Ulcers, Nonsteroidal Antiinflammatory Drugs, and Related Matters ANDREW H. SOLL, JOHN KURA T A, and JAMES E. McGUIGAN Center for Ulcer Research and Education. and the Medical and Research Services. Veterans Administration Wadsworth Hospital Center. the School of Medicine. University of California at Los Angeles. Los Angeles. California. and Department of Medicine. University of Florida. College of Medicine. Gainesville. Florida
This overview is intended to provide perspectives drawn from the proceedings of the Ulcer-Epidemiology Symposium reported in this supplement. This meeting was designed to review the epidemiology and risk factors underlying the development of acid/peptic diseases and to highlight the natural history and therapy of these disorders. These topics served as a focus for the meeting. A second goal was to consider the topics in the context of ulcer complications resulting from use of nonsteroidal antiinflammatory drugs. Although nonsteroidal antiinflammatory drugs are well known to cause gastric damage and ulcers and to lead to complications of preexisting peptic ulcers, the frequency and severity of the clinical problems resulting from their use and the efficacy of different modalities for prevention and treatment have only recently begun to be the subjects of careful analysis. rostaglandins (PGs) enhance gastrointestinal muP cosal resistance to injury; Andre Robert demonstrated that PGs dramatically protected the gastric (1)
mucosa of rats against macroscopic damage by a variety of injurious agents, such as exposure to boiling water or 100% ethanol. These protective effects of PGs appeared to be independent of their ability to inhibit acid secretion. With the description of these mucosal protective properties of PGs, a lot of interest was generated in studies on the effects of PGs on healing of peptic ulcers. Prostaglandins were found as effective, but no more effective, in healing ordinary duodenal ulcer than could be predicted from their antisecretory effects (2,3). The importance of PGs in mucosal resistance to peptic ulcer and other mucosal injury thus came into question. This overview concerns the effects ofPGs on mucosal defense in relation to gastric and duodenal injury induced by nonsteroidal antiinflammatory drugs (NSAIDs). -It is not intended to be comprehensive,
but rather aims at highlighting several unresolved issues and controversies raised during this symposium. In our view further attention to these points may allow a more complete understanding of this topic and its clinical reievance.
Prostaglandins and Mucosal Defense What Is Cytoprotection? Jacobson coined the term cytoprotection to describe the ability of PGs to enhance mucosal resistance to injury independent of acid secretory inhibition. Subsequently, he stated that "anyone who thinks he understands cytoprotection has not read the literature" (Jacobson E, personal communication). The term cytoprotection fell into disrepute after it was reported that the term was technically a misnomer; when histologic sections were examined, PGs did not protect surface epithelial cells from ethanol injury (4). Furthermore, the term cytoprotection implies that PG action reflects direct enhancement of cell resistance, whereas the potential mechanisms by which PGs enhance mucosal defense are probably multiple, including stimulation of bicarbonate and mucus output, and maintenance of mucosal blood flow (5,6). Although PGs may directly enhance cellular resistance to injury (7), this mechanism is probably at most a minor component of the actions of PGs. Rather than cytoprotection, "mucosal protection" is a more accurate term for the effects of PGs on mucosal defense independent of acid secretory inhibition.
Abbreviations used in this paper: NSAID, nonsteroidal antiinflammatory drug; RA. rheumatoid arthritis. © 1989 by the American Gastroenterological Association 0016-5085/891$3.50
SOLL ET AL.
Is "Mucosal Protection" by Prostaglandins an Applicable Concept in Humans? In the rat, gastric mucosal protective effects of PGs are evident at doses less than those necessary to inhibit gastric acid secretion (1), making it relatively easy to separate effects on mucosal defense from inhibition of acid secretion. In humans, PGs induce effects that are likely to enhance mucosal defense by mechanisms such as stimulation of bicarbonate output (8). However, the dose response for these effects have not been established. Jiranek et al. (see page 656) report a dose ranging study for prevention of acute aspirin injury to the gastric mucosa by the PGE 1 analogue ri:tisoprostol. However, present data remain insufficient to distinguish the dose ranges for acid inhibitory and mucosal protective effects of PGs. Another way to establish the existence of mucosal protective actions of PGs that are not due to acid secretory inhibition is to compare PGs with other agents of equal antisecretory efficacy. A recent study by Lanza et al. (9) reports that a PG analogue may be more effective in preventing acute NSAID-induced gastric injury than an H 2 -receptor antagonist of equal antisecretory potency. This finding that PGs are more effective than strictly antisecretory agents in preventing NSAID-induced gastric injury provides support for the concept that PGs enhance mucosal resistance to NSAID-induced gastric injury by mechanisms in addition to inhibition of acid secretion. Is Mucosal Protection by Prostaglandins Relevant to the Healing of Ordinary Peptic Ulcers? This question prompts brief consideration of the healing of ordinary peptic ulcers. Samloff (see page 586) reminds us of the statement from Schwarz in 1910 that "peptic ulcer ... results from an excess of autopeptic power in gastric juice over the defensive power of gastric and intestinal mucosa" (10). Thl1s, it was Schwarz who first recognized that ulcers result from acid/peptic activity only when the ability of the mucosa to defend itself against injury is overpowered. Breaks in mucosal integrity occur continuously, secondary to food and other trauma. Moreover, we do not hesitate to violate the mucosa with biopsies. The normal outcome is rapid healing. Several elements are probably necessary to create a microenvironment that will allow healing to proceed reliably; with minor trauma to the epithelium and little damage below the basement membrane, the central elements of repair are restitution (neighboring cells moving in to fill epithelial defects) and
GASTROENTEROLOGY Vol. 96. No.2. Part 2
mucosal regeneration. It is likely that mucus and bicarbonate secretion and maintenance of mucosal blood flow are important to create a microenvironment that facilitates healing. With deeper injury, an inflammatory response occurs and the wound healing process is probably comparable to that described in skin and other tissues. One can speculate that breaks in the mucosa only become clinically relevant ulcers when the resealinglhealing process fails. Thus, sufficient acid/peptic activity and impaired mucosal defense may both be critical permissive factors in ulcer formation. As an ulcer forms, it is accompanied by a local inflammatory response. Gastric ulcers not associated with NSAIDs develop in mucosa that displays gastritis. Duodenal ulcer is associated almost invariably with antritis, and, to a lesser extent, duodenitis. Although there is insufficient information to sort out the causal aspects of the relationship between ulcers and inflammation, two generalizations are reasonable: (a) gastroduodenitis is an associated factor or a potential predisposing factor in peptic ulcer formation and (b) the formation of ulcers is accompanied by an intensified inflammatory response in the vicinity of the ulcer. An established ulcer lacks the normal local mucosal elements responsible for protection from acid/peptic injury; it is without the cells that secrete bicarbonate and mucus, blood flow may be peiturbed, and the tissue architecture required for restitution and regeneration is altered. Many .of the actions of PGs require intact target cells and tissues (e.g., mucous cells for mucus and bicarbonate production, or capillary endothelium), and ulcers lack these elements. This may explain why PGs enhance healing of ordinary peptic ulcers as effectively-but no more effectively-than antisecretory agents.
Nonsteroidal Antiinflammatory Drugs, Ulcers, and Prostaglandins How Do Nonsteroidal Antiinflammatory Drugs Cause Ulcers?
There is little disagreement that NSAIDs cause gastrointestinal ulcers, but the responsible mechanism or mechanisms remain controversial (see Kauffman, page 606). Aspirin, and other NSAIDs to a lesser extent, are local irritants that produce gastric mucosal damage. Although there are considerable differences among NSAIDs in their propensity to induce acute, superficial damage, the acute injury profile is not a good predictor of clinically significant ulcer formation. Agents that produce little superficial damage still cause ulcers. Acute, superficial damage appears not to be a necessary precursor of
ulcer formation, although variables, such as impaired healing, might allow acute damage to evolve into chronic ulcers. In addition to direct local injury, NSAIDs produce effects on the gastric mucosa when administered systemically. Several lines of evidence indicate that induction of ulcers by NSAIDs reflects this systemic action, as reviewed by Graham (see page 675). Ulcer formation occurs in animal models with systemic administration of NSAIDs (11). Systemic effects are also important in humans, a point particularly highlighted by several observations, including the following: (a) sulindac, a prodrug that has antiinflammatory activity when biotransformed, fails to cause acute gastric damage, but its use is associated with gastrointestinal bleeding at a rate comparable to other NSAIDs (12), and (b) delayed-release and enteric-coated NSAIDs and aspirin produce less acute, superficial gastric damage, but still cause ulcers and ulcer complications (see Graham, page 675). Especially in light of the multiple pharmacologic effects of NSAIDs, the mechanisms underlying ulcer formation have been difficult to sort out. The most consistent and well studied systemic effect of NSAIDs is the dramatic inhibition of PG generation by inhibition of cyclooxygenase. However, NSAIDs also may enhance production of several inflammatory mediators, including lip oxygenase products, platelet activating factor, and oxygen free, radicals, presumably by shifting precursor fatty acids to other metabolic pathways (13). Direct measurement of the rates of PG production in the gastric mucosa is difficult, thereby hampering studies of the role of PGs in ulcer formation in humans. The impact of cotherapy with PGs on evolution of NSAID-induced gastric injury can provide strong supporting evidence for the view that NSAID-induced damage results from selective inhibition of PG formation. As noted elsewhere in this supplement (see Graham, page 675, and Jiranek, page 656), PGs appear to prevent acute NSAID-induced gastric lesions by effects that are independent of acid secretory inhibition. Nonsteroidal antiinflammatory drug-induced gastric ulcers probably result from inhibition of PG production, but further studies are needed to confirm this hypothesis. What Role Do Endogenous Prostaglandins Play in Mucosal Defense? The importance of endogenous PGs to mucosal defense has been debated (2,5,6). This contro~ versy has been fostered by finding that although aspirin and several other NSAIDs consistently induce acute damage to the gastric mucosa, many
ULCERS, NSAIDs AND RELATED MATTERS
individuals ingest NSAIDs without recognized clinical consequences. The gastric mucosa also can adapt to exposure to low doses of aspirin with healing of acute, superficial injury (see Graham, page 675). Is this because endogenous PGs are not crucial to mucosal defense or because NSAIDs fail to inhibit completely PG production? Redfern and Feldman (see page 596) review evidence that antibodies to PGs consistently induce gastroduodenal ulcers in animals. Although the mechanisms operative in this model require further elucidation, a likely hypothesis is that these antibodies induce ulcers by binding and inactivating endogenous PGs. Obviating the effects of endogenous PGs probably impairs mucosal resistance to injury and/or impeqes mucosal repair. The consequences may depend on the degree and duration of inactivation of PGs, the underlying condition of mucosal defenses, and the activities of aggressive factors, such as acid, pepsin, and possibly bile. How Are Nonsteroidal Antiinflammatory Drug-Induced Ulcers Related to Common Peptic Ulcers?
Defining relationships between NSAID-induced ulcers (occurring in individuals consuming NSAIDs) and peptic ulcers (occurring in individuals not taking NSAIDs) rests on knowledge of the pathogenic mechanisms operative in both situations. As ordinary peptic ulcers are common, some NSAIDassociated ulcers will occur on a background of a peptic ulcer predisposition or diathesis. However, NSAID-induced ulcers often occur in individuals who have never manifested any peptic ulcer diathesis. Previous epidemiologic studies have indicated that NSAID use is nlUch more likely to be associated with gastric than duodenal ulcer. However, NSAID consumption is a risk factor for bleeding and perforation of duodenal as well as gastric ulcers (see Langman, page 640, and McCarthy, page 662). The extent to which NSAID-associated ulcer complications are due to ulcers caused entirely by NSAIDs or to exacerbation by NSAIDs of preexisting, often asymptomatic, peptic ulcers is not known. Certain characteristic features may permit distinction of NSAID-induced ulcers from ordinary peptic ulcers. Peptic ulcers in the stomach are usually associated with low or normal acid secretion and a low serum pepsinogen I/pepsinogen II ratio, reflecting an underlying gastritis (see Samloff, page 586). However, NSAID-induced ulcers may occur without an obvious underlying peptic ulcer diathesis, and some NSAID-associated ulcers are not associated with gastritis. Although the causal nature of the
SOLL ET AL.
relationship remains controversial, peptic ulcers of the duodenum, and to a lesser extent of the stomach, show a strong association with gastric colonization with Campylobacter pylori. The possibility that gastric infection with Campylobacter pylori is a risk factor for the development of symptoms, ulcers, or ulcer complications in subjects taking NSAIDs is currently being investigated.
Are Certain Individuals Predisposed to Developing Ulcers With Nonsteroidal Antiinflammatory Drugs? Not everyone treated with NSAIDs develops ulcers and even fewer develop complications, raising the possibility that only a subpopulation of individuals with enhanced susceptibility develop ulcers with NSAID therapy. Genetic predisposition for NSAID ulceration has been proposed. Semble et a1. (14) performed endoscopy on 41 patients with rheumatoid arthritis (RA) being treated with various NSAIDs; they found blood type 0 to be represented in greater frequency in patients with mucosal lesions than in patients with normal endoscopies. It is possible that genetic factors alter gastric mucus, as variations in blood group antigens are expressed in gastric mucus. This hypothesis, however, nemains totally speculative. Nonsteroidal antiinflammatory drugs have been reported to enhance gastric acid secretion (15), suggesting an increased dependency on the acid suppressive effects of endogenous PGs (so that inhibition of PG synthesis increases acid secretion). However, as an increase in gastric acid secretion has not been consistently found in subjects receiving NSAIDs, the effects of NSAIDs on acid secretion remain controversiaL The variability of the observed effects of NSAIDs on acid secretion may reflect the existence of an important feedback, inhibitory role of endogenous PGs on parietal cell secretion in some individuals that is absent in others. However, an increase in gastric acid secretion with NSAID treatment has not been found in all subjects nor has a reproducible effect been verified in a subset of subjects. At the present time one can only hypothesize that individuals may vary in their dependency on endogenous PGs, their acid secretory response to NSAIDs and in their susceptibility to NSAID-induced mucosal injury. If such variability exists, pathophysiologic or pharmacologic abnormalities predisposing to ulcer formation with NSAIDs would be anticipated only in the susceptible individuals, thus complicating study of these mechanisms.
GASTROENTEROLOGY Vol. 96, No.2 , Part 2
Nonsteroidal Antiinflammatory Drugs and Ulcer Complications How Important Are Nonsteroidal Antiinflammatory Drug-Induced Ulcer Complications?
Although there is little question that NSAIDs damage the gastric mucosa and cause ulcers, there has been uncertainty about the magnitude of the risk of serious gastrointestinal complications, such as ulcer hemorrhage and perforation, associated with NSAID use. Langman (see page 640) summarized the four approaches that have been utilized to address this question. (a) Side effects of NSAIDs are most reported to national agencies, but these data are difficult to interpret (especially in the United States) because of highly variable reporting of untoward drug reactions by physicians. (b) Clinical trials of NSAID therapy support the thesis that NSAID consumption is associated with increased rates of ulcer complications, although the low incidence of ulcer complications requires that trials be combined, thus encountering the vagaries of the meta-analysis approach (16). (c) In the United Kingdom several retrospective analyses of ulcer complications assessed at the time of hospitalization indicate a strong association between ulcer complications and NSAID consumption. Langman estimates from these studies that the serious complication rate is about 1 per 5500 prescriptions of an average duration of 1 mo (see Langman, page 640). Studies of ulcer complications in the United States yield conflicting results. However, these studies may not evaluate sufficient numbers of subjects, especially in elderly age groups who have the greatest apparent risk, to clearly detect an association between NSAID consumption and ulcer complications. (d) Fries and coworkers have used a different approach, in that they have prospectively followed a large number of patients with RA. Preliminary data (see Fries, page 647) indicate that NSAID use is a clear risk factor for gastrointestinal hospitalizations in the RA population followed by the American Rheumatism Association Medical Information System. Furthermore, deaths due to gastrointestinal causes are more common in the RA population than expected from statistics for the general population. Although it is not possible to be certain how many of these deaths are due to ulcer complications associated with NSAID consumption, a strong argument can be made that many of these deaths are NSAID-associated ulcer complications. Although the exact risk of gastrointestinal complications imparted by NSAID use remains uncertain, several thoughtful investigators have concluded that NSAIDs impart a significant risk for such ulcer complications (see Kurata, page 569; Langman, page
640; Fries, page 647; McCarthy, page 662; and Graham, page 675). Considering the very large number of NSAID prescriptions issued per year in the United States, the number of patients dying each year from NSAID-associated ulcer complications probably measures in the thousands, with significant morbidity at least ten times that high. Additional studies are needed to determine the magnitude of the risk. The problem is a serious one with substantial, but as yet incompletely defined, costs to society. Studies are currently in progress to estimate the extent of these costs to patients in the United States on chronic NSAID therapy.
Which Risk Factors Predispose to Ulcer and/or Complications? The benefit of being able to identify patients who are at increased risk of developing ulcers and ulcer complications while taking NSAIDs is obvious; individuals at increased risk would clearly warrant preventative ulcer therapy. Retrospective epidemiologic studies have indicated that NSAID complications are most frequent in elderly women. This frequency is explained in part by this group being exposed to greatest risk because they consume the most NSAIDs. There may, however, be an increase in the frequency of ulcer hemorrhage and perforation per NSAID prescription issued in this elderly population. Other likely risk factors include a prior history of peptic ulcer disease and smoking. Concomitant serious disease may precipitate ulcers and their complications, and may compromise a patient's capacity to survive these complications. The effects of concomitant medication, such as corticosteroids, immunosuppressive agents, or consumption of multiple NSAIDs, remain uncertain. It is likely that dose, duration of NSAID action, and duration of NSAID therapy are also risk factors for ulcers. Dose adjustments for body weight are important and the failure to make such adjustments may underlie some of the complications occurring in elderly women. Ironically, compliant patients, who take their NSAIDs regularly as prescribed, may be those most likely to develop the gastrointestinal complications of NSAID use. The complexity of assessing the multiple risk factors for ulcer complications is underlined in the population with RA. Deaths due to gastrointestinal causes occur in RA at a rate greater than expected for the population at large (see Fries, page 647); with the data available at present, many of these deaths appear secondary to the complications of ulcers in patients using NSAIDs. Further studies are needed to establish the extent to which these deaths are due to gastric and duodenal ulcers. Gastrointestinal-related
ULCERS, NSAIDs AND RELATED MATTERS
deaths increase, as expected, with the interrelated variables of age, the severity and duration of RA, and the number of associated diseases. Furthermore, several classes of medications with ulcerogenic potential, including NSAIDs, corticosteroids, gold salts, and antimetabolites, are administered according to disease severity. In light of the interdependence of these variables, it is difficult to sort out the cause and effect relationships between NSAIDs and other risk factors and ulcer disease. Thus, identification of the risk factors for NSAIDassociated ulcers is not an easy task; a suitable data base for assessing risk factors would be likely to come from prospective studies of subjects who at the outset of the study do not have peptic ulcer disease. Information on NSAID consumption and other variables such as age, sex, associated diseases and their duration, and medications prescribed and dosage levels, and the development of ulcers and complications would then be collected in the follow-up period of the study. Since these multiple risk factors are likely to interact, multivariate analysis capable of weighting the interdependency among variables will be required to analyze such study data. How Frequently Do Ulcers Present With Complications?
Nonsteroidal antiinflammatory drug-induced ulcers, probably more commonly than ordinary peptic ulcers, present with complications, rather than with typical ulcer symptoms (see Pounder, page 626). In the study by Armstrong and Blower (17), many of the patients who died or required surgery for ulcer complications never knew that they had ulcer disease. Although they may not realize it, patients with symptoms may be fortunate in that the occurrence of symptoms, such as epigastric pain, usually prompts the patient or physician to stop the offending medication, investigate the complaint, and/or institute antiulcer therapy. In contrast, the clinically silent ulcer remains untreated and therefore at continuing risk for complications. Nonsteroidal antiinflammatory drug-associated ulcers account for a very significant proportion of the silent ulcers that present with life-threatening complications (17). Nonsteroidal antiinflammatory drugassociated ulcers may have a greater risk of complications without heralding symptoms than ordinary peptic ulcers; it has been proposed that the analgesic properties of NSAIDs diminish patient perception of ulcer pain. Although many aspects of silent ulcers remain unknown, it is clear that one cannot rely on symptoms to reveal the presence of an ulcer, particularly when the patient is consuming NSAIDs. What is needed is an algorithm for the early detection,
SOLL ET AL.
treatment and/or prevention of ulcers or ulcer complications in patients using NSAIDs. Do Any Ulcer Characteristics, Such as Depth, Size, and Lesion Duration, Predict the Likelihood of Complications? Depth. Superficial lesions (erythema, red spots and minimal erosions) may bleed occasionally and cause patients to undergo upper gastrointestinal x-ray examination or endoscopy, but otherwise usually carry little clinical significance. Lesions extending through the lamina propria are the ones at significant risk for potentially life-threatening perforation and bleeding. Size. Although inadequately documented, depth is likely to increase in proportion to size of endoscopic lesions. However, narrow, linear lesions may be quite shallow, so that the dimension that probably provides the greatest predictive value for detecting the presence of an ulcer may be diameter rather than length. Therefore, as McCarthy emphasizes (see page 662), a more appropriate dimension for measurement in endoscopic trials using lesion size as the criterion for "ulcers" may be the diameter, rather than length. Furthermore, the greater the size cutoffs for ulcers (e.g., 5-6 mm instead of 3-4 mm), the greater the probability of correctly categorizing lesions as acute ulcers. The variability in ulcer frequency reported in NSAID trials may relate to differences in the diagnostic criteria applied by different groups of investigators. Ulcer duration and degree of chronic inflammatory change. In the absence of longitudinal studies, one can speculate that acute ulcers, present for hours or days, are more likely to be shallow with only acute inflammation, whereas with increasing duration, ulcers might carry an increased risk of being deep and surrounded by a chronic inflammatory infiltrate and scarring. Chronicity of the lesion may not correlate with duration of NSAID therapy; lesions in subjects on sustained NSAID dosage may appear and resolve (see Graham, page 675), as do ordinary peptic ulcers. Unfortunately, no studies have carefully examined the impact of lesion chronicity, size, and depth on outcome, but these factors are probably important determinants of the response to ulcer healing therapy and the risk of complications. Deep lesions may be more likely than superficial lesions to produce significant complications. Large lesions, with chronic inflammatory change, will probably heal more slowly and with more difficulty than small lesions. As such variables may influence healing rates, clinical trials that categorize lesions by these criteria should generate more comprehensive
GASTROENTEROLOGY Vol. 96. No.2. Part 2
data regarding the efficacies of different therapies. Despite the possibility that deep, chronic lesions are those that predispose to complication, patients may develop complications within a few days of starting NSAID therapy. Furthermore, superficial mucosal lesions may result in anemia from chronic blood loss. How Effective Are Various Therapies in Healing Nonsteroidal Antiinflammatory Drug-Induced Ulceration?
As discussed by McCarthy (see page 662) and Graham (see page 675), a variety of therapies have been utilized for healing NSAID-induced ulcers. If NSAIDs are discontinued, administration of most of these treatment modalities is associated with ulcer healing. If NSAIDs are continued, acceleration of healing has been reported with simultaneous treatment with H2 -receptor antagonists. Data suggest that larger NSAID-associated gastric ulcers, however, heal slowly with conventional antisecretory therapy (see McCarthy, page 662). These data require further confirmation. PGs have also been shown to be effective in healing NSAID-associated ulcers (18), but experience is still limited with healing of gastric and duodenal ulcers by PGs in the setting of continued NSAID therapy. Currently, there are no direct comparisons of the relative efficacies of PGs vs. other therapeutic agents in healing ulcers in patients who continue NSAIDs, although such studies are in progress. As mentioned earlier, established ulcers, whether NSAID-related or ordinary peptic ulcers, are surrounded by an inflammatory response, and, therefore, local defense mechanisms may be perturbed as a consequence of disruption of the mucosal architecture. Prostaglandins, even with their dual effects inhibiting acid secretion and enhancing mucosal defense mechanisms, may be as effective, but no more effective, than strictly antisecretory agents in healing NSAID-induced ulcers. Additional prospective, controlled clinical trials are necessary to evaluate the efficacy of PGs for healing NSAID-induced ulcers in comparison to placebo and to strictly antisecretory agents. Prevention of Nonsteroidal Antiinflammatory Drug-Associated Ulcers Are mechanisms underlying ulcer healing and ulcer prevention the same? Although speculative, it is possible that prevention of NSAID-induced ulcers results from mechanisms distinct from those involved in healing established peptic or NSAIDassociated ulcers. The ulcer healing process occurs in a region of mucosa disrupted by the ulcer crater
and surrounding inflammation. In contrast, the goal of prophylactic therapy is to prevent ulcer formation and preserve mucosal architecture. The cellular and tissue substrates for preventative therapy differ from those for enhancement of ulcer healing. Although the importance of the non-antisecretory effects of PGs in healing established ulcers is uncertain, PGs do enhance defense mechanisms in normal mucosa. Based on this line of reasoning, exogenous PGs may prevent NSAID-induced ulceration with efficacy superior to antisecretory agents by compensating for depletion of endogenous prostaglandins induced by NSAIDs, thereby restoring mucosal defense mechanisms. Prevention of nonsteroidal antiinflammatory drug-induced gastric ulcer. Data presently available do not indicate that H 2 -receptor antagonists prevent NSAID-induced gastric ulcers (see McCarthy, page 662). However, a recent study demonstrated that PGs effectively prevented NSAID-induced gastric ulcers occurring during a 3-mo study in patients with osteoarthritis taking NSAIDs (19; see Graham, page 675). Thus, in contrast to ordinary peptic ulcer where healing is enhanced by numerous modalities, available data suggest that prostaglandins, but possibly not strictly antisecretory agents, prevent NSAID-associated gastric ulcer. The hypothesis that NSAID-associated gastric ulcer reflects inhibition of mucosal PG production is supported by the findings that PGs effectively prevent gastric ulcers, a property not apparently shared by strictly antisecretory agents of equal or greater antisecretory potency. However, well controlled clinical trials that directly compare the ability of different modalities to prevent NSAIDinduced ulcers are needed to confirm and extend these findings. Prevention of nonsteroidal antiinflammatory drug-induced duodenal ulcer. In considering prophylactic therapy, conclusions may need to be drawn separately for gastric and duodenal mucosa. H 2 -receptor antagonists and PGs both appear effective in preventing duodenal lesions in short-term, acute experiments in normal subjects (9); however, these data cannot necessarily be extrapolated to NSAID-associated duodenal ulcer in a clinical setting. McCarthy (see page 662) has reviewed preliminary studies suggesting that H2 -receptor antagonists reduce the frequency of acute duodenal ulcers in endoscopic trials. The number of duodenal ulcers occurring in the clinical trial evaluating prevention of NSAID-induced ulcers by PGs in osteoarthritic patients (19) was not sufficiently large to test adequately efficacy for prevention of acute duodenal ulcers.
ULCERS, NSAIDs AND RELATED MATTERS
Will Prevention of Acute Nonsteroidal Antiinflammatory Drug-Induced Ulcers Reduce Ulcer Complications? It is a reasonable hypothesis that reduction in the number of acute ulcers should result in a proportional reduction in the rate of complications. This hypothesis assumes that both classes of lesions (acute ulcers in endoscopic trials and ulcers presenting in real clinical settings) have a similar pathogenesis and are equally prone to complications and equally sensitive to preventative therapy. Confirmation that the prevention of NSAID-induced ulcers in short-term trials will translate into reduced NSAIDassociated chronic ulcers and complications awaits the outcome of long-term clinical trials. "Acute" NSAID-induced ulcers in endoscopic trials are surprisingly common (prevalence -10%-20%), whereas complications from NSAID-associated ulcers, although significant, are considerably less frequent. Ulcers that cause serious complications in patients treated with NSAIDs are deep, usually chronic ulcers, and are sometimes duodenal in location. Therefore, whether the response of acute gastric ulcers in clinical trials predicts the effects of preventative therapy on those gastric and duodenal ulcers destined to produce complications is a hypothesis that requires direct testing. It is important than such long-term trials be initiated as soon as prostaglandins become available in United States for the indication of prevention of NSAID-induced gastric ulcers.
Conclusions Based on available preliminary data, a reasonable operating hypothesis is that a regimen that significantly reduces the incidence or development of acute ulcers in patients on NSAIDs is likely to reduce NSAID-associated ulcer complications. Knowledge regarding the extent to which prevention of acute NSAID-induced lesions in duodenal and gastric mucosa translates into a reduction of ulcer complications awaits direct testing in long-term studies. In our view, three types of studies are required in the further assessment of the effectiveness of a prophylactic regimen in the prevention of NSAID-associated ulcers and their complications: (a) endoscopically-controlled studies comparing the effects of PGs and other antiulcer agents on the healing of gastric and duodenal ulcers during continuing NSAID consumption, (b) similar endoscopically controlled studies comparing PGs with other antiulcer agents in preventing gastric and duodenal ulcers in patients starting NSAID therapy, and (c)
SOLL ET AL.
long-term follow-up studies of patients during treatment with NSAIDs in combination with various prophylactic regimens, to determine the extent to which reductions in NSAID-induced acute gastric or duodenal ulcers, or both, can be translated into reduction of life-threatening ulcer complications and deaths. To reduce costs and make it possible to follow the large number of patients required to gauge the incidence of infrequent complications, this latter study would not necessarily include routine surveillance endoscopy, but rather would utilize endoscopy only when indicated clinically in order to establish the etiology of gastrointestinal hemorrhage, or the presence of ulcers suspected on other clinical grounds. As no clinical trials are yet available that test the effect of prophylactic therapy on complications of NSAID-associated ulcers, we believe that initial trials should compare a placebo limb with the prophylactic regimen. Placebo controls in long-term trials might raise ethical questions, as prophylactic regimens do reduce acute ulcers in patients taking NSAIDs. However, the impact of preventative cotherapy with antiulcer agents on ulcer complications in patients also taking NSAIDs needs to be delineated so that the real impact of cotherapy in reducing complication rates can be determined. A compromise would be to compare only active drugs in patients at higher risk (the elderly, those with serious concomitant disease, or with a history of smoking or peptic ulcer disease), whereas active drugs could be compared to placebo in patients without recognized risk factors for NSAID-induced ulcer complications. Such studies should greatly increase our knowledge of the effects of NSAID therapy on ulcer complications and how best to prevent and treat these complications. Prostaglandins are expected to be introduced in the United States early in 1989 for the primary indication of preventing NSAID-associated gastric ulcers. We can anticipate that this event will provide the impetus to expand our understanding of the pathophysiology and therapy of NSAID-associated gastrointestinal complications.
References 1. Robert A. Cytoprotection by prostaglandins. Gastroenterology 1977;77:761-7. 2. Hawkey q, Walt RP. Prostaglandins for peptic ulcer: a promise unfulfilled. Lancet 1986;ii:l084. 3. Soli AH. Duodenal ulcer and drug therapy. In: Sleisenger MH
GASTROENTEROLOGY Vol. 96, No.2, Part 2
and Fordtran JS, eds. Gastrointestinal disease. 4th ed. Philadelphia: WB Saunders, 1988:814-79. Lacy ER, Ito S. Microscopic analysis of ethanol damage to rat gastric mucosa after treatment with a prostaglandin. Gastroenterology 1982;83:619-25. Whittle BJR, Vane JR. Prostanoids as regulators of gastrointestinal function. In: Johnson LR, ed. Physiology of the gastrointestinal tract. Volume 1. 2nd ed. New York: Raven, 1987:143-80. Silen W. Gastric mucosal defense and repair. In: Johnson LR, ed. Physiology of the gastrointestinal tract. Volume 2. 2nd ed. New York: Raven, 1987:1055-70. Tarnawski A, Brzozowski T, Sarfeh IH, et al. Prostanoid protection of human isolated gastric glands against indomethacin and ethanol injury: evidence for direct cellular action of prostaglandin. J Clin Invest 1988;81:1081-9. Isenberg JI, Hogan DL, Koss MA, et al. Human duodenal mucosal bicarbonate secretion. Evidence for basal secretion and stimulation by hydrochloric acid and a synthetic prostaglandin El analogue~ Gastroenterology 1986;91:370-8. Lanza FL, AspinaH RL, Swabb EA, et al. Double-blind, placebo-controlled endoscopic comparison of the mucosal protective effects of misoprostol versus cimetidine on tolmetininduced mucosal injury to the stomach and duodenum. Gastroenterology 1988;95:289-94. Schwarz K. Uber penetrierende magen- und jejunal geschwure. Beitr Klin Chir 1910;76:96-128. Bugat R, Thompson MR, Aures D, et al. Gastric mucosal lesions produced by intravenous infusion of aspirin in cats. Gastroenterology 1976;71 :754-9. Carson JL, Strom BL, Morse L, et al. The relative gastrointestinal toxicity of the nonsteroidal anti-inflammatory drugs. Arch Intern Med 1987;147:1054-9. Rainsford KD. The effects of 5-lypoxygenase inhibitors and leukotriene antagonists on the development of gastric lesions induced by NSAIDs in mice. Agents Actions 1987;21:316-9. Semble EL, Turner RA, Wu WC. Clinical and genetic characteristics of upper gastrointestinal disease in rheumatoid arthritis. J Rheum 1987;14:692-9. Feldman M, Colturi TJ. Effect of indomethacin on gastric acid and bicarbonate secretion in humans. Gastroenterology 1984; 87:1339. Chalmers TC, Berrier J, Hewitt P, et al. Meta-analysis of randomized controlled trials as a method of estimating rates of complications of non-steroidal anti-inflammatory drug therapy. Aliment Pharm Therap 1988;2(Suppll):9-26. Armstrong CP, Blower AL. Non-steroidal anti-inflammatory drugs and life threatening complications of peptic ulceration. Gut 1987;28:527-32. Roth S, Agrawal N, Mahowald N, et al. Misoprostol heals gastroduodenal injury in patients with rheumatoid arthritis receiving aspirin. Arch Int Med 1989;in press. Graham, DY, Agrawal, N, Roth, S. Prevention of' NSAIDinduced gastric ulcer with the synthetic prostaglandin, misoprostol-a multicenter double-blinded placebo controlled trail. Lancet 1988;ii:1277-80.
Received November 7, 1988. Accepted November 23, 1988. Address requests for reprints to: Andrew H. Soli, M.D., Veterans Administration Medical Center (W151H), Building 115, Room 203, Los Angeles, California 90073. Andrew H. Soli is the recipient of a Medical Investigatorship from the Veterans Administration Wadsworth Hospital Center.