1162 Gut 1993;34: 1162-1166 Duodenal histology, ulceration, and

pylori in the presence or absence of non-steroidal Gut: first published as 10.1136/gut.34.9.1162 on 1 September 1993. Downloaded from anti-inflammatory drugs

A S Taha, S Dahill, I Nakshabendi, F D Lee, R D Sturrock, R I Russell

Abstract lium allowing H pylori to colonise the duodenal Duodenitis and gastric metaplasia, which is mucosa, where it produces an acute inflamma- often colonised by (H tory response.3 Most studies in this field have pylori), are increasingly recognised for their been carried out in patients not taking non- importance in the pathogenesis of duodenal steroidal anti-inflammatory drugs (NSAIDs), ulcers. The situation is not clear in patients and the role of duodenitis and gastric metaplasia receiving non-steroidal anti-inflammatory in mediating NSAID related damage has, there- drugs (NSAIDs), who have a higher risk of fore, remained unclear. peptic ulceration. The aim of this study was to We aimed at studying the histology of the identify the duodenal histological abnormali- duodenal mucosa in the presence or absence of ties in the presence or absence of NSAIDs, chronic NSAID intake, H pylori, and duodenal H pylon, and duodenal ulceration. Endo- ulcers. scopic duodenal biopsy specimens were taken from healthy looking mucosa of 172 patients (74 took NSAIDs, and 98 did not). Duodenitis Patients and methods was graded according to the degree of neutro- Patients, aged 18 years or over, were recruited philic and plasma cell infiltration, villus from the Rheumatology and the Gastro- height, Brunner's gland prolapse, and gastric enterology outpatient clinics, provided they had metaplasia. The activity of duodenitis was no malignancy or previous history of gastric or dependent on the neutrophilic infiltration. A duodenal surgery. NSAIDs had to be taken for a global score covering all the above factors was minimum of four weeks before . constructed, and H pylon in both the stomach Patients were excluded if they had taken ulcer and , was also assessed. The results healing agents, , or cytotoxic drugs showed that duodenitis with varying degrees of within one week of endoscopy. Informed con- neutrophilic infiltration and gastric metaplasia sent was obtained, and endoscopy performed http://gut.bmj.com/ was found in 20 patients (27%) taking NSAIDs, using 3-7 mg midazolam intravenously for compared with 56 patients (57%) not taking sedation. NSAIDs (X2=l6 24, p<0-001). This degree of At endoscopy, an average of two biopsy speci- duodenitis was also found in 20 of 25 patients mens were taken from the anterior and posterior (80%) with duodenal ulcers, regardless of walls of the first part of the duodenum, and NSAID intake (X2=15-38, p<0.001). Gastric immediately fixed in a solution of 10% formalin,' metaplasia was identified in 20 patients (27%) saline. The duodenal histological findings were on October 2, 2021 by guest. Protected copyright. receiving NSAIDs and 38 (39%) not receiving reported as follows: NSAIDs. Duodenal H pylori was only seen Normal - with villus to crypt ratio greater than in patients with gastric metaplasia 10 (50%) 2:1, in the presence or absence of Brunner's receiving NSAIDs, and 34 (89%) not receiving glands above the muscularis mucosa, but with- NSAIDs. H pylori positive , and the out any increase in the numbers of inflammatory combination of active duodenitis and gastric cells. metaplasia were independent predictors of Chronic duodenitis - in the presence ofmucosal duodenal ulceration. It is concluded that active mononuclear cell infiltration and epithelial duodenitis is less common in patients taking damage. NSAIDs, but is strongly associated with Active chronic duodenitis - this was diagnosed gastric metaplasia, H pylori positive gastritis, whenever intramucosal neutrophils were seen, in Departments of , and duodenal ulceration. These findings are the presence or absence of gastric metaplasia, A S Taha relevant to the pathogenesis and treatment of and especially in the background ofmononuclear I Nakshabendi duodenal ulcers in patients taking NSAIDs. cell infiltration. R I Russell (Gut 1993; 34: 1162-1166) The severity ofthe above findings (duodenitis, Pathology, villus height, and Brunner's gland prolapse) was S Dahill graded on a 0-3 scale: 0, normal; 1, mild; 2, F D Lee Gastritis and duodenitis have, for a long time, moderate; and 3, severe. and Rheumatology, been thought to play a part in the pathogenesis of The extent ofgastric metaplasia was assessed as Royal Infirmary, Glasgow , and this concept has been follows: 0, absent; 1, present but consisting of R D Sturrock further consolidated by the recognition of less than five consecutive cells; 2, five consecu- Correspondence to: Dr A S Taha, Helicobacter pylon (H pylon).' 2 Subsequently, it tive cells; and 3, greater than five consecutive Gastroenterology Centre, has been suggested that gastric metaplasia in the cells, seen in 1-3 sections cut at three levels in Southern General Hospital, Glasgow G5 1 4TF. duodenum and H pylori associated gastritis each biopsy specimen. Accepted for publication might be synergistic in the pathogenesis of A global score covering the overall totals of the 5 January 1993 duodenitis, with the metaplastic gastric epithe- scores of all the above assessments was con- NSAID and the duodenum 1163

TABLE I Demographic and endoscopic details ofthe study Gastritis was classified according to the groups Sydney System, which also covers chemical and Receiving Not receiving lymphocytic gastritis.5 8 NSAID NSAID Statistical analyses included the X2 test, Gut: first published as 10.1136/gut.34.9.1162 on 1 September 1993. Downloaded from No 74 98 analysis ofvariance, multiple and logistic regres- Men 24 38 sion where appropriate. All specimens carried Women 50 60 Age (y), median (interquartile 57 (48-65) 52 (40-61) code numbers, and the endoscopist and patholo- range) gists were not aware of any treatment being Smokers 27 56 Drinkers 28 44 received by patients. The histological scoring Abdominal symptoms 43 83 was performed by the two pathologists Gastric ulcers 17 2 Duodenal ulcers 11 14 separately in order to assess reproducibility. Gastric erosions 7 Results structed: in the absence of neutrophils, patients A total of 172 patients were studied: 74 were scoring four or less were considered normal; receiving NSAIDs, and 98 were not. Table I those scoring five or more were considered to shows the demographic data of the patients. have chronic duodenitis, in the absence of Patients ofboth groups were comparable in their neutrophils. Chronic active duodenitis was diag- ages and in their smoking and drinking habits. nosed when intramucosal neutrophils were seen Patients treated with NSAIDs had more gastric in patients scoring four or more. but a similar number ofduodenal ulcers as those The presence or absence of H pylori in the not taking NSAIDs. None of the study subjects duodenal mucosa was also assessed using cresyl received any drug prohibited by the study violet stain, and by culture of single duodenal protocol within six months of endoscopy, apart biopsy specimens. Detection of H pylori by from four patients (not on NSAIDs), who took either histology or culture was considered as antibiotics for chest infection, two to three indicative of active infection in the duodenum. months before recruitment. Also, abdominal Gastric antral biopsy specimens were also complaints in most of the symptomatic patients taken to check for the presence of H pylori not taking NSAIDs have settled once their by histology (2 specimens) and culture (1 endoscopy was reported normal: irritable bowel CTNAritPiAn) -QC nrNIMTr%llichy A-c,rri%AA 4 (or stomach) syndrome, or non-ulcer dyspepsia vjIellalCll)3 as previously Uesc.iUeu. cannot be excluded in these patients. The Figure shows the duodenal global histo- logical scores. The histological scores of one x2= 16-24, p < 0-001 pathologist were reproduced by his other fellow I I~~~~~~~~~~~~~~pathologist in more than 90% of specimens, 13 0 00 regardless of NSAID intake. Patients in the http://gut.bmj.com/ NSAID group had fewer cases of active chronic 12 - 0 duodenitis. There was no significant difference 00°0°00 in the number of patients with chronic duo- 11 0 denitis (11 receiving NSAIDs and 10 not receiv- 000 ing NSAIDs). Table II describes the with active 10 patients 000000 00000 chronic duodenitis. The two groups were similar on October 2, 2021 by guest. Protected copyright. with respect to the number of smokers, patients 0 00000 0 9. * 0 00000 with gastric H pylont, gastric metaplasia, and to duodenal ulcers. Grade 3 (heavy) neutrophilic 8- 0000 0000 4 so and plasma cell infiltration, however, was more prevalent in the absence of NSAIDs. It is also a) 0 7.- * 0 00 worth noting that patients with active chronic 0 duodenitis in both groups had a total of 20 of 25 c;C') duodenal ulcers in all u0 6- (80%) diagnosed patients (U) * 000 00°000 a) (X2=l15385 p<0 001 v patients without active 0 duodenitis). Duodenal H pylori was found in 5. 00*0. 0 000000 4 (9) only 50% ofpatients taking NSAIDs with active duodenitis, and 61% of their counterparts not 0 @0000 4. 00000 0 00000 es h p TABLE II Characteristics ofpatients with active chronic 3 * Neutrophils present 0@'*4 duodenitis Receiving Not receiving 00000 NSAID NSAID (n= 74)(%) (n= 98) (%) 1-. 0@@O@ No 20 (27) 56 (57)* Smokers 12 (60) 37 (66) Gastric H pylori 20 (100) 43 (77) Duodenal H pylori 20 (100) 43 (77) 0- Gastric metaplasia 14 (70) 34 (61) Receiving NSAIDs Not receiving NSAIDs Heavy neutrophilic infiltration 4 (20) 17 (30)** Heavy mononuclear infiltration 2 (10) 12 (21)** The duodenal global histological scores in the presence or absence ofNSAIDs. Patients with Duodenal ulcers 6 (30) 14 (25) intramucosal neutrophils have active chronic duodenitis. The remainder (black dots) have inactive chronic , and their numbers were similar in the two study groups. Significant rise: *p=<0 001; **p=<005. 1164 Taha, Dahill, Nakshabendi, Lee, Sturrock, Russell

TABLE III Details ofpatients with gastric metaplasia in the years, and second line two duodenum drugs years. Indomethacin, naproxen, and diclofenac were Receiving Not receiving the most widely used NSAIDs. Patients taking NSAID NSAID diclofenac, ketoprofen or nabumetone were less Gut: first published as 10.1136/gut.34.9.1162 on 1 September 1993. Downloaded from (n= 74)(%) (n =98)(%) likely to have active duodenitis and gastric No 20 (27) 38 (39) metaplasia than other NSAIDs but the differ- Grade: 1 4 (20) 6 (16) 3 16 (80) 32 (84) ences were not statistically significant. Also, the Smokers 12 (60) 24 (62) intake of second line drugs did not seem to Gastric Hpylori 14(70) 34 (89) Duodenal Hpylori 10 (50) 34 (89) influence the development of active duodentitis Active chronic duodenitis 14 (70) 34 (89) and gastric metaplasia, although such histo- Heavy neutrophilic infiltration 4 (20) 15 (39) Heavy mononuclear infiltration 1 (5) 8 (24) logical abnormalities, as well as gastric H pylori, Duodenal ulcers 6 (30) 10 (26) were less common in patients treated with gold injections.

taking NSAIDs; all cases of duodenal H pylori Discussion coexisted with gastric metaplasia. This study shows that active chronic duodenitis Table III shows the characteristics of patients is less common in patients treated with NSAIDs, with gastric metaplasia; although there were and that the prevalence of gastric metaplasia in relatively fewer cases of metaplasia in patients the duodenum is not significantly reduced by taking NSAIDs, the differences were not chronic NSAID use. The study also shows a statistically significant. Also, patients not treated strong association between duodenal ulcers, with NSAIDs were more likely to have heavy active duodenitis, gastric metaplasia, and neutrophilic and mononuclear cell infiltration. H pylon related gastritis regardless of NSAID The two groups were almost identical, however, intake. in the extent of metaplasia, the number of Active chronic duodenitis in patients not smokers, patients with gastric H pylon, and taking NSAIDs was more commonly diagnosed duodenal ulcers. Similar to patients with active in our study than in that reported by some chronic duodenitis, most of the patients with workers,3 who studied this abnormality in duodenal ulcers (16/25, 64%) were also found to patients with non-ulcer dyspepsia. This can be have gastric metaplasia (x2= 14-00, p<0 001). explained by the absence ofulcers in their study, Table IV shows the prevalence of active duo- and by the nature of our scoring system. Such denitis and gastric metaplasia in the presence or differences could not have influenced the inter- absence of the various types of gastritis. Most pretation of our findings with respect to the cases were found in association with H pylori effect of NSAID on duodenal histology, because positive gastritis, regardless of NSAID intake, of two main reasons. Firstly, the same global although H pylorn was less common in patients score was applied to biopsy specimens taken http://gut.bmj.com/ receiving NSAID. Active duodenitis was from all our patients, regardless of NSAID uncommon in patients with normal gastric intake, under blinded conditions. Secondly, and histology: its prevalence varied between 20% (on allowing for the possibility that our global score NSAIDs) and 24% (not on NSAIDs). It is also might have detected minor increases in the worth noting that 24 of 25 duodenal ulcers mucosal neutrophils in some cases ofmild active (96%), found in all patients, were associated with duodenitis, the number of patients with heavy active antral gastritis, and that gastric H pylori neutrophilic infiltration (severe active duo- on October 2, 2021 by guest. Protected copyright. was identified in 23 of duodenal ulcer patients denitis) was also greater in the absence of (92%). Logistic regression showed that gastric NSAIDs. Hpyloni and the combination ofactive duodenitis Gastric metaplasia tended to occur less fre- and gastric metaplasia were independent pre- quently in our patients receiving NSAIDs, but dictors of duodenal ulceration. On the other the differences did not reach statistical signific- hand, only 11 (nine on NSAIDs) of 16 gastric ance. This, together with its close association ulcers (69%) had H pyloni related gastritis; the with active duodenitis, is in agreement with the remaining five gastric ulcers were found in findings of two other studies.9' High acidity of association with chemical gastritis, which was the duodenal contents has for a long time been found only in patients taking NSAIDs. found to be associated with gastric metaplasia, Table V shows the distribution of NSAIDs, both in humans39 '1 'I and in laboratory animals.'3 with or without second line drugs in patients Acute exposure to NSAIDs is also known to with active duodenitis and gastric metaplasia. stimulate gastric acid secretion,''7 which, at The median duration ofNSAID intake was three least in theory, should increase the prevalence of

TABLE IV Active chronic duodenitis and gastric metaplasia in patients with or without gastritis Receiving NSAID (n= 74) Not receving NSAID (n 98) Gastric Active Gastric Gastric Active Gastric No H pylori duodenitis metaplasia No H pylori duodenitis metaplasia Chronic gastritis: Active (type B) 48 31 9 13 63 54 46 33 Atrophic 3 - 2 2 7 1 2 1 Special forms: Reactive (chemical) 10 - 4 1 - _ _ _ Lymphocytic 3 3 3 2 7 5 3 2 Normal 10 - 2 2 21 2 5 3 NSAID and the duodenum 1165

TABLE V Details ofNSAIDs and second line drugs in apply to gastric ulcers, eight of which (8/14, patients with active chronic duodenitis and gastric metaplasia 57%) were positive for H pylorn.8 Gastric Active Gastric The relative rarity of active duodenitis in the No H pylori duodenitis metaplasia NSAID group might explain, at least in part, Gut: first published as 10.1136/gut.34.9.1162 on 1 September 1993. Downloaded from Indomethacin 13 6 5 4 why duodenal ulcers occur less commonly than Naproxen 12 6 5 5 gastric ulcers in such patients.2' 22 On the other Diclofenac 7 2 - - Ketoprofen 6 3 - 1 hand, duodenal ulcers were still seen in compar- Nabumetone 6 5 2 1 able proportions in the presence or absence of Ibuprofen 5 2 1 2 Others* 24 10 5 6 NSAIDs, which could be because of other Sulphasalazine 18 14 12 8 mechanisms ofNSAID toxicity such as suppres- Gold (im) 13 5 5 2 sion of mucosal prostaglandins. Penicillamine 6 2 3 1 The prevalence of gastric H pylori, active *These included four patients or less for each of the following duodenitis, and gastric metaplasia was similar in agents: azapropazone, fenbufen, flurbiprofen, piroxicam, etodalac, tiaprofenic acid, and sulindac. patients taking the various NSAIDs with the exception of diclofenac and ketoprofen. The prevalence of such findings tended to be lower in patients treated with gold injections than in those gastric metaplasia. The second, however, has not receiving sulphasalazine and NSAIDs.24 A firm been seen in our study or in others9'1 and could statement could not, however, be made because be related to the possibility that gastric acid ofthe relatively small numbers ofpatients taking secretion might not necessarily be raised when the individual NSAID with or without second NSAIDs are taken on longterm basis, three years line drugs in this study. in this study. In conclusion, despite the relative lack of Active chronic duodenitis was also less active duodenitis in NSAID patients, a strong common in our patients taking NSAIDs. The association exists between duodenal ulcers, reason for this is not clear but it could be active duodenitis, gastric metaplasia, and explained by the finding of a lower prevalence of H pylori positive gastritis regardless of NSAID H pylori in NSAID patients,8 18-20 and by the intake. This might be relevant to the under- tendency of NSAID related damage to be standing ofthe pathogenesis and the treatment of maximal in the gastric antrum.21 22 Although no duodenal ulcers in chronic NSAID users. The specific histological picture could be shown in potential benefits of eradicating H pylori in such the duodenal mucosa of NSAID patients, the patients are, however, still speculative, and their relative lack of heavy neutrophilic and plasma value remains to be proved. cell infiltration in the duodenum is reminiscent of some aspects of chemical gastritis.823 In the We wish to thank Miss Pamela Boothman for her help in H pylori absence of gastroduodenal surgery, the last is cultures, and Mrs Ruth Simpson for her secretarial assistance. mostly found in chronic NSAID users, as shown 1 Dixon MF, Sobala GM. Gastritis and duodenitis: the histo- http://gut.bmj.com/ in this study and others.6823 pathological spectrum. EurJ3 Gastroenterol Hepatol 1992; 4 (suppl 2): S17-23. Patients with active duodenitis had compar- 2 Sipponen P. Long-term consequences of gastroduodenal able proportions of cases with gastric H pylori, inflammation. EurJ Gastroenterol Hepatol 1992; 4 (suppl 2): S25-9. regardless of NSAID intake, and despite the 3 Wyatt JI, Rathbone BJ, Dixon MF, Heatley RV. Campylo- differences in the overall prevalence of H pylon bacter pyloridis and acid induced gastric metaplasia in the pathogenesis ofduodenitis. J Clin Pathol 1987; 40: 841-8. in the presence or absence of NSAID. This 4 Taha AS, Boothman P, Holland P, et al. Gastric mucosal emphasises the role ofHpylori in the pathogene- prostaglandin synthesis in the presence of Campylobacter on October 2, 2021 by guest. Protected copyright. 1 pylori in patients with gastric ulcers and non-ulcer dys- sis of active duodenitis. It is worth noting that pepsia. AmJ Gastroenterol 1990; 85: 47-50. in the presence ofgastric metaplasia Hpylori was 5 Misiewicz JJ, Tytgat GNJ, Goodwin CS, et al. The Sydney System: a new classification of gastritis. Working Party almost as reliably isolated from the duodenum as Reports ofthe World Congress ofGastroenterology. Oxford: from the gastric mucosa and this reflects the Blackwell Scientific Publications, 1990: 1-10. 6 Dixon MF, O'Connor HJ, Axon ATR, King RFJG, Johnston natural history ofthis infection, being dependent D. Reflux gastritis: distinct histopathological entity? 7 Clin on the presence ofgastric type epithelium. Pathol 1986; 39: 524-30. 7 Dixon MF, Wyatt JI, Burke DA, Rathbone BJ. Lymphocytic Patients receiving NSAIDs had 11 duodenal gastritis - relationship to Campylobacter pylori infection. ulcers 11/74 (15%) compared with 14/98 (14%) J3Pathol 1988; 154: 125-32. 8 Taha AS, Nakshabendi I, Lee FD, Sturrock RD, Russell RI. in those not receiving NSAIDs. Similar numbers Chemical and Helicobacter pylori related gastritis in patients of duodenal ulcers (25-30%) were also seen receiving non-steroidal anti-inflammatory drugs - compari- son and correlation with peptic ulceration. I Clin Pathol in patients with active duodenitis or gastric 1992; 45: 135-9. metaplasia, or both, regardless of NSAID 9 Carrick J, Lee A, Hazell S, Ralston M, Daskalopoulos G. Campylobacter pylori, duodenal ulcer, and gastric meta- intake. Most patients with duodenal ulcers, plasia: possible role of functional heterotopic tissue in however, in the presence or absence ofNSAIDs, ulcerogenesis. Gut 1989; 30: 790-7. 10 Wyatt JI, Rathbone BJ, Sobala GM, Shallcross T, Heatley were found to have active duodenitis (80%) and RV, Axon ATR, et al. Gastric epithelium in the duodenum: gastric metaplasia (64%). The strong association its association with Helicobacter pylori and inflammation. J Clin Pathol 1990; 43: 981-6. between such histological and endoscopic 11 James AH. Gastric epithelium in the duodenum. Gut 1964; 5: entities is in agreement with the suggestion that 285-94. 12 Patrick WJA, Denham D, Forrest APM. Mucosa change in duodenitis and duodenal ulceration might repre- the human duodenum: a light and electron microscopic sent different points in a disease spectrum with a study and correlation with disease and gastric acid secretion. Gut 1974;-15: 767-76. common underlying pathogenesis.'139 0 There 13 Rhodes J. Experimental production of gastric epithelium in was also a strong association between active the duodenum. Gut 1964; 5:454-8. 14 Gerkens JR, Shand DG, Flexner C, Nies A, Oates J, Data J. duodenitis and H pylon positive gastritis, which Effect of indomethacin and aspirin on gastric blood flow and in turn highlights the potential benefits of acid secretion..7 Pharrnacol Exp Ther 1977; 203: 646-52. 15 Feldman M, Colturic TJ. Effect of indomethacin on gastric eradicating H pylori in minimising duodenal acid and bicarbonate secretion in humans. Gastroenterology damage in NSAID patients. The same would 1984; 87: 1339-43. 1166 Taha, Dahill, Nakshabendi, Lee, Sturrock, Russell

16 Levine RA, Schwartzel EH. Effect of indomethacin on basal pylori (HP) prevalence and mucosal injury in gastric ulcers and histamine stimulated gastric acid secretion. Gut 1984; (GUs): relationship to chronic non-steroidal anti-inflamma- 25: 718-22. tory drug (NSAID) ingestion. Gastroenterology 1991; 100: 17 Levine RA, Jyotirmoy N, King RL. Non-salicylate non- A103. steroidal anti-inflammatory drugs augment pre-stimulated 21 Farah D, Sturrock RD, Russell RI. Peptic ulceration in acid secretion in rabbit parietal cells. Investigations of the rheumatoid arthritis. Ann Rheum Dis 1988; 47: 478-80. Gut: first published as 10.1136/gut.34.9.1162 on 1 September 1993. Downloaded from mechanisms ofaction. Gastroenterology 1991; 101: 756-65. 22 Taha AS, Capell HA, Sturrock RD, Russell RI. Non-steroidal 18 Marshall BJ, McGechie DB, Rogers PA, Glancy RJ. Pyloric peptic damage in rheumatoid patients receiving second-line Campylobacter infections and gastroduodenal disease. Med drugs. AmJGastroenterol 1991; 86: 1588-91. JAust 1985; 142: 439-44. 23 Sobala GM, Fing RFG, Axon ATR, Dixon MF. Reflux 19 Casselli M, Pazzi P, La Corte R, Alcotti A, Trevisani L, gastritis in the intact stomach.J Clin Pathol 1990; 43: 303-6. Stabellini G. Campylobacter like organisms, non-steroidal 24 Taha AS, Sturrock RD, Russell RI. Helicobacter pylori and anti-inflammatory drugs, and gastric lesions in patients with peptic ulcers in rheumatoid patients receiving gold, sulpha- rheumatoid arthritis. Digestion 1989; 44: 101-4. salazine, and non-steroidal anti-inflammatory drugs. Am J 20 Laine L, Marin-Sorensen M, Weinstein WM. Helicobacter Gastroenterol 1992; 87: 1732-5. http://gut.bmj.com/ on October 2, 2021 by guest. Protected copyright.