The use of NSAIDs is ubiquitous in rheumatology because of their effectiveness as anti-inflammatory and analgesic agents. In addition to their use in rheumatoid arthritis (RA) and osteoarthritis (OA), NSAIDs are widely used in the symptomatic management of other rheumatic diseases characterized by chronic musculo-skeletal pain and diverse forms of acute pain. NSAIDs differ widely in their chemical class, but share the property of blocking production of prostaglandins (PGs) [1]. This is accomplished by inhibiting the activity of the enzyme PGG/H synthase, also called cyclooxygenase (COX). COX occurs in two isoforms, COX-1 and COX-2, which differ in their tissue distribution and regulation. The isoforms serve different biological functions, in that COX-1 is expressed under basal conditions and is involved in the biosynthesis of PG serving homeostatic functions while COX-2 expression is increased during inflammation and other pathologic situations [2]. The clinical effects of NSAIDs are evaluated in terms of effects on the different COX isoforms. Inhibition of COX-2 by NSAIDs blocks PG production at sites of inflammation or other forms of tissue damage, while inhibition of COX-1 in certain other tissues - most importantly, platelets and the gastroduodenal mucosa - can lead to common adverse effects of NSAIDs such as bleeding and gastrointestinal ulceration [3].

COX isoform specificity, however, is only one factor that impacts the efficacy and adverse effect profile of individual NSAIDs. Most traditional NSAIDs inhibit both isoforms, albeit with some differences in the relative potency for COX-1 and COX-2. Some NSAIDs lack inhibition of platelet function, which is the operational definition of COX-2-selective NSAIDs [4]. The pharmacologic properties, including chemical class, formulation, and drug half-life, of individual drugs may be equally important in determining the properties of NSAIDs. In light of the widespread use of NSAIDs for common diseases, which are likely to increase in prevalence with the aging of the population, it is critically important to appreciate the potential adverse events associated with NSAIDs to use them safely in patients with rheumatic diseases.

Almost all NSAIDs are >90% bound to plasma proteins. If total drug concentrations are increased beyond the point at which the binding sites on albumin are saturated, biologically active free-drug concentrations increase disproportionately to the increasing total drug concentration. The clearance of NSAIDs is usually by hepatic metabolism, with production of inactive metabolites that are excreted in the bile and urine. Most NSAIDs are metabolized through the microsomal cytochrome P450-containing mixed-function oxidase system. NSAIDs are most often metabolized by CYP3A or CYP2C9, or both. However, some are metabolized by other cytosolic hepatic enzymes. Different patients can respond to the same NSAID in a variety of ways and the basis for this individual variability remains unclear. Several pharmacologic factors related to NSAIDs may influence this variability, such as dose response, plasma half-life, enantiomeric conversion, urinary excretion, and pharmaco-dynamic variation [7]. Such drug factors include protein binding, the metabolic profile of the drug, and the percentage of the drug that is available as the active (S) enantiomer. There is also genetic variability in the cytochrome P450 metabolic enzymes such that some individuals or ethnic groups metabolize drugs more slowly. For example, Asians are frequently slow metabolizers through the CYP2C9 pathway. Finally, the pharmaco-kinetics of some NSAIDs are affected by hepatic disease, renal disease or old age.

NSAIDs exert their actions by inhibiting enzymatic activity of the COX enzymes. These enzymes are the first committed step in the synthesis of PG from arachidonic acid (Figure 1). Arachidonic acid is an omega-6 poly-unsaturated fatty acid commonly found at the sn-2 position of cell membrane glycerophospholipids and cleaved from cell membranes by one of several different phospholipase A₂ enzymes [8]. COX-1 and COX-2 are bifunctional enzymes that mediate a COX reaction whereby arachidonate plus two molecules of oxygen are converted to the cyclic endoperoxide PGG₂, followed by a hydroperoxidase reaction in which PGG₂ undergoes a two-electron reduction to PGH₂ [8]. The unstable inter-mediate PGH₂ spontaneously rearranges or is enzymatically converted by specific synthases to biologically active PG, of which there are many isoforms [9]. The overall regulation of the type and amount of PG produced in a given cell or tissue is determined by the expression levels of COX-1, COX-2, and terminal synthase enzymes.

All of the NSAIDs are synthetic inhibitors of the COX active site, but subtle mechanistic differences in the manner in which individual NSAIDs interact and bind with the active site are responsible for some of the differences in their pharmacologic characteristics [10]. Acetylsalicylic acid is the only covalent, irreversible modifier of COX-1 and COX-2, whereas all of the other NSAIDs are competitive inhibitors, competing with arachidonic acid for binding in the active site.

COX isozyme selectivity is defined most commonly using the concentration of drug required to inhibit PG production by 50% in a particular assay system (inhibitory concentration). Ratios using values obtained for COX-1 50% inhibitory concentrations compared with COX-2 50% inhibitory concentrations can be calculated and used as a standard measure for comparing the degrees of selectivity of a particular NSAID for one or the other COX isoform [6]. PG assay systems can vary widely, however, making it difficult to compare directly results from studies using different assay systems. To circumvent such problems, most clinicians have accepted the use of the in vitro whole-blood assay to compare NSAID selectivities. In this system, COX-1 inhibition is assessed as a function of the reduction of thromboxane made by platelets after clot formation. Inhibition of COX-2 is based on the inhibition of PGE₂ production in a heparinized blood sample after lipopolysaccharide stimulation. A COX-2-selective NSAID lacks an inhibitory effect on platelet COX-1 at concentrations at or above those that maximally inhibit COX-2 [11, 12].

Traditional NSAIDs, such as meloxicam, nimesulide, etodolac, and diclofenac show some selectivity for inhibiting COX-2 over COX-1. After the discovery of COX-2, efforts to further enhance COX-2 selectivity led to the development of celecoxib, valdecoxib, rofecoxib, etori-coxib and lumiracoxib. Most COX-2-selective NSAIDs are diaryl compounds containing a sulfonamide (celecoxib, valdecoxib) or a methylsulfone (rofecoxib, etoricoxib) rather than a carboxyl group, while lumiracoxib is an analog of diclofenac and the only acidic COX-2-selective NSAID. Lumiracoxib is available in only a few countries worldwide. Valdecoxib and rofecoxib are no longer available in any country because of concerns for excess cardiovascular adverse effects. Etoricoxib is approved in the European Union but not in the United States, while celecoxib is available worldwide. Celecoxib and etoricoxib are weak time-independent inhibitors of COX-1, but strong time-dependent inhibitors of COX-2 that require entry into and stabilized binding in the catalytic pocket. Because these drugs lack a carboxyl group, arginine 120 is not involved, but multiple sites of hydrogen and hydrophobic binding stabilize drugs at the catalytic site. The sulfur-containing phenyl ring of COX-2-selective NSAIDs plays a pivotal role in binding stability by occupying the hydrophobic side pocket characteristic of the COX-2 catalytic site. If this side pocket is removed by mutagenesis, all isozyme selectivity is lost [13].

NSAIDs are produced in a variety of dosage forms, including intravenous, slow-release and sustained-release oral preparations, and topical preparations in various forms including gels and patches, and suppositories. Given the desire to reduce NSAID toxicity while preserving drug delivery to a specific site, efforts continue to alter drug formulation and delivery systems. Nanoparti-cles, liposomes, and microspheres are under investigation to allow dose reduction and specific targeting. Intra-articular delivery is under consideration, but because joints have very efficient lymphatic clearance systems the utility of this form of targeting remains to be proved.

Topical NSAID formulations were developed to reduce systemic exposure while preserving efficacy. Several factors - including the drug, formulation, and site of application - are important for efficacy [14]. Diclofenac, for example, is available as a solution, gel, or patch. The systemic effects are directly proportional to the surface area, and this method of delivery results in a relatively stable systemic diclofenac level compared with oral administration [15]. A recent Chochrane review concludes that topical NSAIDs can provide good levels of pain relief and that gastrointestinal adverse events are reduced compared with oral NSAIDs [16].

NSAIDs have also been combined with agents having gastroprotective effects into polypills that are currently available on the market. This strategy may increase compliance with effective protective agents, thereby reducing adverse effects in clinical practice. Combining diclofenac with the synthetic PGE₁ analog misoprostol (arthrotec) is shown to reduce risk of NSAID-related peptic ulcerations and mucosal injury, but utility of the combination is often limited by misoprostol-induced cramping and diarrhea [17]. In population-based studies, arthrotec was more effective than diclofenac and misoprotsol co-prescription in preventing hospitalization for peptic ulcer disease or gastrointestinal hemorrhage [18]. Several poly-pills containing NSAIDs and proton-pump inhibitors are approved for use in rheumatic diseases including ketoprofen with omeprazole (axorid) [19]. The combination of enteric-coated naproxen and the proton pump inhibitor esomeprazole (vimovo) was shown to reduce endoscopically detected gastric ulcers [20]. The combination of ibuprofen and the H₂-blocker famotidine (duexis) was also shown to reduce endoscopically detected gastric and duodenal ulcers [21].

A different strategy is nitric oxide-releasing NSAIDs, which are synthesized by the ester linkage of a nitric oxide-releasing moiety to conventional NSAIDs including aspirin, flurbiprofen, diclofenac, sulindac, and others [22]. The nitric oxide moiety is slowly released by enzymatic activity in vivo, probably by esterases, resulting in slow accumulation of the parent NSAID. The lower rate of gastrointestinal ulceration associated with these drugs is probably related to nitric oxide-associated vasodilation and the relatively lower concentration of parent NSAID.

NSAIDs are frequently used as first-line agents for the symptomatic relief of many different inflammatory conditions. In double-blind, randomized clinical trials of inflammatory arthritis, NSAIDs have been compared with placebo, aspirin, and each other. Clinical trials of NSAID efficacy in RA and OA most often employ a design whereby the current NSAID is discontinued and the patient must have an increase in symptoms or flare to enter the study. Although there is some variation in primary outcome measures, most include parameters that make up the American College of Rheumatology-20. Efficacy superior to that of placebo is easily demonstrated for NSAIDs within 1 to 2 weeks in patients with active RA who are not receiving corticosteroids or other anti-inflammatory medications [23]. Comparisons of adequate doses of traditional NSAIDs or COX-2-selective NSAIDs with one another almost always show comparable efficacy. Despite improvement in pain and stiffness with NSAIDs, these agents do not usually reduce acute-phase reactants, nor do they modify radiographic progression. The anti-inflammatory effects of NSAIDs have also been demonstrated in OA, rheumatic fever, juvenile rheumatoid arthritis, ankylosing spondylitis, gout, and systemic lupus erythematosus. Although not as rigorously proven, their efficacy is also accepted in treatment of reactive arthritis, psoriatic arthritis, acute and chronic bursitis, and tendonitis.

Virtually all NSAIDs relieve pain when used in doses substantially lower than those required to suppress inflammation. The analgesic action of NSAIDs is due to inhibition of PG production in peripheral tissues and in the central nervous system. In the periphery, PGs do not induce pain per se, but sensitize peripheral nociceptors to the effects of mediators such as bradykinin or histamine [24]. PGs released during inflammation or other trauma lower the activation threshold of tetrodotoxin-resistant sodium channels on sensory neurons. In the central nervous system, where NSAIDs and acetaminophen exert analgesic effects, PGs also play an important role in neuronal sensitization. COX-2 is constitutively expressed in the dorsal horn of the spinal cord, and its expression is increased during inflammation [25]. Centrally generated PGE₂ activates spinal neurons and also microglia that contribute to neuropathic pain [26]. Both COX-1 and COX-2 play a role in nociception, as demonstrated by reductions of experimental pain in mice deficient in either COX-1 or COX-2 [27].

Because of the widespread use of prescription and non-prescription NSAIDs, there are ample opportunities for interaction with other drugs and for interactions with patient-specific factors [47]. Specific drug interactions are listed on the package inserts of individual agents.

RA and other diseases (for example, hepatic and renal disease) that decrease serum albumin concentrations are associated with increased concentrations of free NSAIDs. Hepatic and renal diseases may also impair drug metabolism or excretion, and thereby increase the toxicity of a given dose of NSAID to an individual patient. Renal insufficiency may be accompanied by accumulated endogenous organic acids that may displace NSAIDs from protein binding sites.

Aging is accompanied by changes in physiology, resulting in altered pharmacokinetics and pharmacodynamics. Decreased drug clearance may be the consequence of reductions in hepatic mass, enzymatic activity, blood flow, renal plasma flow, glomerular filtration rate, and tubular function associated with aging. Older people are more likely to experience adverse gastrointestinal and renal effects related to NSAIDs. The increased risk of cardiovascular disease in older patients raises concerns of accelerated myocardial infarction or stroke. The use of aspirin for prevention of cardiovascular disease increases the toxicity of NSAIDs and, conversely, the concomitant use of NSAIDs may increase aspirin resistance. Use of proton pump inhibitors for gastroprotection may interfere with the efficacy of antiplatelet agents such as clopidogrel [42]. Older people have more illnesses than younger patients and therefore take more medications, increasing the possibility of drug-drug interactions. Older patients may also be more likely to self-medicate or make errors in drug dosing. For these reasons, frequent monitoring for compliance and toxicity should be a part of the use of NSAIDs in this population.

In choosing an NSAID for a particular patient, the clinician must consider efficacy, potential toxicity related to concomitant drugs and patient factors, and cost [1]. A study to assess patient preferences for treatment-related benefits and risks of NSAIDs for OA suggested that reductions in ambulatory pain and difficulty doing daily activities were the most important benefits. The risk of myocardial infarction and stroke were the most important risk outcomes. However, patients were willing to accept a small increased risk of myocardial infarction to reduce ambulatory, but not resting, pain [51]. Furthermore, patient preference for factors such as the dosing regimen may be taken into account. In addition to choices from the perspective of the individual patient and physician, it may be important to take a broader view. Choice of anti-inflammatory analgesic therapy can also be considered from the perspective of healthcare institutions and payers. The symptoms and conditions for which NSAIDs are used are extraordinarily common. Consequently, the cost of NSAIDs as a proportion of total drug costs can be high when drugs are expensive. The increased cost of branded NSAIDs has an important pharmacoeconomic impact. On the other hand, adverse events can have important economic consequences, and improved safety may be cost-effective.

An absence of anti-inflammatory activity reduces the effectiveness of acetaminophen for diseases accompanied by a significant component of infl ammation (for example, RA, gout). However, acetaminophen is a safe and effective alternative for milder pain conditions, including OA. With respect to patient preference, a survey study demonstrated that only 14% of a large group of rheumatic disease patients (n = 1,799) with RA, OA, or fibromyalgia preferred acetaminophen over NSAIDs, while 60% preferred NSAIDs [52]. In a head-to-head clinical trial of acetaminophen versus diclofenac plus misoprostol, there was significantly greater improvement in pain scores for patients in the diclofenac group. This finding was magnified in those patients with more severe disease at baseline [53].

Acetaminophen should be tried as the initial therapy in patients with mild to moderate pain for reasons of safety and cost. However, if patients have moderate to severe symptoms or if evidence of inflammation is present, moving to treatment with NSAIDs may provide more rapid and effective relief [54].