Non-Steroidal Anti-Inflammatory Drugs and Gastrointestinal Damage

Home , Prostaglandin analogue

82 Postgrad Med J 2001;77:82–88 Postgrad Med J: first published as 10.1136/pmj.77.904.82 on 1 February 2001. Downloaded from Non-steroidal anti-inﬂammatory drugs and gastrointestinal damage—problems and solutions

R I Russell

Non-steroidal anti-inflammatory drugs (NSAIDs) have been used for many years for Box 1: Areas of the gastrointestinal analgesic, anti-inflammatory, and more re- tract that may be damaged by NSAIDs cently in the case of aspirin, antithrombotic x Oesophagus: oesophagitis, ulceration, purposes. The use of NSAIDs continues to stricture increase; over 22 million prescriptions are writ- x Stomach: ulcers, erosions ten every year in the UK, and over 70 million in x Duodenum: ulcers, erosions the US. These figures underestimate their full x Small intestine: ulcers, erosions, protein use as aspirin and other NSAIDs are widely loss, strictures available as “over-the-counter” preparations. x Colon: non-specific colitis, exacerbation The use of these drugs can also be expected of ulcerative colitis and Crohn’s disease to increase in the years to come, partly because of the increasing age of the population and intestinal bleed is between 1/100 and 1/500 partly because of possible new and developing patient years; the risk of a gastrointestinal indications, particularly in vascular disease and related death is between 1/1000 and 1/5000 cancer prevention. It is therefore of importance patient years. to assess the safety and side eVects of NSAIDs, The risk of serious ulcer complications in and to consider how their safety may be patients on NSAIDs has been calculated in improved. It has long been known that various studies, the relative risk (RR) for a seri- NSAIDs may have a range of side eVects, of ous gastrointestinal event overall being re- which the commonest are gastrointestinal. ported as 2.74 in a large meta-analysis, upper It is the purpose of this paper to examine the gastrointestinal bleeding RR 3.09, perforation nature, range, and causation of the gastro- RR 5.93, and ulcer related death RR 7.62.3 intestinal side eVects associated with the use of NSAIDs may have adverse eVects in all parts NSAIDs, and to consider how these may be of the gastrointestinal tract, not only the stom- reduced or modified. NSAIDs may also be ach or duodenum; the oesophagus, small intes- associated with other adverse eVects, although tine and colon may also be aVected (box 1). these are much rarer than gastrointestinal A recent necropsy study of 713 patients has problems; these include nephropathies, skin been reported, of whom 244 had NSAIDs pre- rashes, and hepatitis. NSAIDs may also scribed during the six months before death and http://pmj.bmj.com/ interact with other drugs such as antihypergly- 464 had not; death in all patients was unrelated caemic or antihypertensive agents. to NSAID use. Ulcers of the stomach or duodenum were found in 21.7% of patients on Gastrointestinal problems NSAIDs and 12.3% of those not on NSAIDs NSAIDs may be associated with many gastro- (p<0.001), and small intestinal ulceration was intestinal problems, ranging from mild to found in 8.4% of NSAID users and 0.6% of severe dyspeptic symptoms, the development patients not on NSAIDs (p<0.001).4 Damage on September 30, 2021 by guest. Protected copyright. of gastric or duodenal ulceration, haemorrhage to the small intestine in NSAID users has also or perforation, and other events which may been demonstrated at enteroscopy. Patients lead to hospitalisation or death. with small intestinal damage associated with Endoscopic studies have shown a prevalence chronic NSAID use may present with chronic rate of 14%–25% of gastric and duodenal iron deficiency anaemia or hypoalbuminaemia ulcers in NSAID users,1 although the diYculty due to blood or protein loss, and may develop of having accurate control groups makes exact overt bleeding, perforation, or strictures. figures diYcult to obtain. Although endoscopic In the oesophagus, chronic NSAID use may studies tend to show more gastric than duode- be associated, although rarely, with oesoph- nal ulcers associated with NSAID use, patients agitis, ulceration, or stricture formation.5 presenting with gastrointestinal bleeding on In the large bowel, NSAID use may lead to NSAIDs may have a similar frequency of the development of a non-specific colitis (with University of Glasgow gastric and duodenal ulceration. Dyspeptic and Department of abdominal discomfort, bloody diarrhoea, and Gastroenterology, symptoms occur in up to 60% of patients tak- weight loss). NSAIDs may also cause an exac- Royal Infirmary, ing NSAIDs and there is a poor correlation erbation of ulcerative colitis or Crohn’s disease, Glasgow between symptoms and endoscopically proved if the associated arthropathy in inflammatory lesions; up to 50% of endoscopically confirmed bowel disease is treated with NSAIDs.67 Correspondence to: ulcers associated with NSAIDs are asympto- Professor R I Russell, High risk factors for NSAID related gastro- 28 Ralston Road, Bearsden, matic. intestinal damage include older age group Glasgow G61 3BA It has been estimated that the individual risk (especially over 70), previous history of peptic [email protected] of hospitalisation with gastrointestinal compli- ulceration, and probably the first three months Submitted 26 October 1999 cations of NSAIDs is between 1/50 and 1/150 of NSAID treatment. Other high risk factors Accepted 22 June 2000 patient years.2 The risk of an upper gastro- are smoking, underlying respiratory or cardio-

www.postgradmedj.com Non-steroidal anti-inﬂammatory drugs and gastrointestinal damage 83 Postgrad Med J: first published as 10.1136/pmj.77.904.82 on 1 February 2001. Downloaded from

Box 2: Risk factors for gastrointestinal Box 3: COX-1 and COX-2 damage due to NSAIDs COX-1 x Age: older age group, especially over 70 x Constitutively expressed in most tissues: x Previous ulcer history stomach, duodenum, kidneys, platelets x First three months of treatment x Key role in the production of x Smoking prostaglandins which regulate physi- x Underlying cardiovascular or respira- ological processes: gastrointestinal pro- tory disease tection, blood flow, adaptation, cellular x Concomitant drug use, especially recovery, maintenance of good renal corticosteroids and anticoagulants function, vascular homoeostasis impor- x Use of NSAIDs in high doses, and more tant “housekeeping” role than one NSAID COX-2 x ?Helicobacter pylori x Normally undetectable in most tissues x Can be induced rapidly and in large quantities in the presence of inflamma- vascular disease, and concomitant drug use, tion and other pathological processes particularly corticosteroids, aspirin, and anticoagulants (box 2).8 Perioperative use is also a risk factor. Individual NSAIDs incurring high- maintenance of good renal function and in est risk include azapropazone, ketoprofen and maintaining normal physiological functions in piroxicam, and those with least risk include many other cells. This is an important “house- ibuprofen, diclofenac, and etodolac. Higher keeping” role. COX-2, on the other hand, is doses are associated with increased risk and normally undetectable in most tissues, but it also the use of more than one NSAID. There is can be induced rapidly, and in large also an increased risk of gastrointestinal quantities—to 200–300-fold—by cytokines, complications with relatively low dose prophy- growth factors, and hormones in the presence lactic aspirin, which is widely used nowadays.9 of inflammation and other pathological proc- There is debate on the possible interaction and esses (box 3). Platelets appear to contain only increased risk of Helicobacter pylori causing COX-1. ulceration with NSAIDs, and this will be Most conventional NSAIDs are non- discussed in a later section. The balance of selective in their COX inhibition, exerting their physiological association and clinical evidence anti-inflammatory eVects through the inhibi- tends to support the possibility of some tion of COX-2, but having adverse eVects interaction occurring, especially in those pa- (such as gastrointestinal mucosal damage and tients at high risk and possibly in those who nephrotoxicity) primarily due to inhibition of have bled. Well planned, prospective controlled COX-1. Some existing and longstanding studies are required to provide more infor- NSAIDs, such as etodolac, a known safer mation on this with respect to diVerent types NSAID with respect to gastrointestinal dam- and doses of NSAIDs, diVerent forms and age and which is known to have reduced http://pmj.bmj.com/ strains of H pylori, and diVerent types of adverse eVects on mucosal prostaglandins, has erosions and ulcers. been subsequently found to have a degree of COX-2 inhibition selectivity. Why do NSAIDs cause gastrointestinal Currently, a range of new and specific damage? COX-2 selective inhibitor NSAIDs have been, NSAIDs interfere with the cyclo-oxygenase and are being further developed with the hope (COX) pathways which lead to the production of reducing possible gastrointestinal side ef- on September 30, 2021 by guest. Protected copyright. of prostanoids (prostaglandins, prostacycline, fects; the preliminary clinical results, as we and thromboxane). This interferes with mu- shall see later, are encouraging. One of these, cosal protection by reducing the eVectiveness rofecoxib, is now available in this country. of the mucus-bicarbonate barrier; gastric acid, and possibly also pepsin, are thus more likely to Solutions cause damage. The fact that most NSAIDs are Treatment and prevention of NSAID related also weak acids may also be a contributory fac- gastrointestinal problems are shown in box 4. tor. It is now known that COX exists as two HEALING OF NSAID RELATED ULCERS separate isoforms, COX-1 and COX-2, which There is evidence from both animal and diVer markedly in their tissue expression and human studies that NSAIDs retard the healing regulation.10–12 COX-1 is constitutively ex- of gastric ulcers. Although it is frequently pressed in most tissues, including the stomach, stated that a first step if gastrointestinal duodenum, platelets, and kidneys; COX-1 problems occur with NSAIDs is to withdraw plays a key part in the production of prostag- NSAIDs or reduce the dose, in practical terms landins which regulate important physiological this is generally unrealistic, especially when processes such as gastrointestinal cytoprotec- there is a major inflammatory condition such tion (maintaining an eVective mucus- as rheumatoid arthritis. bicarbonate barrier, submucosal blood flow, It may be possible to reduce the dose of quicker and more eVective mucosal adaptation NSAIDs in some individuals. It is also to initial tissue damage, and more rapid recov- appropriate to try alternative conventional ery when such damage occurs). It is also NSAIDs as there is a wide variability in involved in vascular homoeostasis and the responses to individual drugs. Some NSAIDs

www.postgradmedj.com 84 Russell Postgrad Med J: first published as 10.1136/pmj.77.904.82 on 1 February 2001. Downloaded from NSAIDs—selective COX-2 inhibitors—oVer Box 4: Treatment and prevention of the prospect of greater gastrointestinal safety. NSAID related gastrointestinal problems PROSTAGLANDINS Healing of NSAID related ulcers The prostaglandin analogue misoprostol has x Reduce NSAID dose if possible, or use shown a reduction in ulceration compared with alternative NSAIDs placebo ranging from 50% to 90% over 3–12 x Use of acid suppressive drugs while months; the dose used initially was 200 µg four continuing NSAIDs: high dose times a day and subsequently 400–600 µg per H2-receptor antagonists, or more eVec- day. Protection has been generally similar for tively, proton pump inhibitors both gastric and duodenal ulcers.17 18 A large Prevention of NSAID related gastrointestinal blinded study of misoprostol compared with problems placebo (the “MUCOSA” study) reported a x Avoidance or reduction of risk factors if signiﬁcant reduction of gastrointestinal bleed- possible ing and perforation with misoprostol at 800 µg x Drugs which may improve mucosal pro- daily.19 In this study, 27% of patients withdrew tection: prostaglandin analogues because of side eVects, principally diarrhoea. (misoprostol) Diarrhoea remains a problem for many pa- x Acid suppressive drugs: H2-receptor tients with this preparation. Misoprostol has antagonists (high doses), proton pump also been combined with a standard NSAID, inhibitors diclofenac (arthrotec), and this has proved x COX-2 speciﬁc inhibitors popular with a number of patients. x Eradication of Helicobacter pylori (in Theoretically, prostaglandin analogues have selected cases) an advantage over acid suppressants in that they should provide mucosal protection throughout the gastrointestinal tract. are more likely than others to cause gastrointestinal problems; those associated with ACID SUPPRESSION greatest risk include azapropazone, ketoprofen, As gastric acid is a factor in causing gastroduo- and piroxicam. denal damage associated with NSAID use,

H2-receptor antagonists, especially in high reduction of gastric acid by the coprescription doses, do heal NSAID related ulcers while of acid suppressant drugs has been studied. 13 NSAIDs are continued, but proton pump The use of H2-receptor antagonists in stand- inhibitors may be expected to be more ard doses for the prevention of NSAID associ- eVective, and have now been shown to be so. ated ulcers has shown some protection against Omeprazole has been shown to heal gastric duodenal ulcers.20 and the use of high dose ulcers faster than ranitidine.14 In two large famotidine (40 mg twice a day) has been shown studies comparing omeprazole (20 mg or 40 to signiﬁcantly reduce the cumulative inci- mg daily) with misoprostol (200 µg four times dence of both gastric and duodenal ulcers in a day), the proton pump inhibitors healed sig- patients with rheumatoid arthritis on long term http://pmj.bmj.com/ niﬁcantly more gastric and duodenal ulcers NSAID therapy.21 than ranitidine or misoprostol. Proton pump inhibitors cause more eVective

No extra benefit was gained from using acid suppression than H2-receptor antagonists, omeprazole at the higher dose.15 16 and several randomised controlled trials lasting 3–6 months using omeprazole 20 mg daily have PREVENTION OF NSAID RELATED shown eYcacy in preventing both gastric and GASTROINTESTINAL PROBLEMS duodenal ulcers while continuing long term on September 30, 2021 by guest. Protected copyright. The lowest dose of the safest NSAID which is NSAIDs, with about a 75%–80% reduction in eVective in individual patients should be used if ulcers compared with placebo.15 16 22 possible. The proton pump inhibitors were shown in Avoidance or reduction of risk factors may two of these studies to provide more eVective be possible in some patients, such as smoking, protection than ranitidine (150 mg twice a day) use in patients with a past history of dyspeptic or misoprostol (200 µg twice a day). problems or peptic ulcer, perioperative use, or Omeprazole 20 mg daily has also been cotherapy with other drugs such as cortico- shown to protect against bleeding from ulcers steroids, anticoagulants, or aspirin. Advice in long term NSAID users.23 should be given about “over-the-counter” preparations. NEWER NSAIDS: COX-2 SPECIFIC INHIBITORS There will inevitably be patients in whom The search for safer NSAIDs has recently additional requirements will be necessary, such focused on the development of preferential or as those who continue to require high dose selective COX-2 inhibitors. These compounds NSAIDs or in those with risk factors which aim to exploit the belief that COX-1 is associ- cannot be modified. ated predominantly with the production of In such patients there has been much interest protective prostaglandins and has a “house- in recent years in attempting to reduce risk by keeping” role, whereas COX-2 is induced in the coprescription with NSAIDs of drugs inflammation and associated with inflamma- which may improve mucosal protection (the tory processes. The development of these con- prostaglandin analogue misoprostol) or reduce cepts has been recently reviewed.24

gastric acid (H2-receptor antagonists or proton The deﬁnition of, and screening for, COX-2 pump inhibitors). More recently, newer selectivity has given rise to much debate,

www.postgradmedj.com Non-steroidal anti-inflammatory drugs and gastrointestinal damage 85 Postgrad Med J: first published as 10.1136/pmj.77.904.82 on 1 February 2001. Downloaded from involving a variety of suggested systems that platelet levels of COX, which it acetylates and includes purified recombinant enzyme, trans- to which it binds irreversibly, thus impairing fected cells, and whole blood assays. The latter platelet function. This causes changes in the method may be the most appropriate at bleeding time, platelet aggregation, and the present, but all drugs suggested as COX-2 spe- synthesis of thromboxane, which remain for cific inhibitors require to be shown to have no the life of the platelet. The damage is worse significant inhibition of gastric mucosal pros- with longer acting NSAIDs with long half lives taglandins, in addition to a good clinical safety and an extensive enterohepatic circulation, profile and eVectiveness. such as piroxicam. Conventional longstanding NSAIDs have With increasing use, aspirin is now a major been found to vary with respect to their relative cause of upper gastrointestinal problems, and COX-1 to COX-2 inhibition capacity, but all in particular, upper gastrointestinal bleeding. have a significant eVect on COX-1, thus inter- One recent study reported an overall odds ratio fering with gastric mucosal protection. of 3.2 (95% confidence interval (CI) 2.3 to One NSAID, etodolac, which has been used 4.4) in gastrointestinal bleeding for daily aspi- widely for some years, has been shown in rin use of at least one month, compared with several studies to be clinically eVective and 3.8 (95% CI 3.1 to 4.5) for non-aspirin have reduced gastrointestinal toxicity, demon- NSAID use.9 This can be expected to increase strated endoscopically and clinically; these with increasing use of low dose aspirin, studies also found no significant reduction of especially common in older age groups. gastric mucosal prostaglandins compared with Platelets contain only COX-1; thus only naproxen.25 26 Etodolac has since been found to drugs such as aspirin and those NSAIDs which have some degree of COX-2 selectivity.27 inhibit COX-1 will inhibit platelet function. Other drugs with probable relative COX-2 COX-1 sparing NSAIDs such as etodolac, selectivity include nabumetone, nimesulide, nabumetone, nimesulide, meloxicam, are likely and meloxicam.24 to have little eVect on platelet dysfunction. The Newer more specific COX-2 inhibitors highly specific COX-2 inhibitors (celecoxib, currently under study include celecoxib and rofecoxib), have almost no eVect on platelet rofecoxib (the latter now being available in this function, and are likely, as mentioned previ- country). Preliminary results of studies with ously, to lead to marked reductions in upper these drugs indicate eVectiveness and a good gastrointestinal bleeding. However, if used with gastrointestinal safety profile. Celecoxib (25– low dose aspirin, there is likely to be a loss of 400 mg twice a day) was associated with a sig- benefit. nificantly reduced incidence of adverse upper COX-2 specific inhibitors will not be helpful gastrointestinal problems including ulcers, in vascular disease, and cannot be considered gastrointestinal bleeding, and perforation in as a replacement in this respect for low dose patients with osteoarthritis and rheumatoid aspirin. arthritis compared with standard NSAIDs.28 A new antiplatelet agent, clopidrogel may be Rofecoxib (12.5–25 mg a day) demonstrated a safer than aspirin, although more costly. lower incidence of endoscopic ulcers compared Clopidrogel requires to be further assessed, but http://pmj.bmj.com/ with ibuprofen in patients with osteoarthritis in may be helpful in patients with peripheral a multicentre study over 24 weeks.29 Rofecoxib artery disease, and in stroke or myocardial inf- has also shown reduced occurrence of ulcers, arction in whom aspirin is contraindicated, or perforations, and gastrointestinal bleeding. in whom aspirin fails to achieve the required COX-2 specific inhibitors have little, if any, therapeutic eVect. eVect on platelet function, and may therefore be associated with a marked reduction of WHAT TO DO ABOUT H PYLORI? on September 30, 2021 by guest. Protected copyright. gastrointestinal bleeding associated with H pylori and NSAIDs are the two most NSAID use. common causes of peptic ulceration. These studies are promising and more data The possible interactions of H pylori and are awaited. The long term safety of these NSAIDs have led to much discussion recently drugs remains to be established, particularly and no clear picture emerges that can apply to with respect to the kidney and in the possible, all situations. In many ways, NSAIDs and H although less likely, impairment of healing of pylori have similar adverse eVects on mucosal pre-existing peptic ulcers. It is also theoretically protective mechanisms, and despite H pylori possible that COX-2 inhibitors may have a itself producing small amounts of prostagland- benefit in preventing colon cancer and possibly ins, there remains the possibility of an additive also in Alzheimer’s disease. COX-2 inhibitors damaging eVect when both are present.30 Stud- will not be eVective in cardiovascular protec- ies on mucosal adaptation and on neutrophils tion unlike aspirin, as the protective eVects raise the possibility of some inter-relationship provided by aspirin are mediated through that may allow damage to occur more readily COX-1. when NSAIDs are taken in the presence of H pylori.31 LOW DOSE ASPIRIN The development of an ulcer when NSAIDs Aspirin is widely used in low doses in patients are given to patients who are H pylori positive with suspected or definite vascular disease. may depend on the interaction of a number of Aspirin is a major cause of upper gastro- factors, including previous exposure to intestinal problems, and is a frequent cause of NSAIDs, past history, gastric acid output, and upper gastrointestinal bleeding. Even in low the use of acid suppression drugs such as pro- doses (75 mg daily), aspirin has eVects on ton pump inhibitors.

www.postgradmedj.com 86 Russell Postgrad Med J: first published as 10.1136/pmj.77.904.82 on 1 February 2001. Downloaded from Aspirin may have diVerent interactions with H pylori, particularly with respect to the risk of Questions (answers on p 88) gastrointestinal haemorrhage. 1. Do patients with NSAID related ulcers Gastrointestinal haemorrhage from ulcers always have dyspeptic symptoms? maybeinadiVerent category from non- 2. In a patient on long term NSAIDs who bleeding ulcers, and really requires to be stud- has developed chronic iron deficiency ied separately; other factors may be involved, anaemia, is the cause always likely to be such as the possible antiplatelet eVects of many found in the stomach or duodenum? NSAIDs, which may be a further factor in 3. Is the use of low dose prophylactic bleeding. aspirin safe with respect to the gastro- Clinical studies in this area have given intestinal tract? conflicting results, although one study in 4. Why don’t we just reduce the dose of patients not previously given NSAIDs clearly NSAIDs, or even stop NSAIDs, in demonstrated an advantage in eradicating H patients at high risk or with gastro- pylori before starting NSAIDs; significantly intestinal problems? fewer lesions developed over two months in 5. Is the cost of coprescription of acid patients in whom H pylori had been eradicated suppressive drugs justified in patients before starting NSAIDs.32 In such patients, on long term NSAIDs? especially if risk factors are present (older age 6. Is it likely that selective COX-2 inhibi- group, previous ulcer history), eradication of H tors will reduce serious problems such pylori before starting NSAIDs may be indi- as gastrointestinal haemorrhage? cated. 7. Are patients on corticosteroids in addi- A large multicentre study has found that tion to NSAIDs at greater risk of eradication of H pylori did not reduce the rate gastrointestinal problems? of ulcer relapse in existing long term NSAID 8. Have prostaglandin analogues any users.33 In this study, it was also reported in a advantage over acid suppressive agents subgroup of 41 patients with gastric ulcers in preventing NSAID problems in the found at baseline endoscopy that eradication of gastrointestinal tract? H pylori was associated with delayed ulcer 9. In considering risk factors for NSAID healing. However, in another prospective, related gastrointestinal damage, is there randomised study of 195 patients with H pylori any particular time of greatest danger? infection and NSAID associated bleeding 10. Why should specific COX-2 inhibi- ulcers, eradication of H pylori did not impair tors be safer? healing of gastric or duodenal ulcers.34 A prospective, randomised clinical outcome study compared H pylori eradication alone with corticosteroids and anticoagulants, high dose long term omeprazole for the prevention of and multiple NSAIDs, and possibly in some recurrent ulcer haemorrhage in high risk users cases H pylori. of aspirin or non-aspirin NSAIDs.35 The results

x Low dose prophylactic aspirin may also be http://pmj.bmj.com/ to date indicate that eradication of H pylori associated with adverse gastrointestinal ef- alone did not prevent recurrent ulcer bleeding fects. associated with non-aspirin NSAIDs, but H x Healing of NSAID related ulcers can be pylori eradication alone was as eVective as achieved while NSAIDs are continued by maintenance omeprazole in preventing recur- the use of H -receptor antagonists in high rent haemorrhage associated with low dose 2 doses, or more eVectively by proton pump aspirin. Perhaps the antiplatelet eVects of inhibitors. NSAIDs are relevant there. x Prevention of NSAID related gastro- on September 30, 2021 by guest. Protected copyright. H pylori testing would probably not be indi- intestinal problems may be achieved by cated in patients who have already been on identifying and if possible reducing risk fac- NSAIDs or aspirin for some time without any tors, the coprescription of prostaglandin adverse eVects. In patients with a past history analogues or acid suppressive drugs (espe- of peptic ulcer or dyspepsia on NSAIDs, long cially proton pump inhibitors), or by using term acid suppressive agents such as proton the currently being developed and promising pump inhibitors are indicated even after eradi- COX-2 specific inhibitors. cation of . H pylori x The development of COX-2 specific inhibi- More scientific and clinical data are required tors oVers the hope of real progress in on these complex inter-relationships. producing much safer and eVective NSAIDs. Conclusions x The use of NSAIDs continues to increase, 1 Hudson N, Hawkey CJ. Non-steroidal anti-inflammatory especially in the elderly. drug associated upper gastrointestinal ulceration and x NSAIDs may have adverse eVects in any part complications. Eur J Gastroenterol Hepatol 1993;5:412–19. of the gastrointestinal tract: oesophagus, 2 McCarthy, DM. Nonsteroidal anti-inflammatory drugs: reducing the risks to the gastrointestinal tract. Clinical stomach, duodenum, small intestine, or Perspectives in Gastroenterology 1999;4:219–26. 3 Gabriel SE, Jaakkimaimen L, Bombardier C. Risk for colon. serious gastrointestinal complications related to the use of x Risk factors for gastrointestinal damage due non-steroidal anti-inflammatory drugs—a meta-analysis. to NSAIDs include age, previous ulcer Ann Intern Med 1991;115:787–96. 4 Allison MC, Howatson AG, Torrance CJ, et al. Gastro- history, first three months of treatment, intestinal damage associated with the use of non-steroidal smoking, underlying cardiovascular or respi- anti-inflammatory drugs. N Engl J Med 1992;327:749–54. 5 Mason JC. NSAIDs and the oesophagus. Eur J Gastroenterol ratory disease, concomitant drug use with Hepatol 1999;11:369–73.

www.postgradmedj.com Non-steroidal anti-inflammatory drugs and gastrointestinal damage 87 Postgrad Med J: first published as 10.1136/pmj.77.904.82 on 1 February 2001. Downloaded from 6 Faucheron J-L. Toxicity of non-steroidal anti-inflammatory non-steroidal anti-inflammatory drugs. N Engl J Med 1996; drugs in the large bowel. Eur J Gastroenterol Hepatol 334:1435–9. 1999;11:389–92. 22 Ekstrom P, Carling L, Wetterhus S, et al. Prevention of pep- 7 Forrest E, Russell RI. Peripheral arthropathies in inflamma- tic ulcer and dyspeptic symptoms with omeprazole patients tory bowel disease. Gut 1999;44:439. receiving continuous non-steroidal anti-inflammatory drug 8 Russell RI. Defining patients at risk of NSAID gastropathy. therapy. A Nordic multi-centre study. Scand J Gastroenterol Ital J Gastroenterol Hepatol 1999;31(suppl 1 S):14–18. 1996;31:753–8. 9 Weil J, Colin-Jones D, Langman M, et al. Prophylactic aspi- 23 Chan FKL, Sung JY, Suen R, et al. Eradication of H pylori rin and risk of peptic ulcer bleeding. BMJ 1995;310:827– versus maintenance acid suppression to prevent recurrent 30. ulcer haemorrhage in high risk NSAID users: a prospective 10 Fu JY, Masferrer JL, Seibert K, et al. The induction and randomised study. Gastroenterology 1998;114:A87. suppression of prostaglandin H2 synthase (cyclo- 24 Hawkey CJ. COX-2 inhibitors. Lancet 1999;353:307–14. oxygenase) in human monocytes. J Biol Chem 1990;265: 25 Russell RI. Endoscopic evaluation of etodolac and 16737–40. naproxen, and their eVects on gastric and duodenal prostag- 11 Kujubu DA, Fletcher BS, Varnum BC, et al. TIS 10, a phor- landins. Rheumatol Intern 1990;10(suppl):17–21. bol ester tumor promoter-inducible mRNA from Swiss 3T 3 26 Laine L, Sloane R, Ferretti M, et al. A randomised, double- cells, encodes novel prostaglandin synthase/cyclo-oxygenase blind comparison of placebo, etodolac, and naproxen on homologue. J Biol Chem 1991;266:12866–72. gastrointestinal injury and prostaglandin production. Gas- 12 DeWitt DL, Meade EA, Smith WL. Prostaglandin H trointest Endosc 1995;42:428–33. synthase isoenzyme activity: the potential for safer non- 27 Glaser K. Cyclo-oxygenase activity and NSAIDs: cyclo- steroidal anti-inflammatory drugs. Am J Med 1993;95:40S– oxygenase-2 selectivity of etodolac (Lodine). 6. Inflammo Pharmacology 1995;3:335–45. 13 Hudson N, Taha AS, Russell RI, et al. Famotidine for healing and maintenance in non-steroidal anti-inflammatory 28 Goldstein AL, Agrawal NM, Silverstein F, et al. Celecoxib is drug associated gastroduodenal ulceration. associated with a significantly lower incidence of clinically Gastroenterology significant upper gastrointestinal events in osteoarthritis and 1997;112:1817–22. rheumatoid arthritis patients as compared to NSAIDs. 14 Walan A, Bader JP, Glassen M, et al. EVect of omeprazole Gas- and ranitidine on ulcer healing and relapse rates in patients troenterology 1999;116A:174. with benign gastric ulcer. N Engl J Med 1989;320:69–75. 29 Laine L, Hawkey CW, Harper S, et al.EVect of a COX-2 15 Hawkey CJ, Karrasch JA, Szczepanski L, et al. Omeprazole specific inhibitor Rofecoxib on ulcer formation: a double- compared with misoprostol for ulcers associated with blind comparison with ibuprofen and placebo. Gastroenterol- non-steroidal anti-inflammatory drugs. N Engl J Med 1998; ogy 1999;116A:229. 338:727–34. 30 Taha AS, Sturrock RD, Russell RI. Mucosal erosions in 16 Yeomans ND, Tulassay Z, Juhasz L, et al. Omeprazole com- long-term non-steroidal anti-inflammatory drug users: pre- pared with ranitidine for ulcers associated with non- disposition to ulceration and relation to Helicobacter pylori. steroidal anti-inflammatory drugs. N Engl J Med 1998;338: Gut 1995;36:334–6. 719–26. 31 Taha AS, Dahill S, Morran C, et al. Neutrophils, 17 Graham DY, Agrawal NM, Roth SH. Prevention of Helicobacter pylori, and non-steroidal anti-inflammatory NSAID-induced gastric ulcer with misoprostol: multi- drug ulcers. Gastroenterology 1999;116:254–8. centre, double-blind, placebo-controlled trial. Lancet 1988; 32 Chan FKL, Sung JJY, Chung C, et al. Randomised trial of ii:1277–80. eradication of Helicobacter pylori before starting non- 18 Graham DY, White RH, Moreland LW, et al and the Miso- steroidal anti-inflammatory drug therapy to prevent peptic prostol Study Group. Duodenal and gastric ulcer prevention ulcers. Lancet 1997;350:975–9. with misoprostol in arthritis patients taking NSAIDs. Ann 33 Hawkey CJ, Tulassay Z, Szczepanski L, et al. Randomised Intern Med 1993;119:257–62. control trial of Helicobacter pylori eradication in patients on 19 Silverstein FE, Graham DY, Senior JR, et al. Misoprostol non-steroidal anti-inflammatory drugs: HELP NSAIDs reduces serious gastrointestinal complications in patients study. Lancet 1998;352:1016–21. with rheumatoid arthritis receiving non-steroidal anti- 34 Chan FKL, Sung JJY, Suen R, et al. Does eradication of H inflammatory drugs. A randomised, double-blind, placebo- pylori impair healing of non-steroidal anti-inflammatory controlled trial. Ann Intern Med 1995;123:241–9. drug associated bleeding peptic ulcers? A prospective 20 Ehsanullah RSB, Page MC, Tildesley G, et al. Prevention of randomised study. Aliment Pharmacol Ther 1998;12:1201–5. gastroduodenal damage induced by non-steroidal anti- 35 Chan FKL, Sung JJY, Suen R, et al. Prospective randomised inflammatory drugs: controlled trial of ranitidine. BMJ trial of H pylori eradication versus maintenance omeprazole 1988;297:1017–21. to prevent recurrent upper gastrointestinal haemorrhage in 21 Taha AS, Hudson N, Hawkey CJ, et al. Famotidine for the high-risk aspirin and non-aspirin NSAID users. Gastroenter- prevention of gastric and duodenal ulcers caused by ology 2000;118:A194. http://pmj.bmj.com/ on September 30, 2021 by guest. Protected copyright.

www.postgradmedj.com 88 Russell Postgrad Med J: first published as 10.1136/pmj.77.904.82 on 1 February 2001. Downloaded from

Answers 1. No: Up to 50% of endoscopically con- patients with rheumatoid arthritis and firmed ulcers associated with NSAIDs osteoarthritis. COX-2 specific inhibi- are asymptomatic. Dyspeptic symptoms tors have little eVect on platelet func- occur in up to 60% of patients taking tion and are likely therefore to be NSAIDs and there is a poor correlation associated with a marked reduction of between symptoms and endoscopically bleeding. proved lesions. 7. Yes: Such patients have a greatly 2. No: NSAIDs may have adverse eVects increased relative risk of adverse eVects in all parts of the gastrointestinal tract, with NSAIDs. Many patients with long including oesophagus, stomach, duode- standing rheumatoid arthritis are on num, small intestine, and colon. corticosteroid therapy in addition to A patient with chronic iron defi- NSAIDs. ciency anaemia on long term NSAIDs 8. They may have in that they theoretically may well have small intestinal damage provide mucosal protection throughout associated with NSAIDs, leading to the gastrointestinal tract. This may be of blood loss and chronic iron deficiency particular importance in the small anaemia. Overt bleeding, perforation, intestine . However, many patients or strictures of the small intestine may treated with the prostaglandin analogue also occur. misoprostol develop diarrhoea, and 3. No: There is an increased risk of some patients may require to discon- gastrointestinal complications with rela- tinue treatment because of this. tively low dose prophylactic aspirin, and 9. Yes: It is considered that the first three overt gastrointestinal haemorrhage may months of NSAID therapy is the time occur. With increasing use of low dose of greater risk. This may be related to aspirin, there is likely to be increasing the ability of the gastrointestinal tract to problems with gastrointestinal side adapt to NSAIDs after this period. eVects. A high level of awareness of the 10. COX exists as two separate isoforms, dangers of low dose prophylactic aspi- COX-1 and COX-2. COX-1 is present rin is necessary. in most tissues including the gastro- 4. It is rarely feasible to do so, especially intestinal tract and is involved in regu- in patients who have been on NSAIDs lating physiological processes such for some time. Most of these patients gastrointestinal mucosal protection, have rheumatoid arthritis and require submucosal blood flow, and mucosal the continuation of NSAIDs for adaptation. COX-2 is induced rapidly adequate treatment of the disease. in most tissues in the presence of 5. Yes: In patients at high risk or who have inflammation and other pathological had gastrointestinal problems with processes. Most conventional NSAIDs

NSAIDs, the use of a prostaglandin ana- aVect both COX-1 and COX-2, thus http://pmj.bmj.com/ logue or the more expensive acid sup- interfering with the mucosal protection pressive drugs such as proton pump processes in the gastrointestinal tract. inhibitors, is justiﬁed in preventing Selective COX-2 inhibitors do not gastrointestinal symptoms and the devel- interfere with the protective physiologi- opment of ulceration and possibly cal processes associated with COX-1 gastrointestinal haemorrhage or perfora- and oVer the prospect of greater gastro- tion. The use of COX-2 selective inhibi- intestinal safety. tors may be an important development. It should be noted that COX-2 on September 30, 2021 by guest. Protected copyright. 6. Yes: It is likely that this will be the case, inhibitors will not be eVective in although more data are awaited. There cardiovascular protection as the protec- is early data suggesting that selective tive eVects provided by aspirin are COX-2 inhibitors may reduce the inci- mediated through COX-1. COX-2 dence of ulcers, gastrointestinal bleed- selective inhibitors cannot therefore be ing, and perforation in used to replace aspirin in this way.