Chapter 61. Nonsteroidal Anti-Inflammatory Drugs

Nonsteroidal anti-inflammatory drugs (NSAIDs) are anti-inflammatory, analgesic, and antipyretic agents. They are used to reduce pain, decrease stiffness, and improve function in patients with osteoarthritis (OA), rheumatoid arthritis (RA), and other forms of arthritis. They are also used for the treatment of pain including headache, dysmenorrhea, and postoperative pain.1–3 Whether their effectiveness is solely due to their anti-inflammatory or analgesic effects or other possible mechanisms is not known.4 There are at least 20 different NSAIDs currently available in the United States (Table 61-1). In addition, cyclooxygenase-2 selective inhibitors (COX-2 inhibitors, e.g., celecoxib, rofecoxib), with similar efficacy but significantly decreased gastrointestinal (GI) and platelet effects, are available.5–8

NSAIDs are one of the most commonly used classes of drugs. It has been reported that more than 17,000,000 Americans use these agents on a daily basis for the relief of pain and, at times, swelling related to inflammation.9 With the aging of the US population, the Centers for Disease Control predict a significant increase in the prevalence of painful degenerative and inflammatory rheumatic conditions and thus an increased use of NSAIDs.9,10 Approximately 60 million NSAID prescriptions are written each year in the United States; the number for elderly patients exceeds those for younger patients by approximately 3.6-fold.10 Aspirin, ibuprofen, naproxen, and ketoprofen are also available over the counter. At equipotent doses, the clinical efficacy and tolerability of the various NSAIDs are similar; however, individual responses are highly variable.1,11,12 It is believed that if a patient fails to respond to one NSAID of one class that it is reasonable to try another NSAID from a different class; however, no one has studied this in a prospective controlled manner.11,12

Sodium salicylic acid was discovered in 1763 but more impure forms of salicylates had been used as analgesics and antipyretics throughout the previous century. Once purified and synthesized, the acetyl derivative of salicylate, acetylsalicylic acid (ASA) was found to provide more anti-inflammatory activity than salicylate alone, but at the same time increased the incidence of toxicity, particularly related to the upper GI tract. Phenylbutazone, an indoleacetic acid derivative, was introduced in the early 1950s. This drug was a weak prostaglandin synthase inhibitor, which induced uricosuria, and was rapidly found useful in patients with ankylosing spondylitis and gout. However, owing to concerns related to bone marrow toxicity, particularly in women older than the age of 60, this compound now is rarely prescribed. Indomethacin, another indoleacetic acid derivative, was subsequently developed in the 1960s to substitute for phenylbutazone. It had the potential for significant toxicity as well, and the search for safer (particularly GI safer) and at least equally effective NSAIDs ensued. Other clinical issues have driven the development of newer agents, such as once or twice daily dosing to improve compliance.

Some of the variability in clinical response may be explained by the spectrum of inhibition of prostaglandin synthesis. Some NSAIDs appear to be potent inhibitors of prostaglandin synthesis, while others more prominently affect other nonprostaglandin-mediated biologic events.1,2,13–16 Different responses have also been attributed to variations in the enantiomeric state of the drug or its pharmacokinetics and/or pharmacodynamics.1,2,11,12 Although variability can be explained in part by absorption, distribution, and metabolism, potential differences in mechanism of action must be considered as possibly important to explain observed variable effects.11,12

The NSAIDs are primarily anti-inflammatory and analgesic by decreasing production of prostaglandins of the E series.17 Many of these prostanoic acids are pro-inflammatory, and increase vascular permeability and sensitivity to the release of bradykinins. Decreasing the synthesis of these mediators leads to decreased pain, swelling, and edema in the peripheral tissues. In addition, there is accumulating evidence that central effects of pain modulation may be as important as the effects on the peripheral tissues. The hypothesis is that prostaglandin synthesis is upregulated in the brain with peripheral stimulation of pain particularly associated with inflammation, and those NSAIDs that are more lipophilic penetrate better into the central nervous system (CNS) and inhibit both synthesis of peripheral and central prostaglandins.18

NSAIDs have also been shown to inhibit the formation of prostacyclin and thromboxane, resulting in complex effects on vascular permeability and platelet aggregation in peripheral tissues, which undoubtedly contributes to the overall clinical effects of these compounds.

These prostaglandins are derived from polyunsaturated fatty acids that are constituents of all cell membranes. They exist in ester linkage in the glycerols of phospholipids and are converted through multiple enzymatic steps to prostaglandins or leukotrienes first through the action of phospholipase A2 or phospholipase C.17 Free arachidonic acid, released by the phospholipase from the fatty acids, acts as a substrate for the PGH synthase complex, which includes both cyclooxygenase (COX) and peroxidase. These enzymes catalyze the conversion of arachidonic acid to the unstable cyclic-endoperoxide intermediates, PGG2 and PGH2, which are then converted to the more stable PGE2 and PGF2 compounds by specific tissue prostaglandin syntheses. NSAIDs specifically inhibit COX and thereby reduce the conversion of arachidonic acid to PGG2.

At least two isoforms of the COX enzymes have now been identified. They are products of two different genes yet share 60% homology in the amino acid sequences considered important for catalysis of arachidonic acid. The differences are primarily in their regulation and expression.19,20 COX-1, or prostaglandin synthase H1 (PGHS-1), regulates normal cellular (physiologic) processes and is stimulated by hormones or growth factors. It is constitutively expressed in most tissues, and is inhibited by all NSAIDs to varying degrees depending on the applied experimental model system used to measure drug effects.21–24 It has an important role in maintaining the integrity of the gastric and duodenal mucosa, and many of the toxic effects of the NSAIDs on the GI tract are attributed to its inhibition.25–30 It has been described as a “housekeeping enzyme.”

The other isoform, prostaglandin synthase H2 or PGHS-2 (or COX-2) is an inducible enzyme and is usually undetectable in most tissues. Its expression is increased during states of inflammation or experimentally in response to mitogenic stimuli. In monocyte/macrophage systems, endotoxin stimulates COX-2 expression; in fibroblast studies various growth factors, phorbol esters, and interleukin-1 do so.19,31 This isoform is also constitutively expressed in the brain, specifically the cortex and hippocampus, in the female reproductive tract, the male vas deferens, in bone, and at least in some models, in human kidney.19,20 In the brain it appears that COX-2 is upregulated with increased inflammation-induced pain inpulses; thus, the inhibition of COX-2 in the brain is thought to be an important modulator of pain in states of inflammation.18 The expression of COX-2 is inhibited by glucocorticoids.19,20,31 COX-2 is also inhibited by all of the currently available NSAIDs to a greater or lesser degree, and its inhibition leads to a decrease in those prostanoid products associated with increased pain and swelling.20–23 Therefore, we have observed the effects of prolonged inhibition of COX-2 for the last 25 years as we have used traditional nonselective NSAIDs.

The in vitro systems used to define the actions of the available NSAIDs are based on using cell-free systems, pure enzyme, or whole cells systems.21 Each drug studied to date has demonstrated different measurable effects within each system. As an example, it appears that nonacetylated salicylates inhibit the activity of COX-1 and COX-2 in whole cell systems but are not active against either COX-1 or COX-2 in recombinant enzyme or cell membrane systems. This suggests that salicylates act early in the arachidonic acid cascade similar to glucocorticoids, perhaps by inhibiting enzyme expression rather than direct inhibition of cyclooxygenase.

Recent accumulated evidence has demonstrated that several NSAIDs are selective and inhibit the COX-2 enzyme more so than the COX-1 enzyme. For example, in vitro effects of etodolac and meloxicam demonstrate primary inhibition of COX-2 compared with COX-1, at low doses.32,33 However, at higher approved therapeutic doses this effect appears to be mitigated, as both COX-1 and COX-2 are inhibited to variable degrees. Very potent, specific COX-2 inhibitors that have no measurable effect on COX-1 mediated events at therapeutic doses are now available: celcoxib and rofecoxib.34 Both of these COX-2 specific inhibitors (or COX-1 sparing drugs) have been shown as effective at inhibiting osteoarthritis pain, dental pain, and the pain and inflammation associated with rheumatoid arthritis as naproxen at 500 mg twice daily, ibuprofen 800 mg three times daily, and diclofenac 75 mg twice daily, without endoscopic evidence of gastroduodenal damage and without affecting platelet aggregation.5–8,34–36 Unfortunately, because of the design of the randomized controlled trials, many of the important questions regarding the renal effects of the specific COX-2 inhibitors continue to remain unanswered.19,20

Other possible mechanisms of action that may explain the clinical effects of the NSAIDs include several physicochemical properties of these drugs. NSAIDs are variably lipophilic and become incorporated in the lipid bilayer of cell membranes and thereby may interrupt protein-protein interactions important for signal transduction.13,14 For example, stimulus response coupling, which is critical for recruitment of phagocytic cells to sites of inflammation, has been demonstrated in vitro to be inhibited by some NSAIDs.14 There are data suggesting that NSAIDs inhibit activation and chemotaxis of neutrophils as well as reduce toxic oxygen radical production in stimulated neutrophils.37,38 There is also evidence that several NSAIDs scavenge superoxide radicals.37

Salicylates have been demonstrated to inhibit phospholipase C activity in macrophages. Some NSAIDs have been shown to affect T-lymphocyte function experimentally by inhibiting rheumatoid factor production in vitro. Another newly described action not directly related to prostaglandin synthesis inhibition includes interference with neutrophil-endothelial cell adherence, which is critical to migration of granulocytes to sites of inflammation. These data demonstrate that expression of L-selectins are decreased.16 NSAIDs have been demonstrated in vitro to inhibit NF-kB (nitric oxide transcription factor) dependent transcription, thereby inhibiting inducible nitric-oxide synthetase.15 Anti-inflammatory levels of ASA have been shown to inhibit expression of inducible nitric-oxide synthetase and subsequent production of nitrite in vitro. At pharmacologic doses, sodium salicylate, indomethacin, and acetaminophen when studied had no effect, but at suprapharmacologic doses, sodium salicylate inhibited nitrite production.15

Recently, it has been described that prostaglandins inhibit apoptosis (programmed cell death) and that NSAIDs, via inhibition of prostaglandin synthesis, may reestablish more normal cell cycle responses.19,20 There is also evidence suggesting that some NSAIDs may reduce PGH synthase gene expression, thereby supporting the clinical evidence of differences in activity in NSAIDs in sites of active inflammation.

The NSAIDs have been demonstrated to have variable effects on many biologic processes; however, how important some of these effects are clinically remains unknown. Although nonacetylated salicylates have been shown in vitro to inhibit neutrophil function and to have equal efficacy in patients with rheumatoid arthritis,39 clinically there is no evidence to suggest that these biologic effects are more important than prostaglandin synthase inhibition.

NSAIDs are efficiently absorbed after oral administration, but absorption rates may vary in patients with altered gastrointestinal blood flow or motility. Certain NSAIDs when taken with food have decreased absorption.1,2 Enteric coating may reduce direct effects of NSAIDs on the gastric mucosa but may also reduce the rate of absorption.

Most NSAIDs are weak organic acids; once absorbed they are >95% bound to serum albumin. This is a saturable process. Clinically significant decreases in serum albumin levels or institution of other highly protein-bound medications may lead to an increase in the free component of NSAID in serum. This may be important in patients who are elderly or are chronically ill, especially those with associated hypoalbuminemic states. Importantly, as a result of increased vascular permeability in localized sites of inflammation, this high degree of protein binding may result in delivery of higher levels of NSAIDs.

NSAIDs are metabolized predominantly in the liver by the cytochrome P450 system and the CYP 2C9 isoform, and excreted in the urine. This must be taken into consideration when prescribing NSAIDs for patients with hepatic or renal dysfunction. Some NSAIDs (e.g., indomethacin, sulindac, and piroxicam) have a prominent enterohepatic circulation, resulting in a prolonged half-life and should be used with caution in the elderly. In patients with renal insufficiency, some inactive metabolites may be re-synthesized in vivo to the active compound. Diclofenac, flurbiprofen, celecoxib, and rofecoxib are metabolized in the liver. These agents should be used with care and at lowest possible doses in patients with clinically significant liver disease, if used at all, including patients with cirrhosis with or without ascites, prolonged prothrombin times, falling serum albumin levels, or important elevations in liver transaminases in blood.

As noted, most of the NSAIDs and celecoxib are metabolized through the P450 CYP29 isoenzyme, whereas rofecoxib is not. The exact mechanism to explain the metabolism of rofecoxib remains unexplained.

Salicylates are the least highly protein bound NSAID: approximately 68%. Zero order kinetics are dominant in salicylate metabolism. Thus, increasing the dose of salicylates is effective over a narrow range, but once the metabolic systems are saturated, then incremental dose increases may lead to very high serum salicylate levels. Thus, changes in salicylate doses need to be carefully considered at chronic steady-state levels, particularly in patients with altered renal or hepatic function.

Significant differences in plasma half-lives of the NSAIDs may be important in explaining their diverse clinical effects. Those with long half-lives typically do not attain maximum plasma concentrations quickly and important clinical responses may be delayed. In most chronic conditions that are appropriate for the use of these drugs, the acute effects are not as important as in the treatment of headache or acute pain. Plasma concentrations can vary widely because of differences in renal clearance and metabolism. Piroxicam has the longest serum half-life of currently marketed NSAIDs: 57 ± 22 hours. In comparison, diclofenac has one of the shortest: 1.1 ± 0.2 hours (see Table 61-1). Although drugs have been developed with very long half-lives to improve patient compliance, in the older patient it is sometimes preferable to use drugs of shorter half-life so that, when the drug is discontinued, any unwanted effects may more rapidly disappear.

Sulindac and nabumetone are “pro-drugs” in which the active compound is produced after first-pass metabolism through the liver. Theoretically, pro-drugs were developed to decrease the exposure of the gastrointestinal mucosa to the local effects of the NSAIDs. Unfortunately, as was noted, with adequate inhibition of COX-1 the patient is placed at substantial risk of an NSAID-induced upper GI event as long as COX-1 activity is inhibited. This is true for drugs such as ketoralac given as an injection or by these pro-drugs when given at adequate therapeutic doses.40

Other pharmacologic properties may be important clinically. NSAIDs that are highly lipid soluble in serum will penetrate the central nervous system more effectively and occasionally may produce striking changes in mentation, perception, and mood.41,42 Indomethacin has been associated with many of these side effects, even after a single dose, particularly in the elderly.

Mechanism-Based Adverse Effects

Risk for Anaphylaxis and Pulmonary Effects

Many adverse reactions attributed to NSAIDs are due to inhibition of prostaglandin synthesis in local tissues (Table 61-2). For example, patients with allergic rhinitis, nasal polyposis, and/or a history of asthma, in whom all NSAIDs effectively inhibit prostaglandin synthase, are at increased risk for anaphylaxis. In high doses, even nonacetylated salicylates may sufficiently decrease enough prostaglandin synthesis to induce an anaphylactic reaction in sensitive patients.43 Although the exact mechanism for this effect remains unclear, it is known that E prostaglandins serve as bronchodilators. When COX activity is inhibited in patients at risk, a decrease in synthesis of prostaglandins that contribute to bronchodilation results. Another explanation implicates other enzymatic pathways that utilize the arachidonate pool after it is converted from phospholipase, whereby shunting of arachidonate into the leukotrienne pathway occurs when cyclooxygenase is inhibited.44 The leukotriene pathway converts arachidonate by 5-lipoxygenase, leading to products such as LTB4 as well as others clearly associated with anaphylaxis. This explanation implies that large stores of arachidonate released in certain inflammatory situations lead to excess substrate for leukotriene metabolism. This results in release of products that are highly reactive, leading to increased bronchoconstriction and the risk for anaphylaxis in the right patient. Whether the main mechanism of effect is inhibition of prostaglandin synthesis or shunting of arachidonate into conversion by 5-lipoxygenase or a combination of the two, it is clear that patients who are sensitive are at great risk when NSAIDs are used. The nonacetylated salicylates as a group have been considered a safe choice for these patients because they are known to possess anti-inflammatory activity, but are relatively weak cyclooxygenase inhibitors.

Platelet Effects

Platelet aggregation and thus the ability to clot is primarily induced through stimulating thromboxane production with activation of platelet COX-1. There is no COX-2 in the platelet. NSAIDs and aspirin inhibit the activity of COX-1 but the COX-2 specific inhibitors (or COX-1 sparing drugs) have no effect on COX-1 at clinically effective therapeutic doses.19,20

The effect of the nonsalicylate NSAIDs on platelet function is reversible and related to the half-life of the drug; whereas the effect of ASA is to acetylate the COX-1 enzyme, thereby permanently inactivating it. Since platelets cannot synthesize new cyclooxygenase enzyme after exposure to ASA, the platelet does not function appropriately for its lifespan. Therefore, the effect of ASA on the platelet does not wear off as the drug is metabolized as with the nonsalicylate NSAIDs. Patients awaiting surgery should therefore stop their NSAIDs at a time determined by 4 to 5 times their serum half-life; whereas ASA needs to be discontinued 1 to 2 weeks before the planned procedure to allow for re-population of platelets that have been unexposed to ASA.

We also have little information about the use of the COX-2 specific inhibitors in patients at risk for thrombosis.19 The randomized clinical trials of the COX-2 specific inhibitors were not designed to address this question. Furthermore, we have little information demonstrating that the traditional NSAIDs are safer or more useful than the COX-2 specific inhibitors in this regard. Only aspirin has been studied prospectively, and low-dose aspirin (<325 mg/day) should be given concomitantly with either NSAIDs or specific COX-2 inhibitors in patients at risk for thrombosis.45 Secondary prophylaxis of cardiovascular events with aspirin has been shown to be important. There is about a 23% to 33% decrease in atherothrombotic events in patients with a history of ischemic heart disease when they are treated with <325-mg aspirin/day.45 There are less data about the effectiveness of low-dose aspirin in primary prophylaxis. There is little evidence that the non-aspirin NSAIDs have a role in primary or secondary prophylaxis of cardiovascular or cerebrovascular events. Given the additive ulcerogenic potential associated with the use of multiple NSAIDs, it would be advisable to use specific COX-2 inhibitors with aspirin when considering combination cardioprotective and anti-inflammatory therapies. In the two large bleeding trials describing the effects of celecoxib and rofecoxib, there was a statistically significant difference between the incidence of acute myocardial infarctions (MIs) with rofecoxib (0.4%) and naproxen (0.1%), whereas there was no difference in the incidence of acute MIs with celecoxib as compared with diclofenac or ibuprofen.46,47 It is impossible to know whether these data suggest that rofecoxib induced more MIs than naproxen or whether naproxen exerted a protective effect. The study was too short with relatively too few patients to answer this question.

Gastrointestinal Tract

The most clinically significant adverse effects associated with NSAIDs occur in the GI tract, affecting the GI mucosa.48–57 These events appear to be due to local or systemic inhibition of prostaglandin synthesis. NSAIDs cause a wide range of GI problems including symptoms of intolerance such as dyspepsia, nausea, vomiting, as well as esophagitis, esophageal stricture, gastritis, mucosal erosions, hemorrhage, peptic ulceration and/or perforation, obstruction, and death.27,48,49,58–62 The unpleasant symptoms may be observed in about 40% to 60% of patients but are not related to COX-1 inhibition in the GI mucosa.27,28 The exact cause of these symptoms remains unknown; however, use of H2-receptor antagonists and proton pump inhibitors typically alleviates these unpleasant effects. Erosions and ulcers as well as ulcer complications are predominantly due to the effects of NSAID-induced inhibition of COX-1, thus those prostaglandins that serve to protect the stomach mucosa from toxins. This protection consists of a mucous layer serving as a barrier, a bicarbonate gradient serving to buffer the mucosa layer from the effects of the extremely acidic lumen, developing glutathione to serve as a scavenger of superoxides, prostaglandin-mediated inhibition of gastric acid, and prostaglandin-mediated increases in mucosal blood flow. All of these effects are inhibited by the NSAIDs, which inhibit COX-1 activity. However, in an experiment with mice both COX-1 and COX-2 have to be inhibited in order for mice to have an ulcer. The exact effect of inhibiting COX-2 in this experiment is unknown, but some investigators have speculated that COX-2 plays an important role in healing of mucosal lesions.63,64

The mucosa of the large and small bowel may also be affected by NSAIDs. These agents in the small or large bowel may also induce stricture formation,13,27,52–54,61–63 which may manifest as diaphragms precipitating small or large bowel obstruction, and can be hard to detect on contrast radiographic studies.

Additionally, there is evidence to suggest that NSAIDs interfere with permeability of the GI mucosa. The weakly acidic NSAIDs rapidly penetrate the superficial lining cells of the GI mucosa leading to oxidative uncoupling of cellular metabolism, local tissue injury, and ultimately cell death. This can result in local erosions, hemorrhages, and formation of clinically significant ulcers in the right patients.13

Endoscopic studies have clearly demonstrated that NSAID administration results in shallow erosions and/or submucosal hemorrhages that are observed in the stomach near the prepyloric area and the antrum, although they may occur at any site in the GI tract.27 Typically, these lesions are asymptomatic, making prevalence data difficult to determine.58 As a result, we do not know the number of lesions that spontaneously heal or which will progress to ulceration, frank perforation, gastric or duodenal obstruction, serious gastrointestinal hemorrhage, or subsequent death. Risk factors for the development of GI toxicity in patients receiving NSAIDs include age >60 years, prior history of peptic ulcer disease, prior use of antiulcer therapies for any reason, concomitant use of glucocorticoids, particularly in patients with rheumatoid arthritis, comorbidities such as significant cardiovascular disease, or patients with severe rheumatoid arthritis (Table 61-3).25,27,48,49,64,65 Other risk factors include increasing dose of specific and singular NSAIDs.

The magnitude of risk for GI adverse events is controversial. The US Food and Drug Administration (FDA) reports an overall risk of 2% to 4% for NSAID-induced gastric ulcer development and its complications.25,27,28 In general, based on multiple clinical trials, the relative risk is estimated to be 4.0 to 5.0 for gastric ulcer, 1.1 to 1.6 for duodenal ulcer, and 4.5 to 5.0 for clinically significant gastric ulcers with hemorrhage, perforation, obstruction, or death. The accurate absolute risk is harder to determine.

As noted, other sites in the GI tract, including the esophagus and small and large bowel, may also be affected. Exposure to NSAIDs is probably a major factor in the development of esophagitis and subsequent stricture formation.58,59 Effects on small and large bowel have increasingly been reported.27,53 An autopsy study on 713 patients showed that small bowel ulceration defined as ulcers >3 mm in diameter were observed in 8.4% of patients exposed to NSAIDs compared with 0.6% of nonusers of NSAIDs.61 Ulcerations of stomach and duodenum were observed in 22% of NSAID users compared with 12% of nonusers.

Before the COX-2 selective inhibitors became available, the epidemiologic studies suggested that the nonacetylated salicylates are least likely to result in a NSAID-induced adverse GI event. Agents such as nebumatone are usually considered to be less likely to induce effects.56,57 However, as higher doses of “safe” NSAIDs are used, then more GI damage is reported.57 Thus at equally efficacious doses the non-selective NSAIDs all induce ulcers at variable rates, and any of these ulcers in the right patient could develop a complicated course. NSAIDs with prominent enterohepatic circulation and significantly longer half-lives such as sulindac and piroxicam have been linked to increased potential for GI toxicity attributed to prolonged re-exposure of gastric and duodenal mucosa to bile reflux and the active moiety of the drug.27

Endoscopic data from large numbers of patients treated with COX-2 selective inhibitors strongly suggest that ulcers occur at the same rate as in patients who received placebo; whereas the traditional NSAID active comparators induced ulcers (as documented by endoscopy) in 15% (diclofenac 75 mg twice daily, ibuprofen 800 mg three times daily) to 19% (naproxen 500 mg twice daily) following 1 week of treatment in healthy volunteers and in 26% (naproxen 500 mg twice daily) of patients with OA and RA after 12 weeks of treatment.5–8 Two large outcome trials with rofecoxib and celecoxib have now been published and have shown that rofecoxib and celecoxib both at suprapharmacologic doses (two to four times the treating doses) are associated with about two- to three-fold fewer GI complications than naproxen or ibuprofen, respectively, at standard therapeutic doses.46,47 These data clearly show that compared with the effects of two widely used NSAIDs (ibuprofen 1.2%, naproxen 1.6%), that the COX-2 inhibitors induce bleeding complications at a lower rate (celecoxib 0.4%, rofecoxib 0.6%).46,47 It is possible that patients with preexisting ulcer may experience delay in healing when treated with a COX-2 specific inhibitor, but only long-term outcome clinical trials will clarify if this is a risk.63

Approach to the Patient at Risk for NSAID-Induced GI Adverse Events

The approach to the patient with pain and or inflammation who is about to embark on the use of a NSAID is partly related to dose, consistency, and planned length of therapy. For most patients who will take standard doses of nonselective NSAIDs for less than 10 days, there is little risk unless they have multiple significant risk factors noted previously. However, for the patient with OA or other chronic condition at risk for an NSAID-induced GI event, the decision remains somewhat controversial. Many patients with dyspepsia or upper GI distress have superficial erosions evident on endoscopy that frequently heal spontaneously without change in therapy. Even more difficult to evaluate is whether cytoprotective agents actually alter NSAID-associated symptoms, which may or may not predict significant GI events. Although one clinical study demonstrated that >80% of patients who developed significant NSAID-induced endoscopic abnormalities were asymptomatic,58 several prospective observational trials indicated that patients were more symptomatic with NSAID-induced toxicities than previously thought.55

The patient who develops a gastric or duodenal ulcer while taking NSAIDs should have treatment discontinued and therapy for ulcer disease, either H2-antagonist or proton pump inhibitors, instituted.27–29 If NSAIDs must be continued concomitantly, then the patient will be required to receive antiulcer therapy for longer periods.28 Typically, most patients with uncomplicated gastric or duodenal ulcers will heal within 8 weeks of initiating H2-antagonists. If NSAID treatment is continued, then perhaps 16 weeks of therapy may be necessary for adequate healing. Diagnostic tests to determine if the patient is Helicobacter pylori positive should be performed, and if the patient has positive studies, then specific antibiotic therapy to eradicate the infection should be administered.28

Prophylaxis to prevent NSAID-induced gastric and duodenal ulcers is more complicated. To date there has been no evidence that agents other than concomitant misoprostol therapy will prevent NSAID-induced gastric ulceration and its complications.66–69 Although high-dose H2-antagonists or proton pump inhibitors have been demonstrated to prevent NSAID-induced gastric and duodenal ulcers, prevention of ulceration complications has not been clearly shown. Endoscopic trials have shown that famotidine at twice the approved dose (40 mg twice daily) significantly decreased the incidence of both gastric and duodenal ulcers.69 Similarly, an endoscopy trial demonstrated that treatment with omeprazole (a proton pump inhibitor) decreased gastroduodenal ulcers.68 Both H2-antagonists and proton pump inhibitors decrease dyspeptic symptoms quite effectively.

Misoprostol is a prostaglandin analogue, believed to locally replace prostaglandins whose synthesis in the gastroduodenal mucosa is inhibited by the nonselective NSAIDs.64–66 A large prospective trial evaluated 8,843 patients with rheumatoid arthritis to determine whether misoprostol would decrease the incidence of ulcers and their complications.64,65 Patients received various NSAIDs and were followed for 6 months either on misoprostol co-therapy or placebo. The study was powered on the basis of endoscopic observations of an 80% decrease with concomitant misoprostol therapy in endoscopically proven ulcers >0.3 to 0.5 cm in diameter in the gastric and duodenal mucosa.66 Misoprostol successfully inhibited development of ulcer complications such as bleeding, perforation, and obstruction. There was a 40% reduction in patients treated with misoprostol as opposed to those receiving placebo.64 Further analysis demonstrated that patients with health assessment questionnaire (HAQ) scores >1.5 (thus worse disease) had an 87% reduction in risk for a NSAID-induced toxic event if concomitantly treated with misoprostol.65

These data suggest that high-risk patients may benefit from concomitant misoprostol therapy if NSAID treatment is indicated. Gabriel and colleagues have demonstrated the pharmacoeconomic utility of such therapy in the high-risk patient.70 Unfortunately, the major adverse event causing withdrawal in approximately 10% of patients was diarrhea, and 30% of patients complained of diarrhea. Therefore medications such as stool softeners and cathartics should be stopped. There are data suggesting that concomitant treatment with misoprostol once an ulcer develops will allow the ulcer to heal.68

Renal Adverse Effects

The NSAIDs also have effects on the kidneys. The effects of the NSAIDs on renal function include changes in the excretion of sodium, changes in tubular function, the potential for interstitial nephritis, and reversible renal failure due to alterations in filtration rate and renal plasma flow.71,72 Prostaglandins and prostacyclins are important for maintenance of intrarenal blood flow and tubular transport. All NSAIDs, except nonacetylated salicylates, have the potential to induce reversible impairment of glomerular filtration rate; this effect occurs more frequently in patients with congestive heart failure; established renal disease with altered intrarenal plasma flow including diabetes, hypertension, or atherosclerosis; and with induced hypovolemia, salt depletion or significant hypoalbuminemia.19,71,72 Triamterene-containing diuretics, which increase plasma renin levels, may predispose patients prescribed NSAIDs to acute renal failure. NSAIDs have been implicated in the development of acute and chronic renal insufficiency, due to inhibition of vasodilating prostaglandins, thereby reducing renal blood flow.

NSAID-associated intersitial nephritis is typically manifested as nephrotic syndrome, characterized by edema or anasarca, proteinuria, hematuria, and pyuria. The usual stigmata of drug-induced allergic nephritis such as eosinophilia, eosinophiluria, and fever may not be present. Interstitial infiltrates of mononuclear cells are seen histologically with relative sparing of the glomeruli. Phenylproprionic acid derivatives, such as fenoprofen, naproxen, and tolmetin along with the indoleacetic acid derivative indomethacin, are most commonly associated with the development of interstitial nephritis.

Inhibition of prostaglandin synthesis intrarenally by NSAIDs decreases renin release and thus produces a state of hyporeninemic hypoaldosteronism with resulting hyperkalemia.71 This effect may be amplified in patients taking potassium-sparing diuretics. Salt retention precipitated by some NSAIDs leading to peripheral edema in some patients is likely due to both inhibition of intrarenal prostaglandin production, which decreases renal medullary blood flow and increases tubular reabsorption of sodium chloride as well as other direct tubular effects. NSAIDs have also been reported to increase antidiuretic hormone effects, thereby reducing excretion of free water, resulting in hyponatremia.71 Thiazide diuretics may produce an added effect on the NSAID-induced hyponatremia. All NSAIDs have been demonstrated to interfere with medical management of hypertension and heart failure.

All NSAIDs, including the COX-2 inhibitors with the exception of the nonacetylated salicylates, have been associated with increases in mean blood pressure in hypertensive patients but not in patients with normal blood pressure.74 Patients receiving antihypertensive agents including beta blockers, ACE inhibitors, thiazide, and loop diuretics must be checked regularly when initiating therapy with a new NSAID to ensure that there are no significant continued and sustained rises in blood pressure.

The mechanism of acute renal failure induced in the at-risk patient treated with NSAIDs is believed to be prostaglandin mediated.71–73 However, the role of COX-2 in maintenance of renal homeostasis in the human remains unclear. COX-2 activity is notably present in the macula densa and tubules in animals and man, and is upregulated in salt-depleted animals.75 In humans, COX-1 is an important enzyme for control of intrarenal blood flow. It is believed that COX-2 activity importantly modulates salt and water homeostasis, whereas COX-1 activity seems important in modulating renal plasma flow. In the patient who has decreased renal plasma flow, both COX-1 and COX-2 are upregulated and therefore there is not sufficient evidence to indicate that the COX-2 specific inhibitors will be safer than traditional NSAIDs in terms of renal function. Until the appropriate clinical trials are done, any patient at high risk for renal complications should be monitored very carefully. No patient with a creatinine clearance of <30 mL/minute should be treated with either a NSAID or a COX-2 specific inhibitor.

Nonmechanism-Based Adverse Events

Hepatotoxicity

Elevations in hepatic transaminase levels induced by NSAIDs are not uncommon, although it occurs more often in patients with juvenile rheumatoid arthritis or systemic lupus erythematosus. Unless elevations exceed two to three times the upper limit of normal, or serum albumin falls, or prothrombin times are altered, these effects are usually not considered clinically significant.76–78 Nonetheless, overt liver failure has been reported following use of many NSAIDs, Including diclofenac, flurbiprofen, and sulindac.77,78 Of all NSAIDs, sulindac has been associated with the highest incidence of cholestasis.76 Therefore it is recommended that patients at risk for liver toxicity be followed very carefully. When initiating NSAID treatment, all patients should be evaluated again within 8 to 12 weeks and serious consideration given to performing a blood analysis for serum transaminase changes.

Idiosyncratic Adverse Effects

Many of the toxic effects of NSAIDs are related to their mechanism of action via prostaglandin inhibition, but there are also important potential idiosyncratic effects. A typical nonspecific reaction includes skin rash and photosensitivity, which is associated with all currently available NSAIDs, particularly the phenylproprionic acid derivatives.79 This same class of NSAID derivative may also induce aseptic meningitis, especially in patients with systemic lupus erythematosus. The underlying mechanism of action remains unknown. Ibuprofen has also been associated with a reversible toxic amblyopia.79

Owing to the antiplatelet effects of all NSAIDs, except the nonacetylated salicylates, concomitant therapy with warfarin (Coumadin) puts patients at greater risk for bleeding. As concomitant NSAID therapy would displace warfarin from its albumin binding sites, the prothrombin time may be prolonged. In addition, given the increased relative risk for NSAID-induced gastroduodenal ulcers and bleeding, there is an increased risk for bleeding when the NSAIDs are used concomitantly with warfarin. In that the COX-2 specific inhibitors do not cause ulcers of the GI tract, nor do they alter platelet function, the patient on warfarin would have less risk for a significant GI bleed when treated with these drugs than traditional nonselective NSAIDs. Effects such as these may also be seen with dilantin or other highly protein-bound drugs such as antibiotics.

The NSAIDs inhibit the renal excretion of lithium and should be used with caution in patients taking this drug. Cholysteramine, an anion-exchange resin, reduces the rate of NSAID absorption and its bioavailability.

The CNS side effects of NSAIDs include aseptic meningitis, psychosis, and cognitive dysfunction.1,41,42 The latter changes are more commonly seen in elderly patients treated with indomethacin, whereas the phenylproprionic acid derivatives are more commonly associated with the development of aseptic meningitis and toxic amblyobia. Tinnitus is a common problem with higher doses of salicylates as well as the nonsalicylate NSAIDs. The mechanism is unknown. Interestingly, the young and the elderly may not complain of tinnitus but only of hearing loss. Other NSAIDs may also induce tinnitus in specific patients. Decreasing the dose usually alleviates the effect.

It has been shown that COX-2 is important for ovulation through the PPRA l receptor.19 In addition, COX-2 is upregulated with implantation of a fertilized ovum or in decidualization. Although there are a few case reports of reversible infertility associated with the use of NSAIDs, given the large numbers of patients who regularly use NSAIDs there does not appear to be a generalized epidemic of infertility.80

Use of NSAIDs does not lead to osteoporosis.81–82 The role of COX-1 remains unclear. Although inflammation in the joint leads to juxtaarticular osteopenia, this is the result of increased prostaglandin synthesis in the inflamed joint, which is likely directly related to increased COX-2 activity.

Some of the early available NSAIDs have been associated with an increased risk for bone marrow failure. This is particularly true of phenylbutazone and indomethacin. Strom and colleagues have described the incidence of neutropenia as a toxic effect of the NSAIDs.83 In a case-controlled study performed using Medicaid claims data, these investigators defined that the adjusted odds ratio for neutropenia in patients treated with NSAIDs is 4.2 (confidence interval (CI), 2.0–8.7). When patients treated with either phenylbutazone or indomethacin were excluded, the odds ratio for the development of neutropenia remained quite robust: 3.5 (CI 1.6–7.6). In general, given the common use of NSAIDs, the risk of neutropenia is quite small.

There are little data documenting the effects of the NSAIDs on pregnancy or the fetus. In animal models, the NSAIDs have been shown to increase the incidence of dystocia, post-implantation loss, as well as delay of parturition and miscarriage.80 The effect of prostaglandin inhibition may result in premature closure of the ductus arteriosus. ASA has been associated with smaller babies and neonatal bruising; however, it has been used for many years in the treatment of patients who require NSAIDs while pregnant. Typically, therapy with ASA is stopped about 8 weeks prior to delivery to decrease the risk for interfering with ductus closure. In animals, there is no evidence that ASA is a teratogen. The NSAIDs are excreted in breast milk. It is believed that salicylates in normally recommended doses are not considered dangerous to nursing infants.

Although NSAIDs are known to decrease pain and inflammation and to be antipyretic, they have not been shown to decrease erosions in rheumatoid arthritis, to retard osteophyte formation in osteoarthritis, or to protect cartilage from mechanical or inflammatory injury. However, they continue to be important drugs for the palliation of pain and inflammation. Newer forms of COX-2 inhibitors are in development and are considered even better analgesics because of rapid uptake in the CNS. Time will tell the ultimate role these drugs will play as more is learned about their complex physiologic effects.