8

Nevertheless, the gastrointestinal advantages of these agents must be balanced against their possible adverse cardiovascular effects. Because of their antiplatelet activity, nonselective NSAIDs may be associated with less cardiovascular morbidity than COX-2-selective NSAIDs. However, for many patients, any oral agent may be less appropriate than topical agents such as pharmacist-compounded NSAID creams and ointments or capsaicin cream. Such agents do not pose the risk of systemic side effects, may be as effective as oral agents, and are highly acceptable to patients.

Because of its equivalent efficacy and superior safety compared with that of NSAIDs, acetaminophen has been recommended as the first-line therapy for osteoarthritic hip or knee pain in guidelines issued by the American College of Rheumatology (ACR). However, clinical trial evidence of the analgesic efficacy of acetaminophen is scanty. Only one small, 6-week, placebo-controlled study1 found acetaminophen, 1000 mg/day, significantly reduced osteoarthritic knee pain at rest, but the report of that 1983 study did not indicate whether it incorporated the standard study methodology in use today. Other non—placebo-controlled older studies concluded that acetaminophen was as effective as an NSAID in treating osteoarthritic knee pain. However, acetaminophen produced a statistically significant response in only some of the variables evaluated, whereas the NSAID studied (either ibuprofen or naproxen) did so in all of them. A well-designed, placebo-controlled, 12-week trial whose results were reported in 2003 found that, whereas acetaminophen, 1000 mg, given four times daily, did not improve osteoarthritic hip or knee pain, diclofenac did.² These findings are consistent with those of an earlier study that found that diclofenac was more effective than aceta-minophen for osteoarthritic knee or hip pain. However, because this earlier study was not placebo controlled, it did not establish the effectiveness of either treatment alone compared with placebo.

Because of its tissue-specific prostaglandin selectivity, acetaminophen does not promote bleeding in surgical patients or those with a bleeding diathesis and was not found to cause gastric injury in endoscopic studies. Moreover, no evidence exists to indicate that acetaminophen inhibits COX-1 in gastric mucosa even at high doses. However, results of recent studies indicate that acetaminophen may cause more adverse gastrointestinal events than previously believed. In a retrospective analysis of elderly patients, those taking high-dose acetaminophen (at least 2600 mg/day) had similar rates of adverse gastrointestinal events as those taking high-dose NSAIDs. However, after adjustment for gastrointestinal risk susceptibility, dyspepsia symptoms accounted for most of the gastrointestinal toxicity attributed to high-dose acetominophen.3 In a similar study of a general patient population, high-dose acetaminophen (2000 mg/day or more) conferred the same risk for upper gastrointestinal complications as traditional NSAIDs did and increased the risk of adverse gastrointestinal events even after adjustment for age and other gastrointestinal risk factors. Moreover, in this study, high-dose acetaminophen increased the risk of serious gastrointestinal adverse events such as ulcer and its complications.⁴ Nevertheless, the weight of clinical evidence indicates that the gastrointestinal risk conferred by acetaminophen therapy is less than that conferred by nonselective NSAID therapy, which was associated with more gastrointestinal side effects than acetaminophen in most direct comparisons.

As with gastrointestinal adverse events, acetaminophen may be more strongly associated with renal failure than previously believed. For example, heavy acetaminophen use (more than 365 doses annually) has been found to double the risk of end-stage renal disease.⁵ Although this finding did not prevent the National Kidney Foundation Scientific Advisory Committee from recommending acetaminophen for analgesia in patients with impaired renal function, a later report supported the association between acetaminophen and chronic renal failure.⁶

Because acetaminophen does not affect platelet aggregation, it does not have the antifibrinolytic effects that may be advantageous in patients at elevated risk of thrombosis. Nevertheless, acetaminophen has not been associated with increased cardiovascular risk. Moreover, unlike NSAIDs, acetaminophen has not traditionally been associated with acute bronchospasm. Notwithstanding, one study reported an association between acetaminophen use and asthma. In that study, the odds ratio for asthma was 2.4 in daily users compared with those who never used acetaminophen.⁷ Nevertheless, the causal link between acetaminophen use and asthma remains to be established in prospective studies. There are no data indicating that acetaminophen accelerates cartilage and bone degeneration, whereas some data indicate that NSAIDs such as indomethacin may do so.

Nonsteroidal Antiinflammatory Drugs

Because NSAIDs have both antiinflammatory and analgesic effects, they may be particularly useful for patients with detectable effusion indicating inflammatory osteoarthritis. Moreover, their analgesic effects are almost immediate, although their antiinflammatory effects require higher doses and 1 to 2 weeks of therapy.

Nonselective

Available Agents

In addition to the many different nonselective oral NSAIDs (Table 8–1), nonselective NSAIDs are also available in topical forms. These include methylsalicylate or capsaicin cream, which contains an extract from the chili pepper plant, or pharmacist-compounded NSAID creams such as diclofenac and ketoprofen. Topical agents are popular among patients and most useful in treating the knee or hand, to which they are more easily applied than they are to other areas. Moreover, they are particularly useful in patients who are unwilling or unable to take oral agents, and they are a valuable adjunct to oral agents as well.

^aOral ketorolac should only be used as continuation therapy after intramuscular ketorolac, and the maximum combined duration of use for both parenteral and oral ketorolac is 5 days.

Nonselective NSAIDs have consistently been found superior to placebo and, in general, are similarly effective in treating osteoarthritic pain. For example, in metaanalyses of treatments for hip osteoarthritis,8,9 differences in effectiveness between NSAIDs were rarely found. Nevertheless, indomethacin tended to be more effective as well as more toxic than other NSAIDs. More-over, low doses of naproxen (less than 750 mg/day) and ibuprofen (less than 1600 mg/day) were less effective for hip osteoarthritis than were other NSAIDs. In contrast, meta-analyses of treatments for knee osteoarthritis provide little evidence regarding the superior effectiveness of one NSAID over another at equivalent recommended doses.^10,11 However, although NSAIDS are in general similarly effective, much variability in response to different NSAIDs has been shown both between patients and in the same patient at different times. Topical NSAIDs have proven to be as effective as oral NSAIDs in clinical trials. Meta-analysis has found them to be twice as beneficial as placebo,¹² and substituting a topical NSAID for its oral counterpart can allow dose reduction with equivalent pain control.

The NSAIDs may be more effective than acetaminophen in treating osteoarthritis pain. Meta-analyses have found that NSAID-treated patients with knee osteoarthritis had significantly greater improvement in both pain at rest and pain on motion than acetaminophentreated patients.¹³ However, absolute differences between the two treatments were slight, and no significant difference in walking time or quality of life was found.¹³ Any superior effectiveness of NSAIDs relative to acetaminophen may result from the NSAIDs’ antiinflammatory effects in patients whose osteoarthritis has a low-grade inflammatory component. The possible superior efficacy of NSAIDs compared with that of acetaminophen may be more pronounced in patients with severe pain. However, the results of one study indicating that NSAIDs were superior for treating osteoarthritis patients with severe pain were confounded by reanalysis, which indicated that severity of pain did not predict a better response to ibuprofen therapy when given in antiinflammatory doses.¹⁴

Results of two large surveys indicated that patients with rheumatoid complaints preferred NSAIDs to acetaminophen. However, in one survey, more osteoarthritis patients than rheumatoid arthritis or fibromyalgia patients preferred acetaminophen, and in another, therapy was more likely to be continued for more than 2 years in patients taking acetaminophen than in those taking NSAIDs. Improved compliance with acetaminophen appeared to be related to its superior side effect profile compared with that of NSAIDs.

Adverse Effects and Safety

Nonsalicylate NSAIDs result in adverse effects in 10 to 15% of patients. These effects are mainly related to the ability of these agents to inhibit platelet aggregation, which increases risk of bleeding. As a result, the ACR recommends that a complete blood count (CBC) be obtained before treatment with these agents is initiated, and that CBCs be repeated annually thereafter. More frequent reevaluation of hemoglobin/hematocrit may be prudent in patients at highest risk of gastrointestinal bleeding. Inhibition of prostaglandin synthesis by nonselective NSAIDs in the gastrointestinal mucosa can injure the lining of the stomach or intestines, which, when combined with the increased risk of bleeding posed by these agents, may result in perforated ulcer, gastrointestinal hemorrhage, and death. In a study from the United Kingdom, mortality related to the gastrointestinal complications of NSAIDs was found to be higher than that related to asthma, and an analysis in the United States found that as many as 16,500 NSAID-related deaths occur annually in arthritis patients.

The renal toxicity of these agents is also a concern. Although moderate intake of NSAIDs did not increase the risk of renal toxicity, after more than 5000 doses were taken the odds ratio for renal toxicity increased to 8.8.5 Moreover, daily NSAID use has also been shown to increase the risk of chronic renal disease, although this risk was greatest in the elderly.¹⁵ Nonselective NSAIDs are he-patotoxic as well as renotoxic. As a result, baseline measurements of creatinine and aspartate aminotransferase and alanine aminotransferase levels are recommended before NSAID therapy is initiated. Reevaluation after 1 month of therapy and as clinically indicated thereafter may be warranted in patients at higher risk of renotoxicity such as diabetic and/or hypertensive patients and patients with congestive heart failure.

Nonacetylated salicylates such as choline magnesium trisalicylate or salsalate do not have the antiplatelet effects of nonselective nonsalicylate NSAIDs or as significant a degree of renal toxicity as that associated with nonselective nonsalicylate NSAIDs. As a result, these salicylates may be more appropriate for patients with renal insufficiency or for those at high risk of gastrointestinal or other bleeding than other nonselective NSAIDs. However, at effective doses, these salicylates pose a risk of central nervous system toxicity and ototoxicity. In addition, some nonselective NSAIDs, such as etodolac and meloxicam, are more selective for COX-2 than other nonselective NSAIDs and therefore spare gastric prostaglandin synthesis somewhat. However, clinical evidence of reduced gastrointestinal damage with these agents is sketchier than that for more COX-2-selective agents.

Next to lack of efficacy, gastrointestinal side effects such as dyspepsia, nausea, and vomiting are the main reason patients discontinue or change NSAIDs. Because gastropathy induced by nonsalicylate NSAIDs is dose-dependent, the lowest effective dose should be used. In addition, screening for Helicobacter pylori infection before initiating NSAID therapy is important because such infection increases the risk of NSAID-related peptic ulcer. The association between NSAID use and peptic ulcer was well established by the results of a meta-analysis of current NSAID users, who were found to have a higher incidence of peptic ulcer than nonusers.¹⁶ However, in another meta-analysis that controlled for the ulcer risk posed by H. pylori infection, the odds ratio for endoscopically determined ulcer related to NSAID use alone was 18, and when NSAID therapy was combined with H. pylori infection, it was 61. This controlled meta-analysis also found that, in patients with either NSAID use or H. pylori infection, the addition of either risk factor increased ulcer risk 3.5-fold.¹⁷ H. pylori treatment has been found to reduce the risk of ulcer from 34% to 12% and the risk of complicated ulcer from 27% to 4% over 6 months in patients starting NSAID therapy.¹⁸ Thus, if infected, patients should be treated for H. pylori infection before NSAID therapy is initiated.

In patients at high risk of serious adverse gastrointestinal events, gastroprotective agents should be used even when low-dose nonselective NSAID therapy is prescribed. Although the most commonly used gastroprotective agent is misoprostol, 200 μg three or four times daily, it may cause diarrhea and flatulence when combined with an NSAID. Alternatives include histamine-2 receptor antagonists such as famotidine, which in the usual doses does not prevent gastric ulcers. Other possibilities include omeprazole, which, like high-dose famotidine, has been shown to prevent NSAID-related gastropathy in endoscopic studies. In patients with existing ulcers, omeprazole, 20 to 40 mg/day, is as effective as misoprostol, 200 μg twice daily, but better tolerated. In addition, lansoprazole (Prevacid), and esomeprazole (Nexium) are now Food and Drug Administration (FDA)-approved for the prevention of gastric ulcers in NSAID users at increased risk of developing gastrointestinal bleeding, perforations, or ulcers. Although proton-pump inhibitors are commonly used and have also been shown to prevent ulcer in patients taking NSAIDs, their cost may be limiting. Co-medication may also be required for NSAID-related constipation or diarrhea.

The NSAIDs may also promote fluid retention, edema, and hypertension. Therefore, when NSAIDs are combined with antihypertensive therapy in patients at high risk of these complications because of preexisting heart failure, hypertension, renal insufficiency, or impaired liver function, blood pressure should be monitored regularly. Even in patients who are not at high risk of hypertension, blood pressure should be checked a few weeks after NSAID therapy is initiated and two to three times annually thereafter. Because, like their gastrointestinal adverse effects, the hypertensive effects of NSAIDs are dose-dependent, these effects can be minimized by using the lowest effective dose.

Controversy exists as to whether the use of NSAIDs increases cardiovascular disease risk. Early findings regarding nonselective NSAIDs were inconclusive, suggesting either no effect or cardioprotection. However, a more recent study of more than 180,000 Medicaid patients found that use of various nonaspirin NSAIDs excluding COX-2-selective agents was not associated with altered risk of serious coronary heart disease or stroke compared with that found for controls.¹⁹ Nevertheless, ibuprofen may impair the ability of aspirin to confer car-dioprotection, as indicated by the effect of adding ibuprofen to aspirin on serum thromboxane formation and platelet aggregation in vitro. In contrast, like acetaminophen, diclofenac was not found to impair platelet disaggregation by aspirin.

Nonselective NSAIDs may induce bronchospasm by diverting arachidonic acid to the lipoxygenase pathway and increasing leukotriene synthesis. Thus, these agents should be avoided in patients with aspirin-sensitive asthma. In addition, some NSAIDs have been associated with worsening of degenerative joint disease. NSAIDs in vitro may retard, accelerate, or have no effect on osteoarthritis progression depending on the agent used. Indomethacin in particular may accelerate degeneration of cartilage and bone; piroxicam, diclofenac, and naproxen appear to have no such effect; and meloxicam may be chondroprotective. However, no clinical evidence exists to indicate that NSAIDs have favorable effects on cartilage. Moreover, although simple headache is a common side effect of NSAIDs, these agents may in rare cases produce aseptic meningitis. NSAIDs also interact with several drugs, which may pose problems for patients on multidrug regimens. The adverse effects of NSAID therapy can be avoided by the use of topical agents, which results in a nominal blood level of drug that limits systemic effects and enhances safety compared with that of oral NSAIDs. Meta-analysis has shown that topical NSAIDs are as safe as placebo,12 and other studies have shown no association between topical agents and gastrointestinal bleeding.

COX-2 Selective

Available Agents and Mechanism of Action

Three COX-2-selective NSAIDs are available: celecoxib (Celebrex), rofecoxib (Vioxx), and valdecoxib (Bextra). In addition, although widely used in Europe and Mexico, the COX-2-selective NSAID etoricoxib (Arcoxia) was still under FDA review in the spring of 2005 for approval for U.S. use, as was parecoxib, an injectable prodrug of valdecoxib. At therapeutic doses, COX-2-selective NSAIDs exclusively inhibit prostaglandin synthesis through the COX-2 pathway, thereby sparing prostaglandin synthesis in the gastrointestinal mucosa.

Dosage and Indications

Effective dosages for osteoarthritis pain are celecoxib, 100 to 200 mg/day; rofecoxib, 12.5 to 25 mg/day; valdecoxib, 10 mg/day; etoricoxib, 60 to 90 mg/day; and parecoxib, 20 to 40 mg/dose. An acute pain indication also exists for rofecoxib, 50 mg/day, for up to 5 days, and for celecoxib, 400 mg, for the first dose, which may be followed by subsequent doses of 200 mg. However, no acute pain indication exists for valdecoxib. Parecoxib, which can be given either intravenously or intramuscularly, may be particularly useful for patients who need rapid control of moderate to severe pain or who are unable to take oral drugs because of surgical procedures or serious gastrointestinal distress.

The COX-2-selective NSAIDs can be useful for patients who have not responded adequately to nonpharmacologic treatments and maximum doses of acetaminophen or nonselective NSAIDs. However, greater efficacy of COX-2-selective NSAIDs compared with that of acetaminophen or nonselective NSAIDs has not been shown in these patients in clinical trials. COX-2-selective NSAIDs may also be used as initial therapy, particularly in patients with moderate to severe pain and inflammation. Because of their mechanism of action, COX-2-selective NSAIDs pose less risk of adverse gastrointestinal events than traditional NSAIDs but are similarly effective. As a result, guidelines recommend that COX-2-selective agents be reserved for use in patients at high risk of adverse gastrointestinal events. However, even in patients who are not at such high risk, the reduction in gastrointestinal side effects associated with COX-2-selective agents could be expected to result in better compliance and less need to change drugs, prescribe co-medications, or perform procedures. These advantages may thus support more widespread use of COX-2-selective NSAIDs.

Meta-analysis results have indicated that this class of agents appears to be as effective as nonselective NSAIDs.20 However, no direct evidence indicates that one COX-2-selective agent is more effective than another.

Adverse Effects and Safety

Rofecoxib and celecoxib have an incidence of acute gastrointestinal injury and chronic ulceration that is half of that reported for nonselective NSAIDs. Use of COX-2-selective NSAIDs resulted in less risk of complicated and symptomatic ulcer after 6 to 9 months of therapy than use of nonselective NSAIDs in the Celecoxib Long-term Arthritis Safety Study (CLASS)²¹ and the Vioxx Gastrointestinal Outcomes Research (VIGOR) study.²²However, longer term study of celecoxib did not confirm this risk reduction after 12 or 16 months of therapy.²³These negative results may have resulted from problems with study methodology and concomitant use of low-dose aspirin. For instance, in CLASS, celecoxib only reduced the rate of upper gastrointestinal complications in patients who were not taking aspirin concomitantly.²¹ In contrast, the VIGOR study did not allow patients to take aspirin concomitantly. These studies thus indicate that COX-2-selective NSAIDs pose less risk of gastrointestinal side effects especially in patients who are not currently taking aspirin. However, other data suggest less gastrointestinal bleeding results from using a COX-2-selective NSAID with aspirin than from using a nonselective NSAID with aspirin. COX-2-selective NSAIDs may also be less likely to cause nuisance gastrointestinal side effects than nonselective NSAIDs. For example, celecoxib, 50 to 400 mg twice daily, appears slightly less likely to cause minor gastrointestinal symptoms, such as abdominal pain, dyspepsia, and nausea, than naproxen, 500 mg twice daily. Similarly, in the VIGOR study, rofecoxib was better tolerated than naproxen.

Unlike nonselective NSAIDs, COX-2-selective NSAIDs do not decrease platelet aggregation or increase bleeding time and thus may be used for surgical patients, patients with bleeding diathesis, and cautiously in those taking warfarin. Moreover, although all NSAIDs have been associated with hypertension, the cost of managing blood pressure destabilization was found to be significantly less for celecoxib than for higher-dose rofecoxib or nonspecific NSAIDs (diclofenac, ibuprofen, or naproxen). In addition, users of selective agents are more likely to concomitantly use other medications that may increase blood pressure than users of nonselective NSAIDs. Nevertheless, coprescribing antiplatelet agents with COX-2-selective NSAIDs appears prudent if there are no contraindications to combination therapy, particularly considering that combining 81 to 325 mg/day of aspirin with COX-2-specific NSAIDs has been found to pose a lower risk of endoscopically identified ulcers than combining aspirin with nonselective NSAIDs.

Although COX-2-selective NSAIDs were initially not expected to be renotoxic because of their selectivity, two separate investigative groups concluded that these agents may be so in certain patient populations. Thus, COX-2-selective agents should be used with caution in patients with mild to moderate renal impairment and avoided in patients with severe renal insufficiency. Similar precautions apply in patients with liver disease because, like nonselective NSAIDs, COX-2-selective agents may also be hepatotoxic. In addition, use of celecoxib and valdecoxib is contraindicated in patients who are allergic to sulfonamides because these COX-2-selective agents have a sulfonamide component. However, although some nonselective NSAIDs have been associated with the progression of osteoarthritis in in vitro studies, less is known about the effects of COX-2-selective NSAIDs in this regard. In addition, a few cases of aseptic meningitis related to rofecoxib use have been reported, as were several cases possibly related to celecoxib use.

Although the nonselective NSAID ketorolac is typically used in the emergency department and postoperatively for acute pain, it is associated with gastrointestinal ulceration and renal functional impairment. Moreover, because it predisposes patients to increased perioperative bleeding, it is contraindicated for preoperative use. In contrast, the COX-2-selective NSAID parecoxib does not impair platelet aggregation and produced no more gastrointestinal erosions or ulcers than placebo in healthy volunteers in a short-term trial. Moreover, parecoxib does not appear to have any significant drug interactions that may prevent its use in patients taking other drugs concurrently and appears equal to ketorolac and superior to morphine in controlling postsurgical pain in knee replacement patients. Therefore, once approved, parecoxib will probably replace ketorolac as the injectable NSAID of choice.

Addendum: While this manuscript was in press, rofecoxib was withdrawn worldwide on September 30, 2004, due to an increased risk of cardiovascular events seen in the Adenomatous Polyp Prevention on Vioxx (APPROVe) Trial.

The FDA Arthritis Advisory Committee and Drug Safety and Risk Management Advisory Committee subsequently met (February 16–18, 2005). Following public testimony and scientific presentations, the panel voted unanimously to advise the FDA that Vioxx, Bextra, and Celebrex can cause cardiovascular problems. In a separate vote a majority of the panel advised against removing these medications from the market.

The specific recommendations were: