease (as assessed by Child's grading)5this did not reach statistical significance(0-05<p<0-10). Gastritis occurred in portal

Table 1 Clinicalfeatures of65 patients with gastritiscompared with 62 patients with no gastritis

Patients (no)

Cause ofportal Without Withhypertension gastritis (62) gastritis(65)

Alcoholic cirrhosis 14 23Primary biliary cirrhosis 13 13Chronic active hepatitis 9 12Cryptogenic cirrhosis 10 4Portal vein thrombosis 7 5Others 9 8Child's grading A 21 31

B 15 17C 26 17

Age (Mean+SD) 57 5 (±12.1) 54 5 (±14.1)Sex Male 34 35

Female 28 30

hypertension of all common aetiologies and theproportion of patients who developed gastritis ineach of the aetiological groups was not significantlydifferent (X2 test).

Patients with gastritis had a significantly greater(p<0-005) number of sclerotherapy treatments perpatient (mean 4-0±0-3 SE) than those withoutgastritis (mean 2 3±0 2 SE). The follow up period inpatients with gastritis (mean 13-8 months ± 1.5 SE)was also significantly greater (p<0.005) than in thenon-gastritis group (mean 6*7 months ± 1-2 SE).Mild gastritis was noted in 37 of the 65 patients.

This was not of any clinical significance and in onlytwo patients did this progress to severe gastritis. Onthe other hand severe gastritis led to clinicalbleeding in almost all instances (see below). Gastri-tis both mild and severe was noted on initialpresentation in about one third of patients, anddeveloped during follow up in the remainder. Theage, sex, and Child's grading(s) was similar inpatients with mild transient gastritis and in thosewith severe or persistent gastritis (Table 2), but boththe mean follow up period and the number ofsclerotherapy treatments were significantly greater(p<0-0025 and p<0-005 respectively) in the pa-tients with severe or persistent gastritis. The meannumber of treatments per patient month, however,was similar in both groups.

BLEEDINGClinically significant bleeding from gastritis occur-red on 80 occasions in 29 patients, and bloodtransfusion (2-15 units) was required for 60 bleeds.Bleeds from gastritis accounted for 25% of the totalnumber of bleeds from all sources (Table 3).Gastritis was responsible for only nine of 114

1227

on 21 Septem

ber 2018 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.26.11.1226 on 1 Novem

ber 1985. Dow

nloaded from

McCormack, Simms, Eyre-Brook, Kennedy, Goepel, Johnson, and Triger

Table 2 Clinicalfeatures of65 patients with gastritis

Persistent or Mildsevere gastritis gastritis(28) (37)

Age mean±SE 52-0±2.5 (NS)* 56-9±2-8range (31-79) (21-84)

Sex male 15 20female 13 17

Follow-up (months)mean±SE 17-8±2-3 (p<0.0025)* 8-9+1-5range 1 day-4 years 1 day-2-2 years

Child's gradeA 14 17B 8 9C 6 (NS)t 11

Sclerotherapytreatment per patient

mean+SE 5-3±0-5 (p<0.005)* 2-8+0-3treatment per patient

month mean± 0-39±0-07 (NS)* 0-51±0-09

Students t test for unpaired numbers. t x test.

presenting bleeds, but after initial sclerotherapy, itaccounted for over one third of rebleeding episodes.Fifty one bleeds from gastritis occurred in thepresence of thrombosed or obliterated oesophagealvarices. In the presence of patent oesophagealvarices, bleeding from gastritis was diagnosed only ifit-was seen to be coming from an area of gastritis andno other source of haemorrhage was identifiable.

WEDGED HEPATIC VENOUS PRESSUREWedged hepatic venous pressure (WHVP) wasmeasured in 18 patients with oesophageal varices

Table 3 Sites ofpresenting bleeds in 114 patients and 202subsequent bleeds in 48 patients who rebled

Site of bleeding Presenting bleed Subsequent bleed

Oesophageal varices 88 64Gastritis 9 71Upper GI tract(unknown site) - 13

Post injection slough - 16Gastric varices 6 11Peptic ulcer 4 3Duodenitis - 3Oesophagogastric junction - 14Mallory Weiss 1 1Rectal varices 3 2Oesophagojejunal

anastomosis 1 4Intra-abdominal 1 -

Ileostomy 1 -

114 202

who had not bled before the study. There was nosignificant difference in the mean pressure in the 12with gastritis (mean WHVP 17-3 mmHg ± 1P6 SE)compared with those who had no gastritis (meanWHVP 16*0 mmHg ± 1-6 SE).

RELATION OF MEDICATION TO GASTRITISSeventy patients were receiving diuretics, 11 pa-tients penicillamine, six patients corticosteroids and13 patients were being treated for diabetes. Patientsreceiving these drugs were equally distributed be-tween the gastritis and non-gastritis groups.

EFFECT OF SCLEROTHERAPY ON GASTRITISIn six patients gastritis appeared for the first timeafter successful sclerotherapy while in another fivepatients the obliteration of oesophageal varicesappeared to coincide with the disapppearance ofgastritis.

PATHOLOGICAL CHANGESThe most consistent finding was dilatation of thesubmucosal veins which were tortuous and irregularin diameter. Elastic staining emphasised the irregu-larity of the vein wall and showed foci of intimalthickening (Fig. 1). These features were visible inveins sampled from all areas of the stomach butwere particularly prominent in the proximal part ofthe gastric body and cardia. The mucosal vesselsshowed focal areas of abnormality. These consistedof ectasia of capillaries and veins (Fig. 2) which attimes formed a leash of mucosal vessels. In somecases this could be seen to overlie areas of abnormalsubmucosal veins. Although chronic gastritis ascharacterised by mononuclear infiltrate did occa-sionally occur, the vascular changes were usually inexcess of that expected with the degree and activityof gastritis. We have termed these changes 'conges-tive gastropathy'.

In the jejunal and ileal surgical samples andduodenal biopsies there was also marked vascularectasia in the mucosa, (Fig. 3) with no evidence ofan inflammatory infiltrate.

ENDOSCOPIC SPECIMENSBiopsies were not done in endoscopically severegastritis for fear of inducing major haemorrhage.Classic histological features of chronic inflammatorygastritis were seen in four of 14 patients, while theother 10 had vascular ectasia with little or noinflammatory infiltrate. Six of the nine biopsies frommacroscopically normal mucosa were histologicallynormal, the remainder showing vascular ectasia. Itshould be remembered, however, that endoscopicbiopsies are of necessity small and superficial andtherefore liable to sampling error.

1228

on 21 Septem

ber 2018 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.26.11.1226 on 1 Novem

ber 1985. Dow

nloaded from

Gastropathy in portal hypertension

Fig. 1 Dilated submucosal veins (arrows to walls) with intimal thickening (arrow heads). (Post mortem stomach, elasticvan Gieson, x8 original magnification).

SURGICAL SPECIMENSThese were taken from patients with severe gastritisor bleeding gastric varices and all showed mucosaland submucosal vascular abnormalities.

AUTOPSY SPECIMENSMarked submucosal vascular ectasia was seen in allcases, but mucosal vascularity could not be assessedbecause of mucosal post mortem autolysis.The correlation between the macroscopic appear-

ances and histological findings are summarised inTable 4.

Discussion

In this series, gastritis has been observed at sometime in 51% of patients with oesophageal varicesand it has been responsible for 25% of the totalnumber of bleeds. Other workers have noted thatbleeding in portal hypertension can be caused bygastritis in 30-40% of cases.678 Gastric mucosalchanges were seen in portal hypertension of all

aetiologies and occurred in patients with both patentand obliterated varices.

It should be emphasised that, with the exceptionof the cherry red spots, the gastritis described here ismacroscopically identical to that seen in patientswithout portal hypertension, but the histologicalappearance is quite different from that described by

Table 4 Summary ofmacroscopic and histologicalfindings in gastric specimens

Type of No Macroscopic Microscopicspecimen appearance appearance

Endoscopic 9 Normal 6 normalbiopsy 3 mucosal ectasia

Endoscopic 14 Mild gastritis 4 inflammatorybiopsy 10 mucosal ectasia

4 severe gastritis All mucosal andSurgical 7 3 bleeding gastric submucosal

varices ectasiaAutopsy 11 All submucosal

ectasia

1229

on 21 Septem

ber 2018 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.26.11.1226 on 1 Novem

ber 1985. Dow

nloaded from

McCormiack, Simms, Eyre-Brook, Kennedy, Goepel, Johnson, and Triger

Fig. 2 Gastric antral biopsy showing prominent dilated vessels (arrows) near the surface. (PAS x 160 originalmagnification).

Fig. 3 Duodenal biopsi showing marked vascular ectasia (arrows) particularly in the villi. Acute inflammatory changesare absent. (H & E x20 original magnification).

1230

on 21 Septem

ber 2018 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.26.11.1226 on 1 Novem

ber 1985. Dow

nloaded from

Gastropathy in portal hypertetnsion

Whitehead and colleagues9 where chronic inflam-matory cells predominate and there is a strongassociation with gastric ulcer. Our observations aresimilar to that of Brown and colleagues'( who foundno evidence of an increased incidence of classicalchronic gastritis in a series of cirrhotics, despite thepresence of gastric ulcers and erosions in nearly 20%of their patients. Furthermore, they comment thatgastric erosions tended to occur in histologicallynormal mucosa. Their study, however, did notinclude assessment of the mucosal and submucosalvascular channels.

Gastritis in portal hypertension might be caused byseveral factors. Alcohol can be excluded because wehave observed the changes in non-alcoholic cirrhosisas frequently as in alcoholic cirrhosis. The observa-tion of severe gastritis in two patients who hadundergone truncal vagotomies makes it unlikely thatgastric acid plays a major role.The most important element causing gastritis may

be the raised portal pressure itself. Obstruction ofthe venous drainage from the stomach can inducechanges in the gastric mucosa. Palmer, in 195711induced portal hypertension in dogs by portal veinligation and found that both the mucosal andsubmucosal veins in the stomach wall becamedilated. Both he and Sandblom'2 observed similarchanges in gastric biopsies from patients with portalhypertension. Alternatively gastritis might be be-cause of gastric mucosal ischaemia secondary toarteriovenous shunting which can be demonstratedin the stomachs of both animals13 and humans14 withportal hypertension.The histological changes are entirely consistent

with an increase in venous pressure producing acongested gastric mucosa. The occurrence andseverity of this congestive gastropathy may depend,however, not only on the total level of portalpressure but also on local blood flow characteristicswhich may or may not transmit this increasedpressure to the gastric mucosal and submucosalveins. Differences in local blood flow patterns mayexplain why some patients develop gastropathy andothers do not. Successful sclerotherapy ofoesophageal varices may induce local changes inblood flow patterns and if this results in an increasedvenous pressure in areas proximal to the site ofthrombosis, congestive gastropathy might be pre-dicted.The mean number of sclerotherapy treatments in

patients with gastropathy was significantly greaterthan in those without it. Sclerotherapy appears toincrease long term survival15 and the greater inci-dence of the gastric lesion in the longer survivorsmay be related either to the progression of diseaseor to the greater number of sclerotherapy treat-

ments in these patients. Our experience suggeststhat while gastropathy may develop after sclerother-apy in some individuals the converse is true inothers. In theory obstruction of blood flow at thegastro-oesophageal junction may increase conges-tion by blood vessels flowing from the stomach. Thiseffect is likely to be very variable as recent studiesusing Doppler ultrasound'6 have shown that bloodflow in oesophageal varices may sometimes betowards the stomach and therefore thrombosis ofthese varices would reduce and certainly not inducecongestion in the gastric mucosa. In the majority ofpatients, therefore, the relationship between sclero-therapy and gastropathy is not straightforward andthe presence of the gastric lesion is probablyindependent of the patency of oesophageal varices.

THERAPEUTIC IMPLICATIONSCongestive gastropathy was mild and transient inmore than half of the patients. Complications didnot occur and progression was uncommon duringthe short period of follow up. In contrast, patientswith severe or persistent gastropathy are prone toclinically significant haemorrhage.Twenty five of our patients with severe changes

received H2 receptor antagonists, five receivedsucralfate and all were prescribed antacids at somestage. None of these agents had any significanteffect upon either the gastritis or the bleeding. Thistends to support the hypothesis that it is congestionrather than erosion which is the major factordamaging the gastric mucosa. The rational approachto treatment is therefore a reduction of the portalvenous pressure which should thus reduce thecongestion in the gastic mucosa.2.

Portosystemic shunting effectively reduces portalpressure and bleeding from gastritis is rare after thisprocedure.2 17-19 An alternative surgical approach isto reduce the gastric blood flow alone by devascular-ising the upper two thirds of the stomach andcombining it with an oesophageal transection. Wehave used this procedure in eight patients, specific-ally for controlling severe haemorrhagic gastritis andit has been successful in five. As with portosystemicdecompression this procedure is not without con-siderable risk in patients with advanced liverdisease.Methods aimed at reducing portal blood pressure

or flow by pharmacological means would clearly bepreferable to surgery. Although Lebrec and col-leagues20 reported that propranolol reduced varicealhaemorrhage, close inspection of their paper revealsthat the drug also reduced the incidence of gastritis.On an anecdotal basis we have observed significantimprovement in several patients with severe persis-

1231

on 21 Septem

ber 2018 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.26.11.1226 on 1 Novem

ber 1985. Dow

nloaded from

1232 McCormack, Simms, Eyre-Brook, Kennedy, Goepel, Johnson, and Triger

tent gastritis treated with propranolol and we arecurrently evaluating its effectiveness by means of acontrolled trial.

In conclusion, the macroscopic gastritis noted inpatients with portal hypertension differs from thatseen in the absence of liver disease in that (a) thehistological appearance is quite distinct (b) it isunrelated to the aetiology of the portal hypertension(c) it does not respond to conventional anti-inflammatory drug therapy and (d) the histologicalchanges are found elsewhere in the gastrointestinaltract. In the light of the clinical, haemodynamic andhistological observations the term 'congestivegastropathy' appears to be more appropriate.

We wish to thank those involved with the care ofthese patients. Sister Salisbury and the staff of theendoscopy unit, Sisters Ellis and Barry, and the staffin the operating theatres.

References

1 Wenger J, Huffman RT, Landy MS. Persistent bloodloss from the stomach of patients with cirrhosis andoesophageal varices. South Med J 1970; 63: 560-6.

2 Sarfeh IJ, Juler GL, Stemmer EA, Mason GR. Resultsof surgical management of haemorrhagic gastritis inpatients with gastroesophageal varices. Surg GynecolObstet 1982; 155: 167-70.

3 Taor RE, Fox B, Ware J, Johnson AG. Gastritis -gastroscopic and microscopic. Endoscopy 1975; 7:209-15.

4 Japanese Research Society for Portal Hypertension.The general rules for recording endoscopic findings onoesophageal varices. Jap J Surg 1980; 10: 84-7.

5 Pugh RNH, Murray-Lyon IM, Dawson JL, PietroniMC, Williams R. Transection of the oesophagus forbleeding oesophageal varices. Br J Surg 1973; 60:646-9.

6 Dagradi AE, Mehler R, Tan DT, Stempien SJ. Sourcesof upper gastrointestinal bleeding in patients with livercirrhosis and large esophagogastric varices. Am JGastroenterol 1970; 54: 458-63.

7 Thomas E, Rosenthal WS, Rymer W, Katz D. Uppergastrointestinal haemorrhage in patients with alcoholicliver disease and oesophageal varices. Am J Gastro-enterol 1979; 72: 623-9.

8 Khodadoost J, Glass GBJ. Erosive gastritis and acutegastroduodenal ulcerations as source of upper gastro-intestinal bleeding in liver cirrhosis. Digestion 1972; 7:129-38.

9 Whitehead R, Truelove SC, Gear MWL. The histo-logical diagnosis of chronic gastritis in fibreopticgastroscope biopsy specimens. J Clin Pathol 1972; 25:1-11.

10 Brown RC, Hardy GJ, Temperley JM, MiloszewskiKJA, Gowland G, Losowsky MS. Gastritis and cir-rhosis - no association. J Clin Pathol 1981; 34: 744-8.

11 Palmer ED. Erosive gastritis in cirrhosis. Am J Dig Dis1957; 2: 31-6.

12 Sandblom P. The source of bleeding in portal hyper-tension. Bull Soc Int Chin 1975; 3: 165-7.

13 Manabe T, Suzuki T, Honjo I. Changes of gastricblood flow in experimentally induced cirrhosis of theliver. Surg Gynecol Obstet 1978; 147: 753-7.

14 Hashizume M, Tanaka K, Inokuchi K. Morphology ofgastric microcirculation in cirrhosis. Hepatology 1983;6: 1008-12.

15 MacDougall BRD, Westaby D, Theodossi A, DawsonJL, Williams R. Increased long-term survival invariceal haemorrhage using injection sclerotherapy.Results of a controlled trial. Lancet 1982; 1: 124-7.

16 McCormack T, Smallwood RH, Walton L, Martin T,Robinson P, Johnson AG. Doppler ultrasound probefor assessment of blood-flow in oesophageal varices.Lancet 1983; 2: 677-8.

17 Jackson FC, Perrin EB, Felix WR, Smith AG. Aclinical investigation of the portacaval shunt: V. sur-vival analysis of the therapeutic operation. Ann Surg1971; 174: 672-701.

18 Resnick RH, Iber FL, Ishihara AM, Chalmers TC,Zimmerman H. A controlled study of the therapeuticportacaval shunt. Gastroenterology 1974; 67: 843-57.

19 Rueff B, Prandi D, Degos F et al. A controlled study oftherapeutic portacaval shunt in cirrhosis. Lancet 1976;1: 655-9.

20 Lebrec D, Poynard T, Hillon P, Benhamou JP.Propranolol for prevention of recurrent gastrointestinalbleeding in patients with cirrhosis. N Engl J Med 1981;305: 1371-4.

on 21 Septem

ber 2018 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.26.11.1226 on 1 Novem

ber 1985. Dow

nloaded from