One obvious way to eliminate these serious ulcer complications is to avoid the use of NSAIDs whenever possible. In many people with mild to moderate pain or with fever, acetaminophen can be used for its analgesic or its antipyretic actions. In some patients with osteoarthritis or rheumatoid arthritis, a nonacetylated salicylate can be used as an anti-inflammatory agent. Salsalate, for example, neither inhibits prostaglandin synthesis in the gastroduodenal mucosa nor causes acute gastroduodenal mucosal injury [4]. However, both aspirin and nonsalicylate NSAIDs remain popular agents, a testimonial to their cost-effectiveness and relative safety [5]. Aspirin, for example, is effective in the prevention of myocardial infarction, occlusion of saphenous vein grafts, and strokes and in the treatment of acute myocardial infarction, unstable angina, and transient cerebral ischemic attacks [6]. Furthermore, recent epidemiologic evidence suggests that regular aspirin use may be associated with lower mortality from certain cancers, including colonic cancer [7]. Nonsalicylate NSAIDs may be useful not only in patients with osteoarthritis and rheumatoid arthritis but also in those with juvenile rheumatoid arthritis, ankylosing spondylitis, acute gout, and primary dysmenorrhea. It is unrealistic to believe, therefore, that regular use of aspirin and nonsalicylate NSAIDs will decline in the United States. Just the opposite is happening. Moreover, aspirin and ibuprofen are available over-the-counter, permitting long-term use without medical supervision.

In 1988, a group of American investigators published the first of three carefully performed and well-designed clinical trials on the protective effect of misoprostol (Cytotec; G.D. Searle & Company; Chicago, Illinois), a prostaglandin E₁ analog, on gastroduodenal ulcers in a group of arthritis patients with abdominal pain who had been regularly ingesting ibuprofen, piroxicam, or naproxen [8]. Rather than using clinically diagnosed ulcer disease as an end point, the investigators used a more frequently occurring end point: a new ulcer visible at routine surveillance endoscopy. This study demonstrated striking protection by misoprostol against gastric ulcers. A new gastric ulcer measuring 0.5 cm in diameter developed in the next 3 months in 12.3% of the patients receiving placebo with their NSAID, in 4.2% of patients receiving misoprostol, 100 µg four times a day, with their NSAID, and in 0.7% of those given misoprostol, 200 µg four times a day, with their NSAID. In a second 3-month trial of similar design, these same investigators found that misoprostol, 200 µg four times a day, was superior to sucralfate, 1 g four times a day, in preventing new gastric ulcers in arthritis patients receiving ibuprofen, piroxicam, or naproxen. Gastric ulcers occurred in 9.2% of patients taking sucralfate with their NSAID and in only 0.8% of those taking misoprostol, 200 µg four times a day, with their NSAID [9].

In the original misoprostol study [8], misoprostol did not significantly protect against duodenal ulcers, possibly because development of a new duodenal ulcer during the 3-month trial was relatively uncommon. A new duodenal ulcer occurred in 3.6% of those taking placebo with the NSAID, 2.1% of those taking misoprostol, 100 µg four times a day, with the NSAID, and in 2.9% of those taking misoprostol, 200 µg four times a day, with the NSAID. Around the same time, two placebo-controlled studies investigating the protective effect of ranitidine in chronic NSAID users reported an incidence of new duodenal ulcers that was considerably higher than that reported for placebo in the misoprostol study, namely 8% after just 2 months of an NSAID [10, 11]. Moreover, virtually all of the duodenal ulcers could be prevented by administering ranitidine, 150 mg twice daily, together with the NSAID [10, 11]. In contrast to ranitidine's protective effect against duodenal ulcers in NSAID users, ranitidine was almost completely ineffective in preventing gastric ulcers. The implications from these studies [8, 10, 11] were troublesome: If the objective is to prevent both gastric ulcers and duodenal ulcers in chronic NSAID users, both misoprostol (to prevent gastric ulcers) and ranitidine (to prevent duodenal ulcers) may be required.

Because the H₂-receptor antagonist ranitidine almost certainly prevented NSAID-induced duodenal ulcers through its inhibitory effect on gastric acid secretion in the two trials just cited [10, 11] and because misoprostol in a dose of 200 µg four times a day also reduces acid secretion [12], it was surprising that misoprostol did not significantly prevent duodenal ulcers in the placebo-controlled study by Graham and colleagues. A ß (type II) error was certainly possible in this latter study [8]. In the most recent study from the misoprostol study group, published in this issue of Annals [13], nearly three times as many patients were assigned to receive 200 µg of misoprostol four times a day or placebo for 3 months as in the earlier trial, increasing the power of the study considerably. Unlike the two earlier studies [8, 9], arthritis patients receiving various NSAIDs (ibuprofen, piroxicam, naproxen, diclofenac, indomethacin, or sulindac) were not required to have abdominal pain on entry into the study. Misoprostol, 200 µg four times a day, again prevented NSAID-induced gastric ulcers (placebo, 7.7%; misoprostol, 1.9%), although the protection was not nearly as pronounced as in the earlier trial [8]. Unlike the earlier study, misoprostol significantly reduced the incidence of duodenal ulcers (4.6% in those taking placebo with the NSAID and 0.6% in those taking misoprostol, 200 µg four times a day, with the NSAID).

When all three of the above misoprostol studies from the same group of investigators are combined, the 3-month incidence of new duodenal ulcers in regular NSAID users is 4.3% with placebo and 1.4% with misoprostol, indicating that approximately two thirds of duodenal ulcers can be prevented with misoprostol in a dose of 200 µg four times a day. The 3-month incidence of NSAID-related duodenal ulcers is 0.8% with sucralfate [9] and the 2-month incidence of duodenal ulcers is approximately 1.0% with ranitidine [10, 11]. Thus, misoprostol appears to be superior to both ranitidine and sucralfate for preventing endoscopically diagnosed gastric ulcers in nonsalicylate NSAID users and as effective as ranitidine and sucralfate in preventing endoscopically diagnosed duodenal ulcers. This conclusion is supported by another large 8-week, multicenter study comparing misoprostol, 200 µg four times a day, to ranitidine, 150 mg twice a day, in patients with abdominal pain while taking NSAIDs [14]. It is uncertain, however, whether misoprostol (or any medication) will prevent jejunal/ileal ulceration in chronic NSAID users [15-17].

Will misoprostol prevent gastric ulcers, duodenal ulcers, or small-bowel ulcers during long-term ingestion of aspirin? Unfortunately, none of the three published misoprostol trials included patients taking aspirin [8, 9, 13]. In another study, misoprostol, 200 µg four times a day, was not significantly superior to placebo in preventing new gastric ulcers and duodenal ulcers in rheumatoid arthritis patients receiving aspirin [18]. Despite misoprostol cotherapy, 16 of 86 rheumatoid arthritis patients (19%) developed new gastroduodenal ulcers while taking aspirin (650 to 1300 mg four times a day) during a 2-month period of endoscopic surveillance.

In each multicenter misoprostol study referred to above [8, 9, 13, 14, 18], misoprostol-treated patients had more diarrhea and abdominal pain or cramps than did their respective controls. Thus, misoprostol may prevent ulcers detected endoscopically in long-term nonaspirin NSAID users, but misoprostol paradoxically leads to more gastrointestinal symptoms than does placebo. Lowering the dose of misoprostol to 100 µg four times a day or to 200 µg twice a day causes less diarrhea, but is also significantly less effective in preventing ulcers [8, 19].

Although it appears that misoprostol may be effective in preventing both gastric ulcers and duodenal ulcers in regular users of nonaspirin NSAIDs, we should recall that the primary goal should be the prevention of symptomatic ulcer disease rather than ulcers that only can be detected with the aid of an endoscope and that are often asymptomatic. Of 775 patients receiving placebo along with an NSAID in several trials already discussed [8, 10, 11, 13, 18], only one was reported to have had a serious complication during a 2-to 3-month period of observation, a patient with rheumatoid arthritis who had gastrointestinal bleeding while taking aspirin [18]. Unless it can be shown in a clinical trial containing a sufficiently large number of patients (probably thousands) that misoprostol can prevent serious ulcer complications or symptomatic ulcers, use of misoprostol will rest on the "logical" but unproven assumption (or hope) that prevention of ulcers detected endoscopically will translate to prevention of clinically diagnosed ulcers. Such a large, multicenter clinical trial is now underway and the results are eagerly awaited.

For the time being, misoprostol should continue to be used sparingly in regular NSAID users. To date, the only "high risk" group of NSAID users that has been identified is those with previous ulcer disease, in whom the risk of NSAID-related ulcer is increased 1.8 to 5.5 times [11, 13]. Such individuals ideally should not receive NSAIDs and, indeed, have often been excluded from earlier misoprostol-NSAID trials [8, 9]. Interestingly, older age per se was not an independent risk factor for NSAID-related gastric ulcers or duodenal ulcers in the most recent and largest misoprostol study [13]. We hope we will soon learn which types of patients are at greatest risk for development of clinically significant gastroduodenal and intestinal ulcers while taking aspirin and other NSAIDs. When we do, potentially effective prophylactic therapies, including misoprostol cotherapy, can be carefully evaluated in these individuals.