The prevalence of endoscopically confirmed upper gastrointestinal ulcers in NSAID users has been reported to range between 15% and 31% [12-17]. Despite the predominance of gastric ulcers in chronic NSAID users, ulcer complications in NSAID users are often associated with duodenal ulcers [7, 11, 18-20]. Duodenal ulcers are more common than gastric ulcers in patients not taking NSAIDs, suggesting that many duodenal ulcers associated with NSAID use may reflect NSAID exacerbation of preexisting peptic ulcer disease [21, 22].

Misoprostol, a synthetic prostaglandin E₁ analog, has been shown to be effective in decreasing the incidence of gastric ulcers in chronic NSAID users [17, 23] without interfering with the antirheumatic effects of the NSAID. The H₂-receptor antagonist, ranitidine, and the topical agent, sucralfate, have not been effective in preventing gastric ulcers in long-term NSAID users [23-25]. Ranitidine-NSAID cotherapy decreased the frequency of duodenal ulcers in NSAID-treated patients, suggesting that duodenal ulcers in NSAID users may have a different pathogenesis than gastric ulcers [24, 25]. Our study was designed to investigate the efficacy of misoprostol for the prevention of NSAID-induced duodenal ulcers in arthritis patients receiving long-term NSAID therapy.

Methods

Top Methods Results Discussion Author & Article Info References

Patients were recruited from private practice offices, Veterans Affairs clinics, health maintenance organizations, and academic institutions. The study was done between June 1989 and February 1991. Patients were eligible to enter the study if they had rheumatoid arthritis, osteoarthritis, psoriatic arthritis, ankylosing spondylitis, or the Reiter syndrome. Upper gastrointestinal pain was not required for entry into the study, but patients were expected to require at least 3 additional months of daily NSAID therapy with either ibuprofen, piroxicam, naproxen, sulindac, tolmetin, indomethacin, or diclofenac.

Women were required to be postmenopausal, surgically sterilized, or practicing adequate contraception. Other exclusion criteria were a history of peptic ulcer disease requiring treatment in the 30 days immediately before entry, upper gastrointestinal malignancy, pyloric obstruction, acute hepatitis, pancreatitis, bleeding diathesis, upper gastrointestinal surgery within 30 days, or severe renal impairment. In addition, patients taking antineoplastic drugs, anticoagulants, anti-ulcer drugs other than the study drug, or prednisone at dosages of more than 7.5 mg/d were excluded.

Study Design

The study was a randomized, double-blind, placebo-controlled, multicenter trial of misoprostol, 200 µg given four times a day (with meals and at bedtime with food). Upper gastrointestinal symptoms were scored using a 0 to 3 grading scale; zero corresponded to no abdominal pain, 1 corresponded to mild pain, 2 corresponded to moderate pain, and 3 corresponded to severe abdominal pain.

Patients with gastric or duodenal ulcers of any size as well as those with erosions 0.3 cm in diameter at baseline upper gastrointestinal endoscopy were excluded. Patients who were accepted were randomized (randomization was balanced within each center) and received misoprostol or placebo within 72 hours of the entry endoscopic examination. Patients continued to take NSAIDs at the same dosage administered before the study. Patients were allowed to take up to three aluminum hydroxide antacid tablets (Amphojel [600 mg], Wyeth-Ayerst Laboratories, Philadelphia, Pennsylvania) per day for the first 2 weeks of the study for relief of upper gastrointestinal pain.

After 4 weeks (± 3 days), 8 weeks (± 5 days), and 12 weeks (± 5 days) of treatment with misoprostol or placebo, endoscopic examinations and upper gastrointestinal pain evaluations were repeated. Endoscopists remained blinded to the patient's medication. Patients were instructed to take the study medication the night before the endoscopy and not to take the next dose until after the procedure was completed. Noncompliance was defined as failure to take at least 60% of prescribed medication during the study period, determined by pill count at 4, 8, and 12 weeks of therapy. The protocol was approved by an institutional review board, and each patient gave written, informed consent.

Data Analysis

Any patient who received at least one dose of study medication qualified for analysis of the frequency of adverse experiences. The primary analysis for efficacy was of the intention-to-treat group. In addition, any patient who took at least 60% of the doses of study medication during the time of study participation, who had not missed more than 3 consecutive days of study medication, who satisfied the inclusion or exclusion criteria, and who had an initial and final endoscopic examination was considered evaluable for efficacy.

View larger version (22K): [in this window] [in a new window]   Figure 1. Percentage of evaluable and intent-to-treat patients with duodenal ulcers. Among evaluable misoprostol-treated patients, 0.9% developed a duodenal ulcer compared with 6% in the placebo group (P = 0.004). Among intent-to-treat patients in the misoprostol group, 0.6% developed a duodenal ulcer compared with 4.6% in the placebo group (P = 0.003).

The comparability of the misoprostol and placebo treatment groups with respect to demographic characteristics at admission was assessed for all patients enrolled in the study. These characteristics included age, sex, race, height, and weight. When expected cell sizes were adequate, nominal measures were compared using the Pearson chi-square test. Otherwise, the Fisher exact test was used. Height, weight, and age measures were compared in the two treatment groups using the Wilcoxon rank-sum test.

Treatment group differences in the proportion of patients who completed the study and the proportion of evaluable patients were assessed using the Pearson chi-square test. Patients who discontinued study participation without an endoscopy at week 4, week 8, or week 12 were classified as unknown for each time period at which their endoscopy status was undetermined. A chi-square analysis of duodenal ulcer development during the 12-week period was then conducted.

Secondary analyses using identical methods were done for gastric ulcer development and gastroduodenal ulcer development. The rates of developing ulcers presented above did not account for censoring (that is, patients who discontinued the study before endoscopic verification of outcome or before they finished the full 12 weeks of study participation). Therefore, rates of developing duodenal, gastric, and gastroduodenal ulcer were also compared in the two treatment groups using a life-table analysis, which provided actuarial method estimates of the rate of ulcer development. The grouping intervals used for the life-table were prestudy to week-4 endoscopy, week-4 to week-8 endoscopy, and week-8 to week-12 endoscopy. The actual number of days from the prestudy endoscopy to the week 4, week 8, and week 12 endoscopies was not used to construct the life-table intervals; consequently, they can only be considered nominal categories. The log-rank test was used to compare the survival distributions of the two treatment groups.

View larger version (22K): [in this window] [in a new window]   Figure 2. Percentage of intention-to-treat patients developing ulcers during the 12-week study period. The 95% CI for misoprostol and placebo groups, respectively: duodenal ulcer, 0.1% to 2.6% and 2.8% to 8%; gastric ulcer, 0.8% to 4.4%, and 5.1% to 11.4%; and gastroduodenal ulcer, 1.2% to 5.3% and 8.3% to 15.8%.

The qualitative investigator assessments (daytime and nighttime pain and upper gastrointestinal symptoms) were tabulated at each visit by treatment group for evaluable patients. The distributions of the investigator's severity ratings for daytime and nighttime pain were each compared for the two treatment groups at 12 weeks using the Fisher exact test. For patients who reported daytime or nighttime pain at the prestudy visit, changes from prestudy to week 12 in daytime and nighttime pain were categorized as better, same, or worse and compared in the two treatment groups using the Pearson chi-square test or the Fisher exact test, depending on the distribution of responses. For patients who did not report daytime or nighttime pain at the prestudy visit, the proportion of patients who developed daytime or nighttime pain at week 12 was compared in the two treatment groups using the Pearson chi-square test or the Fisher exact test.

The investigator's assessments of the individual upper gastrointestinal symptoms (nausea, heartburn, and diarrhea) were compared for the two treatments using the statistical methods outlined for daytime and nighttime pain. The proportion of patients experiencing at least one adverse event was compared for the two treatment groups using the Pearson chi-square test.

All statistical tests were two-sided with the nominal level of significance assessed at the 0.05 level. The percentages presented in some tables sometimes may not add up to 100% because of rounding.

Results

Top Methods Results Discussion Author & Article Info References

Demographics

Of 643 patients enrolled in the study, 320 were randomized to receive misoprostol and 323 were randomized to receive placebo. Six hundred thirty-eight patients (319 in each group) took at least one dose of the study medication and therefore qualified for analysis of the frequency of adverse events. Two hundred fifteen of the 320 (67%) patients randomized to receive misoprostol and 240 of the 323 (74%) patients randomized to receive placebo were evaluable for efficacy analysis (P = 0.05). Fewer patients were evaluable in the misoprostol treatment group as compared with the placebo treatment group primarily because of medication noncompliance (24% misoprostol versus 14% placebo) and missed final endoscopic examination (25% misoprostol versus 18% placebo). There was a significant overlap between these two categories, both of which included withdrawals secondary to adverse events.

Among the patients who were randomized (the intent-to-treat cohort), no statistically significant differences existed between the study groups with respect to demographic and clinical characteristics (Table 1). Combining both treatment groups, the mean age was 59 years (range, 21 to 88 years; 82% of the patients were 50 years or older). Forty-seven percent were women and 74% were white. Eighteen percent of all patients reported consumption of alcohol and 15% smoked more than 10 cigarettes per day.

The most common condition in both treatment groups was osteoarthritis (75% of all patients). Thirty percent of all patients used ibuprofen, 24% used naproxen, 17% used piroxicam, and 10% used sulindac at baseline. The two groups showed no significant differences in demographic and clinical characteristics in patients who were evaluable for efficacy (evaluable cohort).

Prevention of Ulcers

Analyzing the 643 patients who enrolled in the study, 2 of 320 (0.6%; CI, 0.2% to 2.6%) patients randomized to receive misoprostol developed a duodenal ulcer compared with 15 of 323 (4.6%; CI, 2.8% to 8%) patients randomized to receive placebo during the 12-week study period (P = 0.002; (Figure 1). Similar results were found in the evaluable patient group.

The assessment of gastric ulcer development indicated a significantly lower frequency in the misoprostol group than in the placebo group in both patient groups. For the intent-to-treat cohort, 6 of 320 (1.9%; CI, 0.8% to 4.4%) patients treated with misoprostol and 25 of 323 (7.7%; CI, 5.1% to 11.4%) patients treated with placebo developed a gastric ulcer during the 12-week study (P = 0.001).

Gastroduodenal ulceration rates for evaluable and for intent-to-treat patients also were lower in the misoprostol than in the placebo group (Figure 2). In the intent-to-treat group, 8 of 320 (2.5%; CI, 1.2% to 5.3%) misoprostol-treated patients and 37 of 323 (11.5%; CI, 8.3% to 15.8%) placebo-treated patients developed a gastroduodenal ulcer by week 12 (P < 0.001).

A subset analysis was performed on patients with and without a history of ulcers. Although this study was not designed for such an analysis and lacks statistical power, the results in both subgroups parallel the results of the overall study (Table 2). In the intent-to-treat cohort, the frequency of gastroduodenal ulcer for patients with a previous history of ulcer was 16.9% (14 of 83 patients) in the placebo group and 3.7% (3 of 81 patients) in the misoprostol group. In patients without a history of ulcer, the frequency of gastroduodenal ulcer was 9.6% (23 of 240 patients) for placebo-treated patients and 2.1% (5 of 239 patients) for misoprostol-treated patients.

To assess the effect of age, sex, alcohol consumption, smoking, type of arthritis, and baseline erosions on ulceration, stepwise logistic regression models were estimated as described in the Methods section. The results from the stepwise logistic regression of duodenal ulceration revealed that treatment (misoprostol compared with placebo) was the only statistically significant [P = 0.008] term retained in the final model. The two design variables representing number of duodenal erosions at entry versus no erosions as reference category) were retained in the final model, although the test for the baseline duodenal erosion effect was not statistically significant (P = 0.10). Age, sex, smoking, alcohol consumption, and type of arthritis were not statistically associated with duodenal ulceration. The odds ratio for treatment effect indicated that placebo patients were 9.3 times (CI, 2.0 to 43.2) more likely to develop a duodenal ulcer than were misoprostol patients.

The stepwise logistic regression analysis for gastric ulceration revealed statistically significant effects for treatment (P = 0.001) and alcohol consumption (P = 0.02). Age, sex, erosions, smoking, and type of arthritis were not significantly associated with gastric ulceration. The odds ratio associated with the treatment effect indicated that placebo patients were 4.1 times more likely to develop a gastric ulcer than were misoprostol patients (CI, 1.6 to 10.5). Patients who consumed alcoholic beverages were 3 times more likely to develop a gastric ulcer than were patients who did not consume alcohol (CI, 1.3 to 6.8).

The stepwise logistic regression for gastroduodenal ulceration examined the effect of age, sex, alcohol consumption, smoking, type of arthritis, and both duodenal and gastric erosions on gastroduodenal ulceration. Treatment was the only statistically significant (P < 0.001) term retained in the final model. Placebo patients were 4.9 times (CI, 2.2 to 10.7) more likely to develop a gastroduodenal ulcer than were misoprostol patients.

The results of the life-table analysis support the previous analysis that did not account for censoring (patients who discontinued the study before endoscopic verification of outcome or before they finished the 12-week study).

Upper Gastrointestinal Symptoms

At week 12 in the evaluable patient group, no statistically significant differences were found between the two treatment groups in assessments of daytime and nighttime pain, regardless of baseline ratings. Similarly, the distributions of the severity ratings for nausea and heartburn were similar in the two treatment groups at week 12. Daytime pain was present in 48% of patients receiving misoprostol and in 47% of patients receiving placebo at baseline. By week 12, 70% of these misoprostol patients and 75% of these placebo patients had a decrease in the severity of daytime pain, whereas 5% of the misoprostol patients and 9% of the placebo patients had an increase in the severity of daytime pain (P = 0.2).

Nighttime pain was present in 33% of misoprostol patients and in 37% of placebo patients at baseline. By week 12, 75% of these misoprostol patients and 77% of these placebo patients had a decrease in the severity of nighttime pain. A similar proportion (5%) of patients in each treatment group had an increase in the severity of nighttime pain (P = 1.0).

Adverse Events

Four patients in the placebo group and one patient in the misoprostol group did not take any study medication and therefore were not evaluable for adverse events; 319 patients in each treatment group were evaluable for these analyses. There was no statistical difference in the percentage of patients with one or more treatment-emergent adverse events in the two treatment groups (misoprostol, 68.0%; placebo, 70.2%), (P = 0.5). For individual adverse events, treatment-emergent diarrhea developed in 103 (32.3%) patients randomized to receive misoprostol and in 57 (17.9%) patients randomized to receive placebo (P < 0.001). Constipation occurred in 8 (2.5%) misoprostol-treated patients and in 23 (7.2%) placebo-treated patients (P < 0.01).

Thirty-eight patients in the misoprostol group and 34 patients in the placebo group dropped out of the study due to adverse events other than gastric ulcer. The frequencies of the most common adverse events that led to patient withdrawal from the study are listed in Table 3. A total of 23 patients (11 misoprostol [3.4%], 12 placebo (3.8%) reported serious adverse events; however, none of these events resulted in death. There were no clinically significant differences between the treatment groups in changes in any of the physical findings, vital signs, or laboratory values during the study.

Discussion

Top Methods Results Discussion Author & Article Info References

This study, along with two ranitidine studies [24, 25], shows that the development of duodenal ulcer is common in chronic NSAID users. The effectiveness of misoprostol cotherapy for preventing gastric ulcers in chronic NSAID users was confirmed [23]; misoprostol cotherapy (200 µg administered four times daily) markedly reduced the frequency of duodenal ulcers in arthritic patients receiving chronic NSAID therapy. Previous studies have shown that the H₂-receptor antagonist ranitidine was ineffective in prevention of gastric ulcers in NSAID users but was effective in prevention of duodenal ulcer in NSAID users [24, 25]. These data suggest that there may be a fundamental difference between gastric ulcers and duodenal ulcers that develop in NSAID users. Both misoprostol and ranitidine are antisecretory drugs and are effective in reducing duodenal ulcer recurrence in patients with typical idiopathic (Helicobacter pylori) duodenal ulcers. The success of both misoprostol and ranitidine in the prevention of duodenal ulcers in NSAID users suggests that a high proportion of duodenal ulcers in NSAID users are acid dependent and that some other mechanism must be active in the development of gastric ulcers in chronic NSAID users. One possibility would be that NSAID use exacerbates Helicobacter pylori duodenal ulcers. If this were the case, it might explain the higher-than-expected frequency of complications related to duodenal ulcers that have been reported in NSAID users.

Misoprostol is the only drug shown to be capable of preventing gastric and duodenal ulcers in chronic NSAID users. The value of sucralfate in prevention of duodenal ulcers in chronic NSAID users is unknown but, in a direct comparative study, misoprostol was significantly more effective than sucralfate in preventing gastric ulcers in patients receiving chronic NSAID therapy [23]. There are, however, few data concerning omeprazole for this indication. Omeprazole was shown to protect the duodenal mucosa of 15 healthy participants from short-term, NSAID-induced damage, but significant protection of the gastric mucosa was not demonstrated [26]. Long-term clinical trials comparing a larger number of patients are required before the efficacy of omeprazole in preventing NSAID-induced ulcers can be adequately assessed.

Because symptoms in NSAID users cannot be used to guide therapeutic decisions, the question arises as to which patients are at risk for NSAID-induced gastrointestinal damage. Although cotherapy with misoprostol will markedly reduce the frequency of both gastric and duodenal ulcers in long-term NSAID users, we cannot yet reliably identify the group of patients who are at highest risk for development of complications of ulcer disease. The present study was not designed to examine subgroups, yet when patients with and without a history of ulcer were analyzed separately, misoprostol was significantly more effective than placebo in preventing NSAID-induced gastroduodenal ulcers in both populations. Several important risk factors predisposing patients to NSAID-induced gastrointestinal damage have been identified, including patient age, prednisone use, a previous NSAID-associated gastrointestinal side effect, and a severe underlying illness or disability [27-34]. Furthermore, a linear response has been found between the NSAID dose and gastrointestinal complications [8]. We may be able to add the presence of gastric erosions and alcohol use as risk factors for ulcer development. In high-risk patients who require NSAID therapy, prophylaxis with misoprostol may be a prudent clinical decision.

The Misoprostol Study Group included Naurang Agrawal, MD, New Orleans, LA; Jacques Caldwell, MD, Daytona Beach, FL; David Cave, MD, and Richard Simms, MD, Boston, MA; Don E. Cheatum, MD, and Richard Stone, MD, Dallas, TX; Martin Collen, MD, and Stanley Benjamin, MD, Washington, DC; Richard Erickson, MD, and Andrezej Tarnawski, MD, Long Beach, CA; Raeburn Evans, MD, Indianapolis, IN; Norman Gitlin, MD, Fresno, CA; John Goff, MD, Denver, CO; Esther Gonzales, MD, Eduard Santiago Delpin, MD, and Esther A. Torres, MD, San Juan, PR; Martin Lidsky, MD, Houston, TX; Oren Gum, MD, New Orleans, LA; Robert Knodell, MD, Baltimore, MD; Mark Korsten, MD, Bronx, NY; Timothy Lipman, MD, and David Nashell, MD, Washington, DC; Jeffrey Lisse, MD, Galveston, TX; Barbara Lister, MD, Houston, TX; William Makarowski, MD, and Paul Gmuer, MD, Erie, PA; Charles Elson, MD, Birmingham, AL; John L. Orchard, MD, Pittsburgh, PA; Keith Peterson, MD, and Clyde Bench, MD, Salt Lake City, UT; Walter Peterson, MD, and Eric Hurd, MD, Dallas, TX; Ron Pruitt, MD, and Daniel Gremillion, MD, Nashville, TN; Jahangir Rao, MD, Wayne, MI; Sanford Roth, MD, Phoenix, AZ; Marshall Sack, MD, Austin, TX; Robert Sanowski, MD, and J. Patrick Waring, MD, Phoenix, AZ; Frank Schmid, MD, Chicago, IL; Elliott Semble, MD, and Wallace Wu, MD, Winston-Salem, NC; David Stanton, MD, Mission Viejo, CA; Charles Colip, MD, Portland, OR; Ernest Katler, MD, Martinex, CA; Jorge Valenzuela, MD, and David Arandia, MD, Los Angeles, CA; Steven J. Wees, MD, and William Palmer, MD, Omaha, NE; Walter Trudeau, MD, Sacramento, CA; Colon H. Wilson, MD, and Eduard Alberto Baetti, MD, Atlanta, GA; Marc Zuckerman, MD, El Paso, TX.