Addressing the Confusion over COXicity

Each year in the United States approximately 110,000 individuals are hospitalized and about 16,500 of them die due to complications associated with treatment with non-steroidal anti-inflammatory drugs (NSAIDs). The death rate is comparable to those of AIDS and leukemia. Potential fatal complications are bleeding ulcers, perforation and gastric outlet obstruction. NSAIDs, both prescription and non-prescription, account for approximately 75% of all peptic ulcer bleeding episodes in individuals 65 years of age and older. The national cost of treating acute complications of NSAID was estimated in 1999 to be approximately $2 billion.

The toxicity of traditional NSAIDs is related to their lack of specificity for the cyclooxygenase type-2 (COX-2) receptor, the inhibition of which diminishes pain. By binding both the COX-2 and COX-1 receptors, these agents reduce pain, but they may also inhibit prostaglandins and other cytoprotective substances in the gastrointestinal system, thus potentially inducing morbidity and, in some complicated cases, mortality.

In recent years, there have been two important breakthroughs in the effort to overcome this sequence of events. First, COX-2-specific medications were designed to reduce pain while sparing COX-1 receptors and avoiding the inhibition of cytoprotective substances. Second, proton pump inhibitors (PPIs), after protonation into their active species in the canalicular region of the parietal cell, reduce acid secretion and maintain a higher and more prolonged intragastric pH. Thus they may be used for prophylaxis during NSAID treatment. During this symposium, the faculty evaluated the roles of these two drug classes in reducing the risk of gastric ulcers and ulcer-related upper gastrointestinal complications.

This symposium was supported by an unrestricted educational grant from TAP Pharmaceuticals Inc.

PPIs to Reduce the Risk of NSAID-Associated Ulcers and Upper GI Complications

The principal risk factors for NSAID-associated ulcers and related upper GI complications, as reported by Rosemary R. Berardi, PharmD, FACCP, FASHP (University of Michigan), are a history of either peptic ulcer or upper gastrointestinal bleeding, the use of high-dose NSAIDs or concurrent corticosteroids, anticoagulation therapy, and the presence of a major illness. The risk appears to be age-dependent, increasing steeply at age 65 years and exponentially thereafter. There is less conclusive evidence that poor general health, excessive use of alcohol and smoking may also increase risk. The contribution of H. pylori colonization remains an unresolved controversy. There is increasing evidence that it plays an important role in ulcers, and it may be involved in NSAID-related bleeding. The presence of multiple risk factors greatly increases risk. For example, a 70-year-old individual with a history of ulcer-related upper gastrointestinal bleeding has a relative risk of up to 20 from taking NSAIDs at average therapeutic doses compared with a younger patient with no prior gastrointestinal bleeding. The risk increases as more risk factors are added.

Table 1 summarizes the principal strategies used to reduce the risk of symptoms, ulcers and complications (primarily bleeding) in patients treated with NSAIDs. Antacids, non-prescription histamine type-2 receptor antagonists (H₂RAs) and standard-dose prescription H₂RA compounds reduce symptoms, but offer no protection from ulceration and bleeding. By decreasing the pain and burning associated with untreated gastropathy, these agents may mask evolving pathological gastric conditions, potentially leading to sudden and catastrophic bleeding in asymptomatic patients. High-dose prescription H₂RAss also reduce symptoms and they may reduce the risk of gastric ulcers. Whether or not they reduce the risk of bleeding is uncertain. Sucralfate has no role in prophylaxis of NSAID-induced ulcers. It has no important effect on acid volume of pH and, therefore, is not as effective as antisecretory drugs in relieving NSAID-related symptoms. Sucralfate is not significantly more effective than placebo in reducing NSAID-associated ulcer or bleeding risk.

Misoprostol was the first agent approved by the FDA with an indication for reducing the risk of NSAID-associated ulcers. In a trial utilizing misoprostol 200 mcg twice, three times and four times per day, all three dosages significantly reduced the incidence of gastric ulcers compared with placebo but the lowest dose of 400 mcg per day was the least effective (Raskin JB et al. Ann Intern Med 1995;123(5):344). Despite its efficacy, however, misoprostol is not widely used because it does not relieve dyspepsia and because of prevalent (up to 40% of patients) diarrhea and abdominal pain at therapeutically-effective doses of 200 mcg three or four times per day. As the dose is decreased to resolve these adverse effects, the efficacy of the drug decreases. In the Raskin trial, for example, 4% of patients treated four times daily experienced gastric ulcers compared with 15.7% of the placebo controls (p<0.001), whereas with twice-daily dosing the incidence was 8.1% (p<0.002). While this is still statistically significant with respect to placebo, in absolute terms individuals with lower daily exposure to misoprostol had double the risk of gastric ulcer experienced by patients with the higher daily exposure.

Because standard doses of H₂RAs have been demonstrated not to be effective in reducing NSAID-associated gastric ulcers, high-dose regimens have been evaluated, primarily with famotidine 40 mg twice daily. In one such trial, which enrolled patients with osteoarthritis, rheumatoid arthritis, and gastric or duodenal ulcers, after 6 months of therapy, treated patients had an ulcer recurrence rate half that of patients given placebo, but the recurrence rate in treated patients remained high: approximately 25% (Hudson N. et al. Gastroenterol 1997; 112(6):1817). Furthermore, there is no established evidence that high-dose H₂RAs decrease the complications of ulcers.

In 1998, the New England of Medicine published two important papers reporting clinical trials in which PPI therapy was compared with an H₂RA and with misoprostol. In the first of these, 425 patients were randomized to omeprazole 20 mg per day or ranitidine 150 mg twice daily. Patients were followed for 6 months while on both NSAIDs and maintenance prophylaxis. The recurrence rates of gastric ulcer were 16.3% in patients treated with ranitidine compared with 4.5% in the omeprazole trial arm (Yeomans ND et al. N Engl J Med 1998;338(11):719). The H. pylori status of patients was not reported.

In the second of these two studies, patients were randomized to omeprazole 20 mg per day (N=274), misoprostol 200 mcg twice daily (N=296) or placebo (N=155) for the purpose of evaluating NSAID-related ulcer recurrence. The recurrence rate of gastric ulcers was significantly lower with both trial drugs with respect to placebo (p<0.001). Although not prospectively evaluated, patients who tested positive for H. pylori were more likely to remain in remission than H. pylori-negative subjects (Hawkey CJ et al. N Engl J Med 1998; 338(11):727). The investigators observed a slightly greater improvement with omeprazole than with misoprostol regarding gastric ulcer. However, the dose of misoprostol used was the lowest therapeutically active dose, and was used to avoid side effects and potential withdrawals from the study that are frequently associated with higher doses.

Graham and colleagues compared misoprostol at the full therapeutic dose — 200 mcg four times daily — with lansoprazole 15 mg and 30 mg daily in a randomized and placebo-controlled study involving 466 H. pylori-negative patients with histories of NSAID-related gastric ulcers but none at the time of enrollment. The primary endpoint was maintenance of healing. All of the treatments were more effective in maintaining healing than placebo after 12 weeks, and there was no difference between the two doses of lansoprazole. With the full dose of misoprostol, the prevalence of diarrhea was 22% compared with 3% and 7% for lansoprazole 15 mg per day and 30 mg per day, respectively. However, when the withdrawals were considered as treatment failures and the data were reanalyzed from patients who completed the study successfully, misoprostol and both doses of lansoprazole were equally efficacious in maintaining ulcer healing (Graham DY et al. Arch Intern Med 2002;162(2):169). Dr. Berardi concluded from these data that while full-dose misoprostol is an effective agent for maintaining ulcer healing, the frequency of intolerable adverse events places it at a practical disadvantage to lansoprazole.

In a 12-month placebo-controlled study involving 123 randomized patients taking low-dose aspirin (100 mg daily), lansoprazole 30 mg daily was associated with a lower rate of ulcer-related complications than placebo (p=0.006) and less recurrence of H. pylori infection than placebo (Lai KC et al. N Engl J Med 2002;346(26)2033). Investigators used occult blood and overt bleeding such as melena as bleeding endpoints. The influence of recurrent H. pylori infection on the trial data is unknown. (See Figure 1.) Thus, initial H. pylori eradication (preferably using a PPI-based three-drug regimen such as a PPI plus amoxicillin plus clarithromycin) followed by the continued use of a PPI may be the preferred option for reducing ulcer complications in H. pylori-positive patients taking low-dose aspirin.

The availability of two classes of drugs designed to reduce NSAID-related ulcers and upper GI complications —PPIs and COX-2-specific inhibitors — has led to comparison efficacy studies. In one such study, 115 high-risk NSAID users with healed NSAID-associated ulcers taking either naproxen 500-700 mg per day plus lansoprazole 30 mg per day or celecoxib 200-400 mg per day were observed for 6 months (Lai KC et al. Gastroenterol 2001:abstr). The complication rates in the two trial arms were essentially the same (3.5% and 3.4%, respectively), but this study was underpowered. In a similar trial, 287 arthritic patients with healed NSAID-associated ulcers were treated with diclofenac 75 mg twice daily plus omeprazole 20 mg daily or with celecoxib 200 twice daily (Chan et al. N Engl J Med 2002; 347:2104). Similar rates of recurrent complications were reported at 6 months with 6.3% for the NSAID plus PPI trial arm and 4.7% for celecoxib. In each of these trials, all patients were H. pylori-negative, and both used intent-to-treat data analysis. Although these studies lacked placebo groups and large numbers of patients, the results suggest that adding a standard dosage of a PPI to a conventional NSAID appears to be a suitable alternative to switching to a COX-2-specific inhibitor to reduce the risk of ulcer-related complications.

COX-2-Specific Inhibitors: Do They Increase GI Complications in Older Individuals?

Henry Cohen, MS, PharmD, BCPP, CGP (Long Island University, Brooklyn) presented an evidence-based evaluation of risks of gastrointestinal complications associated with the use of COX-2-specific inhibitors. Although celecoxib is marketed as a drug of this class, its relatively low specificity for the COX-2 receptor compared with rofecoxib and valdecoxib would disqualify it under the current FDA standard. Conversely, meloxicam, which is significantly more specific for the COX-2 receptor than is celecoxib, is not marketed as a COX-2-specific inhibitor because it did not meet the FDA standard at the time of approval.

In a large 12-week placebo-controlled trial comparing celecoxib 200 mg per day, celecoxib 400 mg per day and naproxen 500 mg twice daily, endoscopically observed gastroduodenal ulcers occurred significantly more frequently in the naproxen arm (p<0.001 vs. both placebo and celecoxib). However, there were no significant differences between naproxen and two dosages of celecoxib with respect to a broader spectrum of adverse gastrointestinal effects (Simon LS et al. JAMA 1999;282:1921). These data suggest that in comparison with NSAIDs, celecoxib causes less gastropathy, but not necessarily fewer gastrointestinal adverse effects such as dyspepsia and abdominal pain.

In the Celecoxib Long-term Arthritis Safety Study (CLASS), a randomized and double-blind study of approximately 8,000 individuals with rheumatoid arthritis or osteoarthritis, patients were treated with celecoxib 400 mg twice daily, diclofenac 75 mg twice daily or ibuprofen 800 mg three times per day for 6 months. The celecoxib dose was twice the usual dose for this population, and the other NSAID doses were standard. Approximately 20% of patients continued to take aspirin in doses up to 325 mg per day (Silverstein FE et al. JAMA 2000;284:1247). Analysis of data from non-aspirin users revealed that conventional NSAID exposure at normal dosages was associated with significantly higher annualized incidences of both ulcer complications and complications plus symptomatic ulcers compared with long-term exposure to celecoxib at double dosages. However, when data from aspirin users were analyzed, there was no statistical difference in the frequency of adverse events. This suggests that the addition of aspirin to celecoxib therapy negates some of the cytoprotective benefit of celecoxib. In addition, extension of therapy to 320 days resulted in an increase in adverse events with all agents both with and without aspirin. In the CLASS population, there were no significant differences in the incidence of cardiovascular or cerebrovascular events in either aspirin users or non-aspirin users.

Rofecoxib was the trial drug in the phase III Vioxx® GI Outcomes Research Trial (VIGOR) involving 8,076 patients with rheumatoid arthritis (Bombardier C et al. N Engl J Med 2000;343:1520) In this trial, patients were randomized to rofecoxib 50 mg daily or naproxen 500 mg twice daily. Low-dose aspirin was excluded, but patients were permitted to continue the use of H2RAs, acetaminophen, non-NSAID analgesics, glucocorticoids and disease-modifying antirheumatic drugs (DMARDs). Patients were followed for a median of 9 months for observation of clinically important upper gastrointestinal events and complications during treatment or within 14 days following discontinuation. Although the incidences of clinically important upper gastrointestinal events and complications of these events rose over time with both agents, the cumulative incidence of both endpoints was significantly higher in patients taking naproxen.

When the CLASS (celecoxib) and VIGOR (rofecoxib) cardiovascular data were compared, the incidence of myocardial infarction was the same with the two coxibs, low with naproxen, intermediate with diclofenac and equal to COX-2-specific inhibitors for ibuprofen.

In explaining the high frequency of peripheral vascular events in the rofecoxib arm of VIGOR, Dr. Cohen disputed the dogma that COX-2-specific inhibitors spare the kidneys. Both the afferent and efferent arterioles of the kidney contain COX-1 and COX-2 receptors, and the COX-2-specific inhibitors profoundly affect the renal vasculature and prostaglandins. To demonstrate this point, Dr. Cohen referred to a placebo-controlled trial comparing the effects of celecoxib 200 mg and 400 mg twice daily with naproxen 500 mg twice daily on cumulative sodium excretion in salt-depleted subjects (Rossat J et al. Clin Pharmacol Ther 1999;66:76). During days 1 through 3, all treatments were associated with significantly reduced sodium excretion relative to placebo that recovered completely during days 4 through 7. (Although this might suggest a transient effect, Dr. Cohen noted that this test was conducted in healthy subjects, not in elderly individuals with underlying diseases such as hypertension and congestive heart failure.) Similarly, the initiation of treatment with these agents resulted in an immediate decrease in urinary output and increased systemic sodium retention. In a series of rofecoxib phase I/II trials and the VIGOR study, escalating doses of rofecoxib, and especially the 50 mg dose, resulted in higher rates of hypertension and lower-extremity edema than ibuprofen, diclofenac and naproxen.

One of the legitimate distinguishing pharmacologic effects of coxibs as compared with NSAIDs is that they do not inhibit platelet aggregation. This has been confirmed, for example, in a study of bleeding times in individuals conducted after 8 days of treatment, 4 hours after dosing, with placebo, high- and low-dose rofecoxib, meloxicam, diclofenac, ibuprofen and naproxen (Van Hecken A et al. J Clin Pharm 2000;40:1109). Whereas ibuprofen and naproxen profoundly increased mean bleeding time, no significant changes were associated with the other medications. Naproxen reached its maximum degree of inhibiting platelet aggregation within 1 hour of dosing and sustained it for 8 hours. With ibuprofen, 40% of the effect was observed after 1 hour and the maximum amount after 2 hours. This was sustained only until approximately the fourth hour, after which the effect declined, reaching approximately 50% at hour 8. Because the COX-2-specific agents do not affect the coagulation cascade, they may have an especially important role in the pain management of patients who take aspirin or are on anticoagulant therapy. These medications will not inhibit platelet aggregation and, therefore, will not increase the risk of GI or non-GI bleeding events.

Following the formal presentations, John K. Siepler, PharmD (University of California Davis), symposium chair, led a series of brief case discussions in which the audience participated.

All contents Copyright © 1999 - 2003 Medical Association Communications