Is the association between non steroidal anti-inflammatory agents and reflux esophagitis significant?
A. Lanas (Zaragoza), B.I. Hirschowitz (Birmingham, Alabama)
It is well known that non steroidal anti-inflammatory drugs (NSAIDs) may induce damage in the gastroduodenal and intestinal mucosa. Mounting evidence suggests that these drugs may also damage the esophagus, but the nature and extent of the problem remain uncertain. Recent clinical and experimental reports indicate that NSAID use might be associated with a wide spectrum of symptoms and lesions related to the esophagus, including heartburn, dyspepsia, chest pain, dysphagia and esophagitis with ulceration, bleeding and even rarely perforation [1-3]. They also suggest that the prevalence and seriousness of the problem may be underestimated and underdiagnosed. The data discussed below indicate also that NSAIDs may be further involved in the development of stricture, but there is no evidence for their involvement in Barrett's esophagus .
Recent studies have shown that NSAID use commonly induces dyspepsia and heartburn. Heartburn was initially present in about 30-40% of patients using different NSAIDs who were enrolled in two different prospective double-blind controlled trials which compared the effects of omeprazole with ranitidine or misoprostol on gastroduodenal ulcer healing . Those patients who used omeprazole obtained faster and better relief of heartburn when compared to other treatments, indicating the probable role of acid exposure in heartburn, including that found in NSAID users. In clinical trials of healthy volunteers taking NSAIDs, endoscopically-diagnosed esophagitis was found in 15.7-20% of cases [6, 7]. It has also been reported that 25% of chronic symptomatic NSAID users had endoscopically proven esophagitis vs 15% in controls , and esophagitis is also a common finding in patients with refractory gastroduodenal ulcer due to chronic aspirin abuse . In a prospective study  we determined objectively the extent of NSAID and aspirin use in patients with upper gastrointestinal acid-peptic diseases by supplementing conventional history taking with two tests of current aspirin use - high performance liquid chromatography (HPLC) for the presence of salicylate in serum, and platelet cyclo-oxygenase activity, which detects the use of aspirin within 5 days of testing. Of
186 consecutive patients undergoing diagnostic upper gastrointestinal endoscopy, 62% of 55 patients with endoscopically proven esophagitis had evidence of current NSAID use, compared to 26% of 42 control patients with normal endoscopy (p < .001). By contrast, 36% of 25 patients with active peptic ulcer but only 12% of 17 patients with recently healed ulcer were positive for NSAID use (Figure 1). Using logistic regression analysis to examine all possible factors that might contribute to esophagitis in this population age, sex, arthritis, alcohol and NSAID-use only NSAID use contributed significantly to the diagnosis, but not specifically or directly to the major associated symptoms (chest pain in 54%, dysphagia in 44%, and heartburn in 67%). In this population, 95% of NSAID use was chronic and 84% of the NSAID was aspirin. We questioned  whether NSAIDs contributed both to the development of esophagitis and the commonly found resistance to treatment with H2-receptor antagonists, and even in the rapid relapse of healing after treatment is stopped. However, in contrast a recent study by Taha et al.  showed that long-term NSAID users had fewer histological abnormalities than patients not receiving NSAIDs, i.e., macroscopic esophagitis was not necessarily dependent on pre-existing microscopic esophagitis.
Figure 1. Prevalence of aspirin/NSAID use in 139 consecutive patients undergoing upper endoscopy and diagnosed of esophagitis, active peptic ulcer (PUD), healed PUD and normal endoscopy. TxI = thromboxane index; patients positive for this test had current aspirin use within 5 days of testing (see  for further details).
Stricture develops in a relatively small percentage of patients with esophagitis  but evidence from a few studies suggests that NSAIDs may play a role in such an outcome in the esophagus as it does in the pylorus  and rectum . Heller  found NSAID use in 31% of patients with stricture of the esophagus requiring dilation, compared to 14% in controls. Wilkins et al.  found 49% of 53 conscutive patients with esophageal stricture had been prescribed NSAIDs in the year before study, compared to 12% of 165 controls. We  found 12 of 55 esophagitis patients requiring dilation; 67% were currently ASA NSAID users compared to 26% of controls. Because of small numbers, the prevalence of use was not different from that with lesser degrees of esophagitis (Table I). However, an extension of the study  by Marks et al., showed 80% of 30 stricture patients were using NSAIDs (mainly ASA) compared to 26% of controls (p < .01).
Mechanism of action
Although, as described above, there is evidence of an association between NSAID use and esophageal symptoms and disease, possible mechanisms of such an effect are not clearly understood. It is not clear whether the way in which such injury could occur would also be responsible for "peptic" ulcer, i.e. acid co-dependent lesions as well as small and large gut ulcers, which are acid-independent. The esophagus could be damaged by local action due to lodging of various different tablets including NSAIDs, producing so-called "pill esophagitis". This is represented by painful ulceration, often severe and which may heal with stricture. The lesion may occur at any age, but most commonly in the elderly due to altered motility, and in those who swallow tablets with little liquid, especially if taken at bedtime. Pill esophagitis may seriously complicate a pre-existing peptic stricture resulting from reflux, and should be considered in the differential diagnosis of stricture with proximal ulceration. Chronic NSAID use may further delay healing of esophageal peptic ulcer as it does for gastroduodenal peptic ulcers, and by promoting collagen production in the scar, additionally lead to stricture .
Mucosal defenses are complex and multiple . NSAIDs may act by disrupting the mucosal barrier to H+. Such a barrier to acid back diffusion has been shown in man . This involves the hydrophobic lipid bilayer cellular membrane, intercellular barriers of the junctional complexe intracellular buffering for hydrogen ions, and the ability to extrude H+ to the interstitial fluid via the Na+/H+ antiport . NSAIDs may act as barrier breakers and allow access of refluxed H+ and pepsin to the basement membrane, with loss of epithelial cells and diffuse mucosal damage. In a recent study Lanas et al.  examined the injurious effects of aspirine acid and pepsin on esophageal mucosa. They studied the effects of both intraluminal and parenteral aspirin on an in vivo rabbit model of esophagitis induced by acidified (pH 2.0) pepsin. The mucosal barrier integrity was determined by H+, K+ and hemoglobin flux rates, and esophageal injury was assessed by macroscopic and microscopic scoring, including cell proliferation indexes. Acidified saline alone caused no damage, but the addition of ASA caused mucosal barrier damage. Exposing esophageal mucosa to acidified aspirin followed by acidified pepsin significantly increased mucosal injury and barrier dysfunction compared to control experiments with acidified saline or acidified pepsin alone (Figure 2). Prostaglandin (PGE2) co-therapy applied before the acidified aspirin exposure significantly (> 40%) reduced the subsequent damage. Histological studies showed a significant increase in cell proliferation in those rabbits that were pretreated with prostaglandin E2. Also, aspirin applied at pH 6 led to less severe damage (Figure 3) but did not totally prevent it. Parenterally administered aspirin also promoted damage by acidified pepsin, but the damage was 23% less than that caused by intraluminal acidified aspirin.
Experimental evidence from these studies shows both local acid exposure (partly acid-dependent) and parenteral aspirin (not acid-dependent) render the esophageal mucosa more permeable to acid and pepsin, which may then damage the mucosa by digesting collagen of the basement membrane and by causing epithelial cell death [17, 21]. Aspirin may thus act both directly on the mucosa via barrier breaking, and indirectly through inhibition of prostaglandin production which could affect other mechanisms of mucosal defense . Indirect effects of inhibition of prostaglandin production could conceivably act via effects on lower esophageal sphincter (LES), and on smooth muscle of the esophageal body, both of which are affected by prostaglandins . Such effects could contribute to increased acid exposure. This was studied  in 9 healthy volunteers given naproxen, which did not modify either the time of exposure to pH < 4.0, nor the number of reflux episodes, esophageal clearance time or LES pressure. Only one subject with an already low LES pressure had an increase in acid reflux. A more recent study  also showed that as a group, patients with esophagitis and chronic NSAID use had similar motility and gastroesophageal reflux (GER) patterns as patients with esophagitis not using NSAIDs. Together these data suggest that NSAID use does not sgnificantly affect LESP or the reflux patterns leading to esophagitis. NSAIDs, especially ASA, contribution to esophagitis is thus more likely to be due to mucosal injury, but not all acid-dependent. The prominent contribution of acid to mucosal injury mechanisms, both via aspirin and pepsin, provides a mechanistic explanation for the therapeutic benefit of strong acid suppression in treating reflux esophagitis.
Figure 2. Effects of both intraluminal and parenteral (I.V.) aspirin on the esophageal mucosal damage induced by acidified pepsin. In the experiments the esophageal mucosa was exposed to acidified saline with/without aspirin followed by acidified pepsin .
Figure 3. Effects of prostaglandin E2 (PGE2)
(2 mg/kg; s.c.) on the esophageal mucosal damage induced by acidified aspirin followed by acidified pepsin. In this model, the esophageal perfusion of a non-acidified aspirin solution (pH 6) induces less damage than acidified aspirin .
16. Hunt TK. The principles of wound healing. In: Halter F, Garner A, Tytgat GNJ, eds, Falk Symposium 59. Mechanisms of peptic ulcer healing. Dordrecht, The Netherlands: Kluwer Academic Publisher, 1991:1-12.
21. Hirschowitz BI. Pepsin in the pathogenesis of peptic ulceration. In: Halter F, Garner A, Tytgat G, eds. Mechanisms of peptic ulcer healing. Dordrecht, The Netherlands: Kluwer Academic Publishers, 1990:183-194.
22. Baunack AR, Demol P, Froese G, et al. Effect of rioprostil, an oral prostaglandin E1 analogue, on lower oesophageal sphincter pressure and on the motility of the distal oesophagus in healthy volunteers. Scand J Gastroenterol 1989;24(suppl164):120-126.
23. Scheiman JM, Patel PM, Henson EK, Nostrant TT. Effect of naproxen on gastroesophageal reflux and esophageal function: a randomized, double-blind, placebo-controlled study. Am J Gastroenterol 1995;90:754-757.