Nonsteroidal Antiinflammatory DrugInduced Ulcers: Management by Traditional Therapies DENIS M. McCARTHY University of New Mexico and Veterans Administration Medical Center, Albuquerque, New Mexico
Endoscopic distinction between ulcers and erosions is difficult. Consequently, existing literature, which must be taken at face value, may be misleading. Nevertheless, from published studies most gastric and duodenal ulcers associated with nonsteroidal antiinflammatory drugs appear to heal on antacids or H2 -antagonists. Sucralfate appears useful for duodenal but not gastric ulcers. Continuing nonsteroidal antiinflammatory drugs does not prevent or delay healing of duodenal or small gastric ulcers; their effects on large gastric ulcers remain uncertain. Thus far, only full doses of H2 -antagonists, or their combinations with antacids, have been shown to heal ulcers and prevent recurrences. Ulcer recurrences and complications have occurred in small numbers of patients on maintenance doses of H2 antagonists. Available antiulcer drugs (antacids, H2 -antagonists, sucralfate) reduce severe acute injury when taken before or with nonsteroidal antiinflammatory drugs. They also reduce ulcerlike symptoms due to nonsteroidal antiinflammatory drugs. Inexplicably, chronic prophylaxis with H2 -antagonists for 4 wk or more appears ineffective in preventing gastric ulcers, although duodenal injury is reduced. As the efficacy of available prophylactic therapy (H2 -antagonists, sucralfate, and antacids) has not been established, routine use in all cases seems unjustified at present.
entral to evaluating the success of various therapies for gastroduodenal ulcers encountered during the use of antiinflammatory drugs, is the basic issue of what constitutes an ulcer. Before the widespread availability of endoscopy, ulcers were defined by pathologists, based on operative or autopsy specimens, as losses of epithelial surface deep enough to reach or penetrate the muscularis mucosae. However, in contemporary practice and in clinical trials ulcers are seen face on, using endoscopic
techniques that do not permit accurate measurements of lesion depth (1). In this setting, the pathological definition has not proved useful, especially in conducting clinical trials. Particularly -in the past 5 yr, investigators have tried repeatedly to come up with a satisfactory alternative working definition of an ulcer, but despite a conference on the topic in 1987 sponsored by the Food and Drug Administration, no consensus has been reached. There is general agreement that an ulcer should possess depth, but how much depth, and how it should be measured, have not been established. Over this same time has emerged an increasing emphasis on routine measurement of the diameter(s) of mucosal surface lesions encountered at endoscopy, and the recording of the presence of exudate, fibrosis, or scarring. Also important is whether or not the lesion can be grasped with a biopsy forceps and readily moved over the muscularis (implying acuteness and superficiality of the lesion) or is immovable and tethered (implying deep extension and chronicity). Lesions whose diameters are less than an arbitrary size (see below) should be called "erosions," lesions with larger diameter that are movable and not scarred should be called "acute ulcers," and large lesions that are clearly deep or relatively fixed should be called "chronic ulcers." None of the studies reviewed here applied these kinds of criteria to define the nature of the treated lesions. This is of some importance as the therapeutic responses of the various lesions are almost certainly different, and grouping them all together under headings such as nonsteroidal antiinflammatory Abbreviations used in this paper: DU, duodenal ulcer; GU, gastric ulcer; NSAID, nonsteroidal antiinflammatory drug; PU, peptic ulcer. © 1989 by the American Gastroenterological Association 0016-5085/89/$3.50
drug (NSAID)-induced lesions (1), obscures important differences between studies. This in turn affects the apparent efficacies of the drugs being studied. The lesions associated with NSAID use include erythema; edema; various types of mucosal hemorrhage (friability, petechiae, subepithelial hemorrhages, intramucosal hemorrhages, red stre~ks, and ecchymoses); erosions (e.g., large, small, deep, confluent); and finally ulcers, some of which are acute and some of which are chronic. Faced with this diversity, investigators have devised a variety of different scoring systems (2-4) that attach some numerical value to each lesion. Next, they compute by simple addition a total "lesion score" for each patient, and use changes in the lesion score as the basis for claiming a therapeutic effect or a difference in injury potential between different NSAIDs (this term is used throughout to include all salicylates and other nonsteroidal drugs). The scores are computed from many different numbers and kinds of lesions. Despite the innate attractiveness of this simple approach, the assumptions and data weighting involved in using such scoring systems have not been validated. Many believe that the endoscopic assessment of edema and erythema is so subjective that systems that rate them are liable to serious error. When scoring systems are used it is usually impossible to identify which patients had the kind of chronic ulcers (as distinct from acute ulcers or erosions) whose progress on treatment interests the clinician. For these and other reasons the majority of investigators currently conduct studies that attempt to separate "ulcers" from lesser lesions, employing criteria for an ulcer similar to those of Larkai et al. (5) who state: "ulcers are three-dimensional, circumscribed mucosal defects, associated with an exudate, and having a diameter of 5 mm or more." In different studies, various diameters (e.g., 3, 4, 5, or 6 mm) have been used to separate erosions and ulcers, but the smaller the figure the less reliable it is in identifying a true ulcer. In my view, studies that use 3- and 4-mm criteria describe too many erosive lesions as "ulcers"; the 5-mm criterion is probably acceptable if the lesion is "three dimensional," but flat lesions up to 7-8 mm may still be "erosive" in nature, and may heal much more easily than deep chronic ulcers. However, there appear to be a number of differences between erosions and chronic ulcers, which warrant their being regarded as distinct lesions, associated with different natural histories, clinical outcomes, and responses to therapy. These are summarized in Table 1. Perhaps the most important point to be made is that erosions occur with the onset of therapy, heal rapidly if drug is discontinued, or gradually dimin-
NONSTEROIDAL DRUG-INDUCED ULCERS
Table 1. Comparison of Erosions and Chronic Ulcers Erosions
Shallow, no scarring Onset immediate (15 min) Caused mainly by topical injury Very rare with oral salsalate, nonacetylated salicylates, diflunisal, and pro-drugs. Usually multiple Diffuse, all over stomach Usually associated with mucosal hemorrhages Symptoms absent Lessen with adaptation Correlate strongly with anti platelet activity (6)
Deep, scarred Onset delayed (weeks-months) Caused mainly by systemic effects Occur in time with all NSAIDs and all routes of administration Often solitary «3) Mainly distal stomach Mucosal hemorrhages usually absent 20%-50% symptomatic Occur despite adaptation Correlate more with inhibitory effect on mucin synthesis (7)
NSAIDs, nonsteroidal antiinflammatory drugs.
ish with time-on-drug as mucosal adaptation occurs. But despite this; chronic ulcers develop in NSAID users, often in the antral portion of the stomach, and emerge clinically after 1-3 mo of NSAID treatment. Furthermore, there are numerous NSAIDs available, including salsalate, diflunisal, enteric-coated aspirin, and pro-drugs such as sulindac (which has no antiinflammatory or other biologic action until after it has been absorbed and biotransformed by the liver), which cause very little or no initial erosive injury. Despite this, solitary or multiple chronic antral ulcers develop with all of these drugs in time in some subjects, and sulindac has emerged as a leading cause of upper gastrointestinal bleeding within 30 days of starting NSAID therapy (8). There is no credible direct evidence that antral ulcers develop from acute erosions. From the previous section it will be apparent that efforts to examine the prevalence of ulcers in NSAID users and the outcome of therapy with different drugs over the past 10-20 yr suffer from the lack of a clear and standardized operational definition of an ulcer. The reviewer is forced to accept the diagnosis of an ulcer at face value, as no criteria were defined, applied, or published for most of the studies. The literature on the subject is inadequate and is mostly comprised of studies that confuse prevalence and incidence. Studies were often retrospective, uncontrolled, unblinded, and unrandomized. They often combined distinct lesions under such headings as "peptic ulcers," "mucosal lesions," or "NSAID gastropathies," they pooled patients using multiple drugs in varying dosages, and they ignored other risk factors (e.g., age, smoking, alcohol consumption, history of ulcer disease), resulting in data that defy analysis. I have therefore based this review on those studies from which relatively clear data can be
abstracted. This applied both to conclusions about prevalence and to evaluations of responses to drug therapy. Conclusions drawn about a particular study may differ from those of the authors.
Prevalence of Ulcers The percentage of subjects taking long-term NSAIDs who have an ulcer at any given time (point prevalence) is given in Table 2. None ofthese studies (5,9-16) were truly prospective, in that the diagnostic test used to detect the ulcer was not applied to the subjects before starting NSAIDs. Age- and sex-specific prevalence rates for peptic ulcer (PU), duodenal ulcer (DU), or gastric ulcer (GU) in subjects with rheumatoid or other forms of arthritis are not available but may differ from rates in the normal population (17). The data do not establish that the observed ulcers developed during the study, or that their development was causally related to the use of the drug. Nevertheless, the data are striking. The crude prevalence estimate for GU in 13% of arthritic NSAID users (Table 2), when compared with the GU prevalence of 0.28% in the normal population (18), yields a relative risk of x46, i.e., a 46-fold greater chance of GU in the group. The crude prevalence estimate for DU of 11%, when compared with a normal population prevalence of 1.4% (18), yields a relative risk of X 8. Thus, if NSAID use is a causal factor, the user is about six tim~s more likely to develop GU than DU in consequence of NSAID therapy. There are many uncertainties that weaken such a conclusion, including the following facts. Duodenal ulcer has a considerable prevalence in the population (life-time prevalence, -10%); is a chronic recurrent disorder with many of the recurrences being asymptomatic; may have antedated the start of NSAID therapy; may have been activated by NSAIDs or by the stress of the disease that led to their use; or may have recurred for other reasons unrelated to NSAID use (e.g., smoking) during the period of NSAID therapy. For these and other reasons the available data must be regarded with caution at the present time. Studies of NSAID use in patients with ulcers or ulcer complications also strongly support an increased relative risk of ulcer disease and complications in the NSAID user (19,20). In the case of aspirin, data from Cameron (21) indicated that the risk of developing GU was dose-related, with consumption of 14 or less tablets per week being associated with no increase in ulcer risk. Only when consumption exceeds 22 tablets per week does the increased risk achieve statistical significance (22). A study from Boston also observed a similar threshold
GASTROENTEROLOGY Va\. 96, No. 2, Part 2
at 14 tablets per week (23). The capacity of other NSAIDs to cause chronic ulcers may also be doserelated, but so far such dose-responsiveness has only been shown in the case of acute injury due to ibuprofen (24) or for the risk of gastrointestinal bleeding (cause undetermined) from a number of NSAIDs (25). Although there is a considerable body of epidemiologic evidence linking consumption of aspirin with the development of GU, so far there is little epidemiologic evidence to suggest that aspirin or other NSAIDs playa part in causing DU (22). However, the high prevalence of DU in many arthritic NSAID users, and the fact that ulcer complications in NSAID users are as commonly complications of DU as of GU (19,20,26), suggest some important connection between NSAIDs and DU or its complications. It may be that those with a history of DU, a scarred crater, or an active DU at the start of therapy are at increased risk for an adverse outcome of their ulcer disease (e.g., reactivation, exacerbation, or complication of the disease), accounting for the high prevalence noted above. Several studies have indicated that a previous history of PU disease appears to be a risk factor for developing ulcer complications during NSAID therapy.
Incidence Studies Despite much .current interest, prospective studies of the incidence of developing an ulcer during NSAID administration have been very few. The ideal study should submit all patients to endoscopy before starting NSAID therapy; follow established methods of blinding, randomization, and compliance monitoring; and submit all patients to repeat endoscopy at predetermined intervals for at least 2-3 mo (preferably 6-12 mol. Published studies fall short of these objectives, but cast some light on the dimensions of the problem. In 1980 Caruso and Bianchi-Porro (26) followed 164 patients with rheumatoid arthritis and 85 with osteoarthritis, who were being treated with 12 different NSAIDs. All had normal endoscopy before entry. Endoscopy at 3, 6, and 12 mo disclosed development of "gastric lesions" in 31%, with both types of arthritis patients being similarly affected. In patients receiving a single NSAID the incidence was 23%, and in those receiving two or more drugs it was 51%. It is not possible to determine what proportion of these lesions were ulcers or erosions. No data on duodenal lesions were presented. Later in 1987, in a prospective study of 20 normal volunteers taking aspirin (650 mg q.i.d for 2 wk), 11 of 20 (55%) developed gastric "ulcers" and 8 of 20 (40%) developed duodenal "ulcers," with erosions also being present at 2
1987 1988 1988
Larkai et al. Farah et al. Yeomans et al.
1974 1981 1987 1988 1988
1980 1981 1981
Lockard et al. Morris et al. Morris et al.
Sun et al. Gerber et al. Roth et al. Farah et al. Yeomans et al.
1987 1988 1988
Larkai et al. Farah et al. Yeomans et al.
Sun et al. Gerber et al.
1974 1979 1981 1981 1981
Sun et al. Silvoso et al. Gerber et al. Morris et al. Morris et al.
16 13 14
5 13 14
15 12 12
5 13 14
9 10 11 12 12
X-ray X-ray Endoscopy Endoscopy Endoscopy
Endoscopy Endoscopy Endoscopy
Endoscopy Endoscopy Endoscopy
Endoscopy Endoscopy Endoscopy
X-ray Endoscopy X-ray Endoscopy Endoscopy
Various Various Various Various Various
Aspirin Aspirin Aspirin + 2nd drug Various Various Various
Various Various Various
Various Aspirin Various Aspirin Aspirin
65 (Various) 230 (Various) 50 (Various)
56 (Various) 39 (RA) 26 (RA)
140 (RA) 225 (RA)
140 (RA) 225 (RA) 131 230 50 (Various)
Total 831 In some 50% Some No No
No 4/230 No
No No Some
(RA) (RA) (RA) (RA) (RA)
65 (Various) 230 (Various) 50 (Various)
140 82 225 39 26
Total 857 In some 50%
No 4/230 No
In some No 50% No Yes
28% 19% 19% 31% 20%
Crude prevalence rated
Sum of GU + DU; also includes study (16) where author quotes only data for "peptic ulcer"
Crude prevalence rated
11%d 91 39 43 27
Many asymptomatic Risk increased by smoking Risk increased by smoking
Mostly in NSAID + steroid cases Steroid vs. no steroid; p < 0.05, NS All kinds of aspirin similar Computed from authors' data Computed from authors' data
Crude prevalence rated
Many on several drugs Focus on stomach only Symptomatic patients only Some patients taking low-dose prednisone ± other drugs; all on aspirin Most ulcers asymptomatic Smoking increased risk Smoking increased risk
1.5% 16% 2%
4% 8% 0%
14% 15% 18%
9% 17% 10% 15% 8%
1 37 1
2 3 0
9 35 9
12 14 23 6 2
No. of ulcers
DU, duodenal ulcer; Dx, diagnosis; GU, gastric ulcer; NSAIDs, all nonsteroidal antiinflammatory drugs, including aspirin and nonacetylated salicylates; PU, peptic ulcer; RA, rheumatoid arthritis. Studies that described pathology in terms of lesions, lesion scores, erosions, erosions-complete and incomplete have been omitted. Only numbers of ulcers are given; prevalence of other lesions is considerably higher. For the most part these studies provide no criteria for use of the term "ulcer," do not establish activity of the ulcer, do not give indications of lesion size, and may in some cases erroneously include erosions. b No data on dosage are provided, nor are doses constant. Dose-response effects on prevalence are not clear. "Various" refers to other musculo-skeletal disorders, mostly osteoarthritis. C Where specified, prepyloric ulcers have been grouped with gastric ulcers. The numbers so specified were small. d Crude prevalence estimates were calculated by expressing, as a percentage of total patients, the numbers of the specified type of ulcers, adding up ulcers from all analyzable studies. e This term is used to imply the total number of deep injuries. It does not imply that these are of peptic etiology.
No. of patients (DX)
Table 2. Ulcer Prevalence Observed in Arthritis Patients During Long-Term Treatment With Various Nonsteroidal Antiinflammatory Drugs
wk in most subjects (27). The term ulcer was used for a lesion >3 mm in diameter (27). Depth was not measured or used as a diagnostic criterion. It is probable that many of these lesions were large erosions or small acute ulcers, not comparable to the kinds of chronic ulcers described in the prevalence studies. These two prospective studies, showing a high incidence of development of persistent gastric and duodenal lesions in both normal control and arithritic patients observed for both long (26) and short (27) periods of follow-up, strongly support a role for NSAIDs in causing erosions in most cases, and "ulcers" in a clinically significant minority. Several on-going studies are adpressing these issues with greater precision.
Traditional Therapy In the light of current knowledge one can fairly say that previous studies were not of high quality and the numbers of patients treated not large. Despite limitations, each of the studies yielded some useful information. Reviewing them chronologically helps the reader to appreciate the growth in our awareness of the issue over the past 10 yr. Study 1. In 1978, a National Institutes of Health study section, troubled by ethical questions as to whether or not one should continue NSAID therapy or stop it if subjects developed an ulcer, prompted a study by Gerber et fll. (11). All patient records of the Clinical Center, National Institutes of Health, from 1953 to 1975, listing the diagnosis of rheumatoid arthritis (225 patients) and the dual diagnoses of rheumatoid arthritis and PU (41 patients) were reviewed (Table 2). The ulcers were diagnosed by x-ray (only when symptomatic;) and treated with various antacids in various doses. Healing was deemed successful if all symptoms and radiographic findings had disappeared within 6 rna of diagnosis. In 35 of 41 cases treatment response of the ulcer could be evaluated. Most patients were taking aspirin, more than 50% were also taking steroids, and a minority were taking various doses of other NSAIDs at different times. Eight patients (9 ulcers: 8 GU, 1 DU) stopped taking NSAIDs, and all healed on antacid therapy. Twenty-seven continued on NSAID~ and antacids. Of these, 21 patients (22 ulcers: 10 GU, 12 DP) healed, 5 patients (4 GU, 1 DU) did not, and one "GU" proved to be a gastric cancer. Among those who continued NSAIDs, 1 patient (3.7%) died of hemorrhage following ulcer surgery, and 3 patients (11%) had occasional microbleeding during follow-up. Nevertheless, the data indicated that in most cases GUs and DUs were successfully treated with antacids, without stopping NSAIDs. Overall healing was 85%.
GASTROENTEROLOGY Vol. 96. No.2. Part 2
Study 2. In another study in 1981, LoIudice et al. (28) randomized 70 patients with NSAID-induced GU to either cimetidine (300 mg q.i.d.) plus antacid (7 doses, 854 mEq/day) or placebo plus the same dose of antacid. Nonsteriodal antiinflammatory drugs were discontinued. Repeat endoscopy at 6 wk showed 66% healing for the combination and 25% for antacids alone. Patients unhealed on combination therapy were sent to surgery, whereas patients on antacids alone continued treatment to 12 wk. By this time healing on antacid alone had reached 67%, and total healing for both groups had reached 79%. Note the shorter period of treatment compared with the study of Gerber et al. Healed patients were restarted on NSAIDs and cimetidine was continued. At 1 yr symptomatic recurrences were 6.5%. From these two studies antacids alone or combined with cimetidine appeared effective, whether or not NSAIDs were discontinued. Study 3. At about the saIIle time C:;rol
Despite the poor initial response, therapy was continued to endoscopic healing; 17 of 18 ulcers healed with 6-26 mo of therapy. One went to surgery. Slow healers did not appear at increased risk of recurrence. Study 6. Despite growing awareness that GUs healed slowly, Davie~ and colleagues (32) in 1986 carried out a 6-wk study in 27 ulcer patients (21 GU, 6 DU) randomized to cimetidine (300 mg q.i.d.) plus P.R.N. antacid (dose not recorded) or placebo plus P.R.N. antacid (dose not recorded). Nonsteroidal antiinflammatory drug therapy was continued. At 6 wk antacid healing was 50% versus 65% with the combination. The differences were not significant and treatment was not continued. Study 7. In an effort to resolve some of the imperfections of earlier studies, Manniche et al. (33) in 1987 carried out a randomized, endoscopic, single-blind study in 67 patients with ulcers (diameter >3 mm: 57 DU, 14 GU, 2 GU + PU), who were receiving 11 different NSAIDs and who were treated for 9 wk with either ranitidine (150 mg b.i.d.) or sucralfate (1 g bj.d.). Nonsteroidal antiinflammatory drugs were continued in half and stopped in half the cases. Patients not healed at 9 wk were crossed over to the other drug and treated for another 3-9 wk; ultimately 59 of 62 (95%) healed. When ulcer therapy was discontinued 14 of 59 (23%) relapsed in 1 yr, 9 while taking NSAIDs and 5 after discontinuing the drugs. However, in view of the larg~ excess of DU patients in the trial, and uncertainty as to how long they were on NSAIDs before entry, the 9-wk results bear further analysis. 25/30 (83%) healed on sucralfate 27/32 (84%) healed on ranitidine 29/32 (91%) healed with NSAID stopped 23/30 (77%) healed with continued NSAID 6/7 (86%) with GU healed with NSAID stopped 3/6 (50%) with GU healed with NSAID continued This was a competently executed study, but the number with GU was very small. Five of the GUs were >10 mm in diameter. For these reasons, type II statistical error may be masking a tendency for slow healing or nonhealing of GU when NSAID is continued, the trend being similar in studies 1, 2, and 5. This needs further evaluation. Study 8. In contrast to the apparent efficacy of sucralfate in treating small DUs not clearly related to NSAIDs, a recent study by Caldwell et al. (34) of 229 patients on long-term NSAIDs with mucosal damage failed to show much benefit. Sixty patients had GUs or erosions (number of ulcers not specified). These were treated with sucralfate (1 g q.i.d) or placebo, with the NSAIDs continued. Data were expressed as
NONSTEROIDAL DRUG-INDUCED ULCERS
lesion scores, which were calculated after 2- and 6-wk courses of th~rapy. Lesion scores were unchanged between 2 and 6 wk in the placebo group and showed only a small change in the mean lesiQn score, from 3.50 to 2.46, between 2 and 6 wk (p < 0.005) in those given sucralfate. However, a direct comparison of the mean decrease in score for patients treated with sucralfate versus placebo was not significant. This study casts doubt on the efficacy of sucralfate in healing gastric lesions. The numbers of patients with ulcers could not be abstracted. Study 9. In 1988, Farah et al. (13) published a report of an open, nonrandomized, uncontrolled, endoscopically monitored study of the response to PU treatment ip 56 patients with arthritis (29 GU, 27 DU), who Were treated with ranitidine (150 mg b.Ld.) for 12 wk with NSAID therapy continued. "Ulcer" was not defined nor ulcer size measured, but the prevalence of PU in the population studied [80 of 230 (35%)] was high, suggesting perhaps the inclusion of some erosions in addition to ulcers. Ranitidine healed 26 of 29 (90%) patients with GUs and 23 of 27 (85%) patients with DUs in 12 wk, results comparable to those seen in non-NSAID users. It is of interest that smoking appeared to be a major factor predisposing to both GUs and DUs in patients with all forms of arthritis. Study 10. The most recent study (35,36) is an open, prospective, uncontrolled, multicenter Euro" pean study of 187 patients on NSAIDs (median drug exposure, 8 wk), 79 of whom had ulcers (60 DU, 19 GU) and an additional 84 of whom had erpsions (56 with gastric erosions, 28 with duodenal). Patients were treated with cimetidine (800 mg h.s.) for 4-8 wk with NSAIDs continued. In the case of gastric lesions, by 8 wk 14 of 19 ulcers (74%) and 53 of 56 erosions (95%) were healed. In the duodenum 51 of 60 ulcers (85%) were healed and 28 of 28 erosions (100%) were healed at 8 wk. The overall PU healing rate was 81% at 8 wk. However, GU healing (75%) with cimetidine at this lower dose (800 mg h.s., 8 wk) did not appear to have been as good as that achieved with cimetidine at higher doses (1.0-1.2 g/day, 12 wk; see studies 1, 3, and 7). After healing 113 patients entered a maintenance phase, taking 400 mg of cimetidine at night and continuing NSAID therapy. The symptom!ltic re" lapse rate (a figure of dubious value given the high proportion of asymptomatic ulcers caused by NSAID) was 12% at 6 mo, a relapse rate similar to that observed in patients with uncomplicated ulcers off NSAIDs. This figure (12% at 6 mol should not be taken as establishing the efficacy o~ low-dose ma~ntenance therapy in patients using NSAIDs. One report has described 2 patients on indomethacin who perforated DUs during maintenance cimetidine therapy (37).
Study 11. An additional recent study by Roth et al. (16) is difficult to evaluate because of methodologic complexity. Many patients were taking, in addition to NSAIDs, other drugs that might adversely affect healing (e.g., penicillamine, methotrexate, steroids) and throughout the study NSAID doses were often adjusted by attending physicians (without restriction). Furthermore, at entry the study included patients with erythema and erosions without regard to symptoms, and excluded patients with normal mucosa and those with ulcers. It is thus difficult to generalize from this study. However, the study does suggest that, during 10 mo of a maintenance trial of 400 mg of cimetidine h.s. versus placebo h.s., progression to frank PU occurred in 2 of 12 (16%) of those on cimetidine and in 2 of 14 (14%) of those on placebo (p > 0.05, NS). The figures of 14% (placebo) and 16% (cimetidine) at 10 mo are not different and are comparable to the European maintenance recurrence rate of 12% at 6 mo (36). Parenthetically, the number of such recurrences may be less important than their severity and their association with complications such as hemorrhage, perforation, and death (38). Conclusions-Therapy for Chronic Ulcers Most gastric and duodenal ulcers heal with adequate doses of antacids or H2 -antagonists, or a combination of both drugs, in 12 w~ or less. Large GUs may take longer to heal. Continuing NSAID therapy does not appear to prevent the healing of DUs or small GUs, but it cannot yet be concluded that continuing NSAID therapy during ulcer treatment has no adverse effects [e.g., delayed healing or increased risk of complication(s)], particularly on large GUs. Additional studies are needed to address this issue. Duodenal ulcers appear to be less commonly associated with NSAID use, but to heal more easily than GUs (including prepyloric ulcers). Initial ulcer size affects the time required to heal GUs. Variations in the size of ulcers included may explain the apparent variations in drug efficacies between different studies, particularly when small numbers of GUs are involved. To date only full doses of antacids or H2 -antagonists have been clearly shown to be effective in healing ulcers and in preventing recurrences. There are no data establishing a clinically relevant efficacy for sucralfate in treating chronic GUs caused by NSAIDs, but DUs appear to heal normally on the drug.
Prophylaxis for Chronic Ulcer From studies 1, 3, and 10, data suggest that when ulcers heal on H2 -antagonists, continued ther-
GASTROENTEROLOGY Va!. 96, No.2, Part 2
apy with the healing agent is associated with a low recurrence rate (6%-16% over 1 yr). One might therefore expect that coadministration of H2 -antagonists and NSAIDs, from the start of NSAID therapy would prevent development of ulcers during longterm treatment. This has not been established. Berkowitz et al. (39) have shown, in a 4-wk study in 43 healthy volunteers given aspirin (650 mg q,i.d.), and either ranitidine (150 mg b,i.d.) or placebo, that ranitidine significantly reduced punctate and confluent hemorrhages in the stomach but not erosions. Erosions were significantly reduced in the duodenum, where they were the predominant lesion. Interestingly only 1 of the 43 subjects was judged to have a DU, in striking contrast to the study of Kimmey et al. (27) who used the same dose of aspirin and claimed that at 14 days 40% of patients treated with aspirin and placebo had DUs, a very high figure. In a second long-term prophylactic study, Lanza et al. (40) used ranitid~ne (150 mg b,i.d.) or placebo in 119 arthritic patients being treated with five different NSAIDs. On follow-up endoscopy (using a lesion scoring system) there were no significant differences in gastric damage at week 4 or week 8. However, at both times the duodenum was significantly protected. Altogether, 4 of 57 patients on placebo (7%) and no ranitidine patient developed DUs, as judged by an experienced investigator. Finally, Ehsanullah et al. (41) undertook a randomized, double-blind, endoscopically monitored, multicenter trial of ranitidine (150 mg b.Ld.) versus placebo (b,i.d.) in 263 arthritic patients free of lesions on entry and treated with one of four NSAIDs (naproxen, piroxicam, diclofenac, or indomethacin). Again, overall protection by ranitidine was better in the duodenum, with DU incidence being 1.5% in the ranitidine-treated group versus 8% in the placebo group (p = 0.016) at 4 or 8 wk. However, GU incidence was 6% in both groups at week 8. Conclusions-Chronic Prophylaxis These studies suggest that coadministration of full doses of ranitidine with NSAIDs protects against chronic duodenal but not gastric injury. All three studies (39-41) appear of high quality and claim a lower incidence of true ulcers than suggested in many other papers for aspirin or other potent NSAIDs. However, despite fewer ulcers and more erosions, gastric protection during continued NSAID administration was not achieved.
Prophylaxis-Acute Studies The apparent discrepancy between the efficacy of H2 -antagonists in treating established gas-
tric erosions or ulcers, and their inability to prevent such lesions developing during 2 mo of NSAID therapy, prompts a brief review of their efficacy in acute prophylactic studies. The studies are summarized in Table 3. All but two of the six Hz-antagonist studies used aspirin as the damaging NSAID, a point of some importance as its effects are very pH-dependent and topical, features that do not apply to other NSAIDs. In general, gastric injury predominated over duodenal injury. For this reason (i.e., the paucity of duodenal lesions ) it is more difficult to assess protection against acute duodenal injury by any of the drugs, and failure to observe it more commonly may simply be due to small numbers of subjects with relatively little injury. With these reservations some generalizations can be stated. Cimetidine dose-dependently reduces acute gastric and duodenal damage due to aspirin. Benefit in the duodenum was greater in two studies. When conventional doses of cimetidine (300 or 400 mg) are taken with the aspirin, the effects reducing acute injury are marked and immediate. The effects of ranitidine appear similar, but numbers of treated subjects are small. Although cimetidine significantly lessened the injury due to indomethacin, the benefit was much less than observed in studies of aspirin injury. When tolmetin was the offending drug, cimetidine protected 63% of subjects but to a lesser extent than observed with misoprostol (protected 93%) in the same study. The latter was, given in a dosage equal in acid-inhibitory potency to the dose of cimetidine used. Protection by sucralfate was not impressive. Of the four studies lasting 5-7 days, statistically significant protection of the stomach was not achieved in three and in the fourth was weak. Significant protection of the duodenum was observed or inferred in two studies. In comparative trials, the protection achieved was inferior to that seen with Hz-antagonists or the prostaglandin drug misoprostol. Conclusions-Acute Prophylaxis These studies suggests that Hz-antagonists afford significant protection, particularly against severe gastroduodenal injury due to aspirin (28). However, their efficacy may not extend to other NSAIDs or equal that of prostaglandins, which appear to possess a therapeutic advantage not attributable to acid secretory inhibition. Sucralfate (Table 3) appeared to possess the weakest protective action of all the compounds studied, and its use cannot be advocated based on human endoscopic studies published to date (total treated, 51 subjects). There is no evidence that the efficacy of any of the drugs in chronic prophylaxis or in treating chronic ulcers is predict-
NONSTEROIDAL DRUG-INDUCED ULCERS
able from its prophylactic efficacy in acute injury studies. Adequate follow-up studies linking acute and chronic effects of NSAIDs are lacking. Furthermore, the use of NSAIDs that avoid initial acute erosive injury (as is possible with salsalate, diflunisal, and "pro-drugs" such as sulindac) does not protect against the development of a solitary chronic ulcer at a later time. With the apparent efficacy of Hz-antagonists in acute prophylaxis, and in the treatment of chronic ulcers, their apparent failure to prevent GU in chronic prophylactic studies is puzzling at this time.
Role of Mucosal Adaptation Whereas Graham et al. (51) propose that gastric mucosal adaptation occurs in response to chronic aspirin therapy, others have failed to find it in humans using the same doses of aspirin (28,39). In a recent study (52) Graham et al. have advanced evidence that the time needed for adaptation may be longer when the dose of aspirin is increased, and acknowledge that the importance of adaptation may be mainly at lower doses of the drug. However, at any dose the adaptive process, which must follow aspirin injury, may contribute to the healing rate observed in drug efficacy trials. This contribution to healing may extend to other NSAIDs (53). If acute prophylaxis lessens or abolishes initial NSAID injury, either because it protects the mucosa or greatly reduces mucosal uptake of the damaging drug (e.g., aspirin), full adaptation may not occur, and healing may not be stimulated. There is some evidence from Domschke and coworkers (54,55) that adaptation may be altered with antacid therapy. Using the hemoglobin washout method to measure gastric microbleeding in healthy volunteers who were being started on aspirin therapy, they divided subjects into two groups of 6, one group to receive coadministration of aspirin and Maalox-70 (Rorer Pharmaceuticals Inc., Fort Washington, Pa.; Maalox-70 is not currently approved by the Food and Drug Administration for use in the United States), and one group to receive aspirin and placebo. Initial injury (day 3) was more severe in those on aspirin alone, but by day 28 microbleeding had dropped back almost to preaspirin levels. In the antacid plus aspirin group, injury (microbleeding) was less initially (day 3), was almost back to basal by day 14, and was at basal on day 28. On day 28, antacids were stopped and aspirin continued. By day 29, microbleeding continued near basal in the aspirin-alone group, but rose to the highest observed level (higher than day 3 in either group) in those who had had a month of Maalox-70. Thus, those cotreated with antacid, although bleeding less, had not devel-
(c) 200 mg
(b) 200-400 mg
400 mg b.i.d.
(a) 10-400 mg
150 mg b.i.d.
Patients with skeletal disorders
50 mg t.i.d.
650 mg q.i.d.
400 mg Li.d .
650 mg q.i.d.
900 mg t.i.d.
Findings and comments
(a) Dose-response. Four hundred milligrams of cimetidine p.o. 1 h before aspirin administration protected stomach in all cases. Between 10 and 200 mg, a dose-response effect existed. Two hundred milligrams of cimetidine protected > 75% of cases with 70%-80% confidence. (b) Dose ranging. Eight dose regimens were studied. These established that 200 mg of cimetidine taken with the aspirin was the lowest effective protective dose. (c) Placebo-controlled trial. Cimetidine and aspirin reduced mucosal injury in 14 of 20 subjects (70%) compared to 0 of 10 (0%) subjects by placebo, i.e. , compared to aspirin alone (p < 0.001). The end point was of gastric hemorrhages 2 h after aspirin administration. Using a lesion score of 2 or less as "successful protection," cimetidine protected gastric mucosa in 19 of 30 subjects (63%) versus 8 of 30 (27%) subjects for placebo. However, 200 JLg of misoprotol q.i.d. protected 27 of 29 (93%) subjects and achieved significantly better lesion scores than cimetidine (p < 0.001) . One cimetidine- and 7 placebo-treated subjects worsened and were withdrawn. Cimetidine in the remainder protected strongly against both gastric and duodenal ulcers, and severe gastric and duodenal erosions. Milder lesions were equal in both groups; those > 3 mm in diameter were " ulcers."
Cimetidine-treated patients (10) had significantly fewer total gastroduodenal lesions and antral lesions than placebo-treated (10) patients. Duodenal injury was less but similar in both groups. Ranitidineiaspirin-treated subjects had significantly less mucosal damage in stomach (p < 0.01) and duodenum (p < 0.05) than placebo/aspirin-treated subjects. This came from lowering hemorrhages; erosions did not differ. Prevention of severe erosions was greater in duodenum than in stomach. Compared to placebo, cimetidine decreased the number of patients with mucosal damage from 33% to 27% (NS) and significantly decreased mucosal lesion scores (stomach and duodenum combined) in those with injury, but significant damage remained compared to normal mucosa.
Table 3. Prevention of Acute Mucosal Nonsteroidal Antiinflammatory Drug Injury by Hz-Antagonists and Sucralfate
S1' Z ? !'"
0 r 0
0 t%l Z
NONSTEROIDAL DRUG-INDUCED ULCERS
oped mucosal adaptation. If the same kind of interaction occurred during cotreatment with aspirin and H 2 -antagonists, it might suppress the adaptive response and account for the disparity between the poor efficacy of Hz-antagonists in chronic prophylaxis, despite their efficacy in healing severe lesions in chronically injured (? adapted) mucosa. However, the puzzle remains, and other factors such as trial design, lack of blinding, or operation of unidentified risk factors (e.g., smoking) may be playing a significant role in the apparent discrepancy.
Effect of Therapy on Symptoms
As noted earlier, NSAID-induced ulcer symptoms are not predictably related to the presence of gastroduodenal lesions. Ulcer patients using NSAIDs seem to have fewer symptoms before complications than those not using the drugs (20,56-58). Among all patients using NSAIDs, up to 50% have symptoms (about half of whom have lesions) and about 1 in 8 patients discontinues therapy because of intolerable gastrointestinal side effects (35) . Regardless of the presence or absence of an ulcer, relief of ulcerlike symptoms in itself may be desirable so that NSAID therapy can continue. Whether suppression of symptoms, while continuing NSAIDs, will in the end prove beneficial or hazardous to the patient is not known at this time. Two studies cast some light on symptom relief attending the use of antiulcer drugs in NSAID users . In a European study of 127 lesion-free symptomatic patients on NSAIDs, 95% had upper abdomip.al pain, 75% had heartburn, and almost 60% had nausea; less than 1 in 5 had vomiting (35). In a randomized, controlled, double-blind study of cimetidine (400 mg b.Ld.) versus placebo for 4 wk, with NSAIDs continued, cimetidine was significantly better than placebo, causing complete remission in 72% of cases (49% in placebo p < 0.01) and marked improvement or better in 92% (72% in placebo p < 0 .01) (35) . When symptom-free patients were treated with cimetidine (400 mg h.s.) or placebo for 6 mo, relapses of symptoms occurred with equal frequency in the treated (11%) and control (13.5%) groups. Symptoms returned more rapidly in those on placebo, suggesting again that low-dose maintenance therapy may be suboptimal in this setting. Patients with lesions present on endoscopy also experienced significant symptom relief: 55% were asymptomatic at 4 wk, and 78% at 8 wk. This response is slower than that observed in lesion-free patients (36) . In another study (34) that included patients with and without endoscopically verified lesions, therapy with sucralfate (1 g q.i.d.) was also significantly superior to placebo in reducing symptoms when
patients were taking long half-life or nonsalicylate NSAIDs. Patients taking salicylates or short half-life NSAIDs did not benefit. Whether or not lesion status affected the outcome was not stated. So far, prostaglandin analogues have not been shown to be very effective in relieving symptoms, and antacids are not well studied. The marked amelioration of symptoms with time in the placebo groups in all of the published studies suggests that patients may develop symptomatic tolerance to NSAIDs similar to that observed in the case of gastric lesions or microscopic bleeding. However, at present gastrointestinal side effects of NSAIDs account for 42% of all medical admissions in patients with rheumatoid arthritis, with 15% requiring at least one admission to the hospital per year (50). The importance of treating NSAID users solely on the basis of symptoms warrants further prospective studies.
Clinical Implications Discussion of the management of "peptic" ulcers in patients who need to continue taking NSAIDs is limited here to recommendations regarding the antiulcer drugs available for use in the United States at the present time. Based on the published data, ulcers should be treated with full doses of Hz-antagonists, cimetidine (300 mg q.i.d.) or ranitidine (150 mg b.i.d.), alone or in combination with antacids, for at least 8 wk in the case of DU and for 12 wk in the case of GU (60). Pending welldesigned studies that establish their efficacy and safety in the patient taking NSAIDs, other forms of therapy (e.g., h.s. dosing) should be avoided. A case can be made for continuing indefinitely with full doses of drug in those who must continue NSAIDs and who are known to have ulcer disease or its complications, particularly if patients are over 60 yr of age (19,20). Low-dose maintenance therapy has not been proved to be satisfactorily safe and effective. There is as yet no basis for the indiscriminate long-term, prophylactic use of Hz-antagonists in ordinary patients requiring NSAID therapy. In the case of aspirin use, there may be some place for antacid therapy, but data are lacking for other NSAIDs. Only the weakest case can be made, based on published endoscopic studies, for any use of sucralfate in the current management of GU in the patient on NSAIDs. In short-term studies, cotherapy with NSAIDs and Hz-antagonists abolishes or significantly reduces ulcer symptoms in patients with or without lesions at endoscopy. As many trials are on-going, these recommendations may well change over the next few years. The use of Hz-antagonists, particularly cimetidine, seems well validated by
GASTROENTEROLOGY Vol. 96, No.2, Part 2
numerous studies, particularly for DU. It remains to be seen whether or not continued use of NSAIDs delays the healing of large GUs, but for the moment continuing these drugs seems reasonably safe in most cases.
References 1. Butt JH, Barthel JS, Moore RA. Clinical spectrum of the upper gastrointestinal effects of nonsteroidal anti-inflammatory drugs. Am J Med 1988;84(Suppl 2A):5-14. 2. Eliakim R, Ophir M. Rachmilewitz D. Duodenal mucosal injury with non-steroidal anti-inflammatory drugs. J Clin GastroenteroI1987;9:395-9. 3. Lanza F, Peace K, Gustitus L, Rack MF, Dickson B. A blinded endoscopic comparative study of misoprostol versus sucralfate and placebo in the prevention of aspirin induced gastric and duodenal ulceration. Am J Gastroenterol 1988;83:143-6. 4. Malfertheiner P, Stanescu A, Rogatti W, Dieschuneit H. Effects of microencapsulated vs. enteric coated acetylsalicylic acid on gastric and duodenal mucosa: an endoscopic study. J Clin GastroenteroI1988;10:269-72. 5. Larkai EN, Smith JL, Lidsky MD, Graham DY. Gastroduodenal mucosa and dyspeptic symptoms in arthritic patients during chronic nonsteroidal anti-inflammatory drug use. Am J Gastroenterol 1987;82:1153-8. 6. Rainsford KD. Relationship between drug absorption, inhibition of cyelo-oxygenase and lipoxygenase pathways and development of gastric mucosal damage by nonsteroidal antiinflammatory drugs in rats and pigs. In: Bailey JM, ed. Prostaglandins, leukotrienes and lipoxins. New York: Plenum, 1985:639-54. 7. Garner A, Allen A, Rowe PH. Gastroduodenal defense mechanism and the action of non-steroidal anti-inflammatory agents. Scand J GastroenteroI1987;22(Suppl 27):29-34. 8. Carson JL, Strom BL, Morse ML, et al. The relative gastrointestinal toxicity of the non-steroidal antiinflammatory drugs. Arch Intern Med 1987;147:1054-9. 9. Sun DCH, Roth SH, Mitchell CS, Eglund DWW. Upper gastrointestinal disease in rheumatoid arthritis. Dig Dis 1974;19: 405-10. 10. Silvoso GR, Ivey KJ, Butt JH, et al. Incidence of gastric lesions in patients with rheumatic diseases on chronic aspirin therapy. Ann Intern Med 1979;91:517-20. 11. Gerber LH, Rooney PJ, McCarthy DM. Healing of peptic ulcers during continuing anti-inflammatory drug therapy in rheumatoid arthritis. J Clin GastroenteroI1981;3:7-11. 12. Morris AD, Holt SD, Silvoso GR, et al. Effect of antiinflammatory drug administration in patients with rheumatoid arthritis. Scand J GastroenteroI1981;16(Suppl 67):131-5. 13. Farah D, Sturrock RD, Russel RI. Peptic ulcer in rheumatoid arthritis. Ann Rheum Dis 1988;47:478-80. 14. Yeomans ND, Elliott SL, Edwards J, Buchanan R, Sturrock D, Smallwood R. Gastroduodenal damage during therapy with nonsteroidal antiinflammatory drugs (abstr). Gastroenterology 1988;94:A510. 15. Lockard 0, Ivey KJ, Butt JH, Silvoso GR, Sisk C, Holt S. The prevalence of duodenal lesions in patients with rheumatic diseases on chronic aspirin therapy. Gastrointest Endosc 1980;26:5-7. 16. Roth SH, Bennett RE, Mitchell CS, Hartman RJ. Cimetidine therapy in nonsteroidal anti-inflammatory drug gastropathy. Arch Intern Med 1987;147:1798-801. 17. Atwater EC, Morgan ES, Wieche DR, Jacox RF. Peptic ulcer and rheumatoid arthritis. Arch Intern Med 1965;115:184-90.
18. Kurata JH. Epidemiology of peptic ulcer disease. Pract GastroenteroI1983;7:13-7. 19. Langman MJS. Ulcer complications and non-steroidal drugs. Am J Med 1988;84:15-9. 20. Armstrong CP, Blower AL. Non-steroidal anti-inflammatory drugs and life threatening complications of peptic ulceration. Gut 1987;28:527-32. 21. Cameron AJ. Aspirin and gastric ulcer. Mayo Clin Proc 1975;50:565-70. 22. Graham DY, Smith JL. Aspirin and the stomach. Ann Intern Med 1986;104:390-8. . 23. Levy M. Aspirin use in patients with major upper gastrointestinal bleeding and peptic ulcer disease: a report from the Boston Collaborative Drug Surveillance Program. N Engl J Med 1974;290:1158-62. 24. Lanza FL. Endoscopic studies of gastric and duodenal injury after the use of ibuprofen, aspirin, and other non-steroidal anti-inflammatory agents. Am J Med 1984;77:19-24. 25. Carson JL, Strom BL, Soper KA, West SL, Morse ML. The association of non-steroidal anti-inflammatory drugs with upper gastrointestinal tract bleeding. Arch Intern Med 1987; 147:85-8. 26. Caruso I, Bianchi-Porro G. Gastroscopic evaluation of antiinflammatory agents. Br Med J 1980;1:75-8. 27. Kimmey MB, Silverstein FE, Saunders DR, Frank WOo Timecourse of aspirin-induced gastroduodenal mucosal injury and gastro-intestinal symptoms: the influence of cimetidine (abstr). Gastroenterology 1987;92:1466. 28. LoIudice T A, Saleem T, Lang JA. Cimetidine in the treatment of gastric ulcer, induced by steroidal and non-steroidal antiinflammatory agents. Am J GastroenteroI1981;75:104-10. 29. Croker Jr, Cotton PB, Boyle AC, Kinsella P. Cimetidine for peptic ulcer in patients with arthritis. Ann Rheum Dis 1980; 39:275-8. 30. O'Laughlin JC, Silvoso GR, Ivey KJ. Healing of aspirin-associated peptic ulcer disease despite continued salicylate ingestion. Arch Intern Med 1981;141:781-3. 31. O'Laughlin JC, Silvoso GK, Ivey KJ. Resistance to medical therapy of gastric ulcers in rheumatic disease patients taking aspirin. Dig Dis Sci 1982;27:976-80. 32. Davies 1, Collins AJ, Dixon A St J. The influence of cimetidine on peptic ulcer patients with arthritis taking anti-inflammatory.drugs. Br J RheumatoI1986;25:54-8. 33. Manniche C, Malchow-Moller A, Andersen JR, et a1. Randomized study of the influence of non-steroidal anti-inflammatory drugs on the treatment of peptic ulcer in patients with rheumatic disease. Gut 1987;28:226-9. 34. Caldwell JR, Roth SH, Wu WC, et al. Sucralfate treatment of non-steroidal anti-inflammatory drug-induced gastrointestinal symptoms and mucosal damage. Am J Med 1987;83(Suppl 38):74-82. 35. Bijlsma JWJ. Treatment of endoscopy negative NSAID-induced upper gastrointestinal symptoms with cimetidine-an international multicenter collaborative study. Aliment Pharmacol Therap 1988;2(Suppl 2):75-84. 36. Bijlsma JWJ. Treatment of NSAID-induced gastrointestinal lesions with cimetidine - an international multicenter collaborative study. Aliment Pharmacol Therap 1988;2(Suppl 2):85-96. 37. Mitchell WS, Sturrock RD. Ulcers and anti-inflammatory agents. Br Med J 1982;284:731. 38. McCarthy DM. Duodenal ulcer: is there a role for chronic prophylactic therapy? In: Barkin J, Rogers A, eds. Difficult decisions in digestion. 1st ed. Chicago: Year Book Medical Publishers, 1988:95-107. 39. Berkowitz JM, Rogenes PR, Sharp JT, Warner CWo Ranitidine protects against gastroduodenal mucosal damage associated
NONSTEROIDAL DRUG-INDUCED ULCERS 673
with chronic aspirin therapy. Arch Intern Med 1987;147: 2137-9. 40. Lanza F, Robinson M, Bowers 1. Griffin J, Kogut D, Kogan F. A multi-center double-blind comparison of ranitidine vs. placebo in the prophylaxis of non-steroidal anti-inflammatory drug (NSAID) induced lesions in gastric and duodenal mucosae (abstr). Gastroenterology 1988;94:A250. 41. Ehsanullah RSB, Page MG, Tildesley G, Wood JR. NSAIDinduced gastroduodenal damage: effect of ranitiqine prophylaxis. Gastroenterology 1988;94:A111. 42. Stiel D, Ellard KT, Hills L1, Brooks PM. Protective effect of enprostil against aspirin-induced gastroduodenal mucosal injury in man. Am J Med 1986;81(Suppl 2A):54-8. 43. Berkowitz JM, Adler SN, Sharp JT, Warner CWo Reduction of aspirin-induced gastroduodenal mucosal damage with ranitidine. J Clin GastroenteroI1986;8:377-80. 44. Rachmilewitz D, Polak D, Eliakim R, Stalnikowicz R, Wengrower D. Cimetidine significantly decreased indomethacin induced gastroduodenal mucosal damage (abstr). Gastroenterology 1986;90:1596. 45. Kimmey MB, Silverstein FE, Saunders D~, Chapman Re. Reduction of endoscopically assessed acute aspirin-induced gastric mucosal injury with cimetidine. Dig Dis Sci 1987;32: 851-6. 46. Kimmey MB, Silverstein FE. Role of H 2 -receptor blockers in the prevention of gastric injury resulting from nonsteroid"l anti-inflammatory agents. Am J Med 1988;84(Suppl 2A):4Q52. 47. Lanza FL, Aspinall RL, Swabb EA, Davis RE, Rack MF, Rubin A. Double-blind placebo-controlled endoscopic comparison of the mucosal protective effects of misoprostol versus cimetidine on tolmetin-induced mucosal injury to the stomach and duodenum. Gastroenterology 1988;95:289-94. 48. Tesler MA, Lim ES. Protection from ASA induced gastric erosions by sucralfate. J Clin Gastroenterol 1981;3(Suppl 2): 175-9. 49. Wu WC, Castell DO. Does sucralfate protect against aspirin induced mucosal lesions? Yes & no (abstr). Gastroenterology 1984;86:1303. 50. Stern AI, Ward F, Hartley G. Sucralfate protects the human stomach from the damaging effects of aspirin. Am J Med 1987;83(Suppl 38):83-5. 51. Graham DY, Smith JL, Dobbs SM. Gastric adaptation occurs with aspirin administratic)ll in man. Dig Dis Sci 1983;28:1-6. 52. Graham DY, Smith JL, Spjut HJ, Torres E. Gastric adaptation: studies in humans using continuous aspirin administration. Gastroenterology 1988;95:327-33. 53. Kuwayama H, Eastwood G, Miyake S, Furokawa H. Adaptation of gastric mucosa to repeated indomethacin administration in the rat (abstr). Gastroenterology 1986;90:1506. 54. Hagel HJ, Wild H, Kachel G, Ruppin H, Domschke W. Adaptation of gastric mucosa to chronic acetylsalicylic acid and cytoprotection by antacid pretreatment (abstr). In: Proceedings of the Second International Congress of Gastroenterology (Lisbon), 1984:A614. 55. Domschke W, Hagel J, Ruppin H, Kaduk B. Antacid protection of gastric mucosa. Klin Wochenschr 1986;64(Suppl 7): 28-31. 56. Mellem H, Stave R, Myren 1. et a1. Symptoms in patients with peptic ulcer and hematemesis and/or melena related to the use of non-steroid anti-inflammatory drugs. Scand J GastroenteroI1985;20:1246-8. 57. Collins A1. Davies J, Dixon A St J. Contrasting presentation and findings between patients with rheumatic complaints taking non-steroidal anti-inflammatory drugs and a general population referred for endoscopy. Br J Rheumatol 1986;25: 50-3.
58. Morgan R. Non-aspirin non-steroidal anti-inflammatory drugs and active gastroduodenal ulceration in the elderly. Br J RheumatoI1987;;!6:158-9. . 59. Wolfe F, Kleinheksel SM, Spitz PW, et a!. A multicenter study of hospitalization in rheumatoid arthritis: frequency, medical-surgical admissions, and charges. Arthritis Rheum 1986; 29:614-9. 60. Howden CW, Jones DB, Peace KE, Bllrget DW, Hunt RH. The
GASTROENTEROLOGY Vo!' 96, No.2, Part 2
treatment of gastric ulcer with antisecretory drugs. Dig Dis Sci 1988;33:619-24. Received September 1, 1988. Accepted October 18, 1988. Address requests for reprints to: Denis McCarthy, M.D., Veterans Administration Medical Center (lllF), 2100 Ridgecrest Drive S.E., Albuquerque, New Mexico 87108.