Aspirin has been on the market for 115 years. Beginning

Transcription

1 Contemporary Reviews in Cardiovascular Medicine Nonsteroidal Anti-Inflammatory Drugs and the Heart Carlo Patrono, MD; Colin Baigent, BM, BCh Aspirin has been on the market for 115 years. Beginning with the marketing of indomethacin for the treatment of rheumatoid arthritis in 1963, at least 20 other nonsteroidal anti-inflammatory drugs (NSAIDs) with aspirin-like actions have been developed over the past 50 years, 1 culminating with the introduction of a new class of selective inhibitors of cyclooxygenase (COX)-2, the coxibs, approximately 15 years ago. 2 The NSAIDs represent the single most crowded family of drugs sharing the same therapeutic activities and mechanism of action, perhaps reflecting the unmet therapeutic need in the area of pain management and the large interindividual variability in response to these agents. NSAIDs provide symptomatic relief of pain and inflammation associated with a variety of human disorders, including the rheumatic diseases. Their shared therapeutic actions (ie, analgesic, anti-inflammatory, and antipyretic) are usually accompanied by mechanismbased adverse effects that can, at least in part, be attenuated as a function of individual pharmacokinetic or pharmacodynamic properties. 1 Nuances in the tolerability of different NSAIDs had been described in the precoxib era on the basis of clinical trials of a few hundred patients treated for up to a few months with soft end points. More recently, however, important differences in safety have been demonstrated in head-to-head randomized comparisons of individual coxibs and 1 or more traditional NSAIDs that involved tens of thousands of patients treated for up to a few years with hard end points. Even more significantly, the interpretation of the effects of NSAIDs has been greatly enhanced by the availability, for the first time, of large, placebo-controlled trials that aimed to assess the potential for chemoprevention of colorectal cancer by rofecoxib or celecoxib. As a result of these developments, the whole field of NSAIDs has been illuminated during the past 15 years. Although epidemiological studies had previously associated regular use of NSAIDs with some aspects of vascular toxicity such as enhanced risk of congestive heart failure, 3 a novel aspect of cardiotoxicity associated with COX-2 inhibition emerged from the coxib trials, that is, increased risk of atherothrombotic vascular events. 4,5 This was largely unexpected and paradoxical, given that the prototypic COX-2 inhibitor, aspirin, had been clearly shown to be cardioprotective over a wide range of doses up to 1500 mg daily. 6 The aim of this article is to review the evidence and to discuss the mechanisms underlying the cardiovascular effects of traditional NSAIDs and coxibs, trying to reconcile the pharmacology of COX isozyme inhibition with its clinical readouts. In doing so, we shall rely primarily on the results of the recently completed meta-analysis of individual participant data from randomized trials 7 because it provides the most reliable information on the effects of the most widely prescribed NSAIDs on serious clinical outcomes. Arachidonic Acid Metabolism in Human Platelets and Endothelium Stimulation of platelet aggregation by various agonists leads to the activation of membrane phospholipases, with the consequent release of arachidonic acid and its cyclooxygenation to form the prostaglandin (PG) endoperoxides PGG 2 and PGH 2. PGH 2 is further metabolized by thromboxane (TX) synthase (the most abundant PGH 2 isomerase in human platelets) to form TXA 2 as the prevailing platelet prostanoid. 1 The COX isoform involved in this rapid chain of enzymatic reactions is COX-1 in mature human platelets, with transient COX-2 expression being restricted to newly formed platelets. 8 Prostanoids are usually not required to produce a functional response but rather to modulate its intensity and duration. Thus, human platelets can aggregate and form a hemostatic plug in response to thrombin even in the absence of TXA 2. However, TXA 2 amplifies the activation signal by virtue of its being synthesized and released in response to a platelet agonist and in turn inducing platelet aggregation and further TXA 2 production through its interaction with the platelet thromboxane receptor. 9 Thus, COX-1 dependent generation of TXA 2 initiates an amplification loop that propagates the activation signal to adjacent platelets by inducing further platelet activation and TXA 2 formation. 1,9 The rate of TXA 2 synthesis and release can increase a few thousand-fold from the resting state and is driven largely by enhanced substrate availability. 9 Indeed, the maximal biosynthetic capacity of human platelets to produce TXA 2 when challenged by thrombin in vitro ( ng/ml of whole blood in 1 hour) exceeds the basal rate of TXA 2 biosynthesis in vivo (0.1 ng kg 1 min 1 ) by 3 orders of magnitude. 10 The remarkable ability of platelets to increase their production of TXA 2 by virtue of an explosive chain reaction may explain, at least in part, the unusual requirement for virtually complete (ie, >97%) suppression of platelet COX-1 activity for pharmacological inhibition to translate into functional impairment 11,12 (Figure 1). From the Department of Pharmacology, Catholic University School of Medicine, Rome, Italy (C.P.); and Clinical Trial Service Unit and Epidemiological Studies Unit, Nuffield Department of Population Health, University of Oxford, Oxford, UK (C.B.). Correspondence to Carlo Patrono, MD, Istituto di Farmacologia, Università Cattolica del Sacro Cuore, Largo F. Vito 1, Rome, Italy. carlo.patrono@rm.unicatt.it (Circulation. 2014;129: ) 2014 American Heart Association, Inc. Circulation is available at DOI: /CIRCULATIONAHA

2 908 Circulation February 25, 2014 Figure 1. Maximal biosynthetic capacity of human platelets (left), rate of thromboxane production in healthy subjects (middle), and nonlinear relationship between inhibition of platelet cyclooxygenase activity and thromboxane biosynthesis in vivo (right). Left, Thromboxane production stimulated by endogenous thrombin during wholeblood clotting at 37 C (based on data from Reference 13). Middle, The metabolic fate of thromboxane A 2 (TXA 2 ) in vivo and the calculated rate of its production in healthy subjects based on thromboxane B 2 (TXB 2 ) infusions and measurement of its major urinary metabolite (data from Reference 10). Right, The nonlinear relationship between inhibition of serum TXB 2 measured ex vivo and the reduction in thromboxane metabolite (TXM) excretion measured in vivo (data adapted from Reference 12). Prostacyclin (PGI 2 ) is the major product of COXcatalyzed metabolism of arachidonic acid in human macrovascular endothelium. 1 Cultured human endothelial cells synthesize PGI 2 ( 20 ng/10 6 cells) when stimulated with arachidonate in vitro. 14 PGI 2 is a potent inhibitor of platelet aggregation and a vasodilator. 15 Its basal rate of secretion into the systemic circulation is as low as that of TXA 2, 0.1 ng kg 1 min 1 in healthy subjects, 16 resulting in blood levels (a few picograms per milliliter) that are at least 10-fold lower than the minimal concentration that inhibits platelet function. However, the in vivo biosynthesis of PGI 2, as reflected by PGI 2 metabolite excretion, is enhanced in patients with severe atherosclerosis and platelet activation, 17 suggesting that it functions as a homeostatic response to accelerated platelet vessel wall interactions. 15 The pathophysiological importance of PGI 2 has undergone ups and downs during the past 40 years. In 1979, Moncada and Vane 15 proposed that vascular homeostasis is determined by a balance between the platelet production of TXA 2 and the endothelial production of PGI 2 and that the occurrence of thrombotic disorders might be influenced by factors that alter this balance. Lack of specific inhibitors of PGI 2 synthesis or action in the preknockout era eventually led to a less enthusiastic view of its role in vascular disorders, 18 The hypothesis put forward by Moncada and Vane 15 had to wait another 20 years or so before being properly tested by the large, postmarketing trials of coxibs and by the use of mice genetically deficient in the PGI 2 receptor. 19 The latter approach led FitzGerald s group to demonstrate for the first time that PGI 2 modulates platelet-vascular interactions in vivo and specifically limits the cardiovascular response to TXA Urinary PGI 2 metabolite excretion is reduced by 60% to 80% by therapeutic doses of selective COX-2 inhibitors, implying that PGI 2 biosynthesis is largely a COX-2 driven process in humans. The hypothesis that COX-2 is normally expressed by endothelial cells in response to steady laminar shear stress 24 is consistent with the finding that deletion of COX-2 in the mouse vasculature reduces the urinary excretion of PGI 2 metabolite. 25,26 Compared with platelets, human endothelial cells have a considerably lower biosynthetic capacity to generate prostanoids. Although the maximal biosynthetic capacity of endothelial cells cannot be assessed ex vivo, one can use the lipopolysaccharide-induced expression of COX-2 in circulating monocytes 27 to assess the influence of systemic plasma levels of COX-2 inhibitors on the maximal prostanoid biosynthetic capacity of blood cells exposed to the same inhibitor concentration as vascular endothelial cells. Paired measurements of the inhibition of monocyte COX-2 activity, as assessed ex vivo, and the reduction in urinary PGI 2 metabolite excretion, as assessed in vivo, have established a linear relationship between the two, 28 in contrast to the strikingly nonlinear relationship between the inhibition of platelet COX-1 activity and the reduction in urinary thromboxane metabolite excretion 11,12 (Figure 1). The markedly different requirements for inhibition of TXA 2 - versus PGI 2 -dependent functions may have important implications for the interpretation of the cardiovascular effects of different classes of COX inhibitors, as discussed below. Platelet and vascular responses to a PGI 2 analog and time to thrombotic carotid arterial occlusion were modulated by PGI 2 receptor deletion in a gene/dose-dependent fashion, 29 consistent with the concept of a linear relationship between inhibition of PGI 2 -evoked signaling and reduction in PGI 2 -mediated functional effects. Mechanism of Action of NSAIDs The best-characterized mechanism of action of NSAIDs is inhibition of the COX activity of PGH synthase-1 and -2 (also referred to as COX-1 and COX-2; Figure 2). 1 Given the role that prostanoids such as PGE 2, PGI 2, and TXA 2 play in the local modulation of many important cellular functions, this mechanism of action is probably sufficient to explain the clinical effects of NSAIDs. 1 The administration of PGE 2 and PGI 2 causes erythema, an increase in local blood flow, and, in concert with other inflammatory mediators (eg, bradykinin), hyperalgesia and enhanced vascular permeability. 1 Moreover, PGE 2 interacting

3 Patrono and Baigent COX-2 Inhibitors and the Heart 909 Figure 2. Mechanism of formation and action of prostanoids. Arachidonic acid, a 20-carbon fatty acid containing 4 double bonds, is liberated from the sn2 position in membrane phospholipids by phospholipases, which are activated by diverse stimuli. Arachidonic acid is converted by cytosolic prostaglandin (PG) G/H synthases, which have both cyclooxygenase (COX) and hydroperoxidase activity, to the unstable intermediate PGH 2. The synthases are colloquially called cyclooxygenases and exist in 2 forms: COX-1 and COX-2. PGH 2 is converted by tissue-specific isomerases to multiple prostanoids. These bioactive lipids activate specific cell membrane receptors of the superfamily of G protein coupled receptors. cpla 2 indicates cytosolic phospholipase A 2 ; spla 2, secretory phospholipase A 2 ; EPs, PGE 2 receptors; IP, PGI 2 receptor; TP, TXA 2 receptor; and TX, thromboxane. with its EP 3 receptor can produce fever. Thus, prostanoids reproduce the main signs and symptoms of the inflammatory response. 1 Because of the redundancy of mediators of this response, it is not surprising that NSAIDs exert only a moderate anti-inflammatory effect, are effective only against pain of low to moderate intensity, reduce fever, but do not interfere with the physiological control of body temperature. 1 The production of prostanoids involved in these responses appears to be triggered by the immediate availability of constitutively expressed COX-1 and to be amplified and sustained by the local induction of COX-2 in response to inflammatory and mitogenic stimuli. 30 Although the analgesic, anti- inflammatory, and antipyretic actions of traditional NSAIDs are closely reproduced by coxibs, this finding does not exclude a potential role for COX-1 in mediating, at least in some individuals, the PG-dependent contribution to pain and inflammation. 2 The ability of acetaminophen to inhibit COX-1 and COX-2 is importantly conditioned by the peroxide tone of the local environment. 31 This may explain, at least in part, the clinical observation that, while sharing the analgesic and antipyretic effects of NSAIDs, acetaminophen has relatively poor anti-inflammatory activity at conventional doses. 1 High concentrations of leukocyte-derived peroxides accumulate at sites of inflammation and may impair the ability of acetaminophen to inhibit COX However, circulating plasma levels of the drug after the administration of 1000 mg inhibit systemically COX-2 activity to a degree comparable to that of traditional NSAIDs and coxibs. 32 Clinical Pharmacology of COX Isozyme Inhibition Low-dose aspirin provides a paradigm of COX isozyme selective and cell-specific inhibition in vivo by virtue of its short half-life and ability to inactivate COX irreversibly. 33,34 The relative COX-1 selectivity of low-dose aspirin in vivo arises from both pharmacokinetic determinants such as the acetylation of platelet COX-1 that occurs in portal blood before first-pass metabolism 35 and pharmacodynamic determinants such as the cumulative nature of platelet COX-1 inactivation on repeated daily dosing. 13,36 Thus, the daily administration of 30 mg aspirin results in virtually complete suppression of platelet TXA 2 production after 1 week of dosing through a cumulative process of inactivation of platelet COX-1 by successive daily doses of aspirin. 13,36 The unique features of the antiplatelet effect of aspirin explain the highly predictable, virtually complete inhibition of platelet TXA 2 biosynthesis in subjects who take the drug regularly, 12 the persistence of this effect over the 24-hour dosing interval in the vast majority of subjects with an 8- to 10-day platelet life span, 12 and the clinical efficacy of doses as low as 30 to 50 mg daily in high-risk patients. 34 Traditional NSAIDs lack these unique pharmacokinetic/ pharmacodynamic features and do not usually achieve the same degree of persistent platelet COX-1 inhibition as is obtained with low-dose aspirin. 34,36 Moreover, several traditional NSAIDs (eg, ibuprofen and diclofenac) are characterized by very short half-lives (1 2 hours), contributing to the transient nature of platelet COX-1 inhibition. 34,36 The COX isozyme selectivity of a particular inhibitor is critically dependent on its concentration. One can compare the selectivity profiles of different COX-2 inhibitors by plotting the drug concentrations required to inhibit the activity of human platelet COX-1 and those required to inhibit human monocyte COX-2 by 50% (IC 50 ), as measured by widely used whole-blood assays of COX activity 27 (Figure 3). This type of analysis established 2 important facts: COX-2 selectivity is a continuous variable that does not justify a dichotomous definition of selective and nonselective inhibitors, and there Figure 3. Cyclooxygenase (COX)-2 selectivity as a continuous variable. Concentrations of various COX-2 inhibitors to inhibit the activity of platelet COX-1 and monocyte COX-2 by 50% (IC 50 ) are plotted on the abscissa and ordinate scales, respectively. The solid line describes equipotent inhibition of both COX-1 and COX-2. Symbols to the left of this line denote greater inhibition of COX-1 than COX-2. Symbols to the right of this line indicate progressively greater inhibition of COX-2 than COX-1, that is, increasing degrees of COX-2 selectivity. 6-MNA indicates 6-methoxy-2-naphthylacetic acid, the active metabolite of nabumetone (data adapted from Reference 2).

4 910 Circulation February 25, 2014 is an appreciable overlap in COX-2 selectivity between some first-generation coxibs (eg, celecoxib) and some traditional NSAIDs (eg, nimesulide and diclofenac). 2 Moreover, there is large variability among NSAID-treated patients in the plasma concentrations of the COX-2 inhibitor after oral administration of a standard therapeutic dose (pharmacokinetic variability) and in the extent of inhibition of each COX isozyme, corresponding to any given concentration of the inhibitor (pharmacodynamic variability). Therefore, one can pragmatically characterize different levels of COX-2 selectivity in terms of the probability of sparing platelet (and presumably gastrointestinal mucosa) COX-1 at therapeutic plasma levels: low (eg, acetaminophen), intermediate (eg, celecoxib, nimesulide, and diclofenac), and high (eg, rofecoxib, etoricoxib, and lumiracoxib; Figure 3). Potential variables contributing to different COX-2- dependent effects include the daily dose of the inhibitor determining the extent of COX-2 inhibition, the half-life and dosing interval of the inhibitor determining the duration of COX-2 inhibition, and the patient substrate, inasmuch as the importance of COX-2 dependent PGI 2 biosynthesis is likely to vary in different clinical settings. Although the daily dose, half-life, duration of treatment, and COX-2 selectivity are all continuous variables that could influence the cardiovascular outcomes resulting from COX-2 inhibition, the clinical readouts of such inhibition are constrained by the dichotomy of a nonlinear relationship between the inhibition of platelet COX-1 activity and the downregulation of TXA 2 -dependent platelet activation (Figure 1). Therefore, inhibiting platelet COX-1 activity by 0% to 20% (as typically achieved with highly selective COX-2 inhibitors), by 20 to 50% (as typically achieved with COX-2 inhibitors endowed with moderate COX-2 selectivity), or by 50% to 90% (as typically achieved with most traditional NSAIDs) will result in only similarly modest suppression of TXA 2 biosynthesis in vivo. 11,12 Naproxen, when taken regularly at high doses and an adequate dosing interval (ie, 500 mg twice daily), is the only traditional NSAID that has been reported to inhibit platelet COX-1 activity by >95% throughout the dosing interval and to suppress TXA 2 biosynthesis in vivo to the same extent as low-dose aspirin. 37,38 Lower doses and less frequent dosing of naproxen are not expected to reproduce this aspirin-like antiplatelet effect. 38 Some NSAIDs favoring COX-1 versus COX-2 inhibition such as ibuprofen and naproxen (Figure 3) may interfere with the antiplatelet effect of low-dose aspirin by competing with acetylsalicylic acid for a common docking site (arginine 120) within the COX-1 channel. 39 Prior occupancy of arginine 120 by an NSAID will prevent aspirin from acetylating a serine residue (Ser-529) just below the COX catalytic site of the enzyme, which forms the basis of the unique mechanism of action of aspirin, resulting in permanent inactivation of platelet COX-1. 6 Drugs favoring COX-2 versus COX-1 inhibition such as acetaminophen and diclofenac do not interfere with the antiplatelet effect of low-dose aspirin, similar to celecoxib and rofecoxib (Figure 3). 39,40 Using a current analysis of the available data, the Food and Drug Administration has issued a statement informing patients and healthcare professionals that ibuprofen can interfere with the antiplatelet effect of low-dose aspirin (81 mg/d), potentially rendering aspirin less effective when used for cardioprotection and stroke prevention ( Clinical Readouts of COX Isozyme inhibition Clinical evidence from randomized trials is consistent with the available evidence on COX isozyme specific effects on particular outcomes from laboratory studies. We now describe the relationship between the known pharmacology and evidence from randomized trials as recently summarized by the Coxib and Traditional NSAID (CNT) Collaboration. 7 Effects on Pain and Inflammation The results of phase 3 randomized trials in osteoarthritis, rheumatoid arthritis, and various clinical models of acute pain have established that the analgesic and anti-inflammatory efficacy of coxibs is similar to that of several NSAID comparators (typically diclofenac, ibuprofen, or naproxen), independently of variable COX isozyme selectivity. 1,2 This finding is consistent with the COX-2 dependence of PG-mediated pain and inflammation and similar COX-2 inhibition at comparable doses of different drugs. Effects on Atherothrombotic Outcomes Meta-analyses of randomized trials and observational studies have previously shown that coxibs and some traditional NSAIDs are associated with an increased risk of atherothrombotic events. 28,41 43 The CNT Collaboration 7 has recently completed meta-analyses of individual participant data from all available randomized trials of an NSAID versus placebo or 1 NSAID regimen versus another NSAID regimen, allowing detailed assessments of the cardiovascular effects of high-dose regimens of the 3 NSAIDs that are most widely prescribed worldwide for chronic inflammatory disorders: naproxen 500 mg twice daily, diclofenac 75 mg twice daily, and ibuprofen 800 mg 3 times daily. 7 Compared with placebo, allocation to a coxib (mainly rofecoxib 25 mg daily or celecoxib 400 mg daily) increased the risk of major vascular events by 40%, largely as a result of an increase in major coronary events, with no clear evidence of an increase in stroke risk (Figure 4). 7 There was no statistically significant heterogeneity in the increased risk of major vascular events associated with rofecoxib and celecoxib despite a 20-fold difference in COX-2 selectivity between the two (Figure 3), consistent with the COX-2 dependence of PGI 2 -mediated thromboresistance of the vessel wall. 26,29 Moreover, there was no significant difference in the incidence of vascular events between a coxib and traditional NSAIDs (with the exception of naproxen [see below]). Given the nonlinear relationship between the inhibition of platelet COX-1 activity and the inhibition of platelet activation in vivo, 11,12 it is perhaps not surprising that the cardiovascular safety profiles of coxibs and some traditional NSAIDs appear similar because they both fail to inhibit platelet activation adequately regardless of their COX-2 selectivity. Compared with placebo (using indirect comparisons 44 and direct comparisons combined), however, high-dose naproxen and other traditional NSAIDs (mostly diclofenac and ibuprofen) had contrasting effects on major vascular events (Figure 5).

5 Patrono and Baigent COX-2 Inhibitors and the Heart 911 Figure 4. Effects of coxib therapy on major vascular events (MVEs), heart failure, cause-specific mortality, and upper gastrointestinal (GI) complications. Actual numbers for participants are presented, together with the corresponding mean yearly event rate (in parentheses). Participants can contribute only once to the total of MVEs. Major vascular events include myocardial infarction (MI), stroke, or vascular death. CHD indicates coronary heart disease. Rate ratios (RRs) for all trials are indicated by squares; their 99% confidence intervals (CIs), by horizontal lines. Subtotals and their 95% CIs are represented by diamonds. Squares or diamonds to the left of the solid line indicate benefit. *Includes another 25 vs 21 MVEs among patients randomized into trials for which only tabular information was available. Reproduced from Reference 7 with permission from the publisher. Copyright 2013 Elsevier B.V. There was no evidence of atherothrombotic risk from naproxen, consistent with an aspirin-like phenotype conferred by the unique pharmacokinetic/pharmacodynamic features of a twicedaily high-dose naproxen regimen noted above. 37,38 The increased risk of coronary events associated with vascular COX-2 inhibition is mechanistically consistent with the pattern of cardioprotection associated with platelet COX-1 inhibition by low-dose aspirin in low-risk subjects, that is, a clear decrease in coronary risk. 45 These findings suggest an important role of PGI 2 and TXA 2 in modulating platelet activation and its contribution to coronary atherothrombosis. 9 A reliable assessment of the role of these prostanoids in cerebrovascular ischemic events requires a much larger number of strokes than available in the CNT database. There was no convincing evidence of a latent period before these vascular hazards emerge, consistent with the proposed mechanism of action 4 and with evidence of early hazard observed with intravenous administration of parecoxib followed by oral valdecoxib in 2 short-term trials among patients undergoing coronary bypass surgery. 46,47 The CNT analyses did not provide any evidence that the effect of coxib therapy on the risk of major vascular events is attenuated by concomitant aspirin use at baseline. 7 A mitigating effect of low-dose aspirin seems biologically plausible because COX-1 knockdown in mice attenuates the prothrombotic effect of COX-2 inhibition. 4 Moreover, the neutral coronary phenotype of high-dose naproxen 7 suggests that a concomitant aspirin-like antiplatelet effect (ie, profound and persistent inhibition of platelet COX-1 activity throughout the 12-hour dosing interval of a naproxen 500 mg twice daily regimen) 37 may mitigate or abolish the prothrombotic consequences of COX-2 inhibition. However, because there was disproportionate underrepresentation of aspirin-treated participants in the coxib trials, this analysis lacked statistical power to assess whether the vascular hazards of a coxib were modified by low-dose aspirin. 7 Unfortunately, reliable information on the cardiovascular safety of other traditional NSAIDs is lacking. The vast majority of these agents, with the notable exception of indomethacin, failed to achieve >95% inhibition of serum TXB 2 after single oral dosing. 36 Moreover, both nimesulide 48 and meloxicam 49 have been reported to inhibit platelet COX-1 activity by <50% on repeated daily dosing. Therefore, the cardiovascular effects of these high-dose agents would not be expected to differ from those of diclofenac and ibuprofen unless convincingly shown otherwise. Effects on Blood Pressure and Renal Outcomes Both traditional NSAIDs and coxibs have been associated with renal and renovascular adverse events, 1,2 which is consistent with the important role of constitutively expressed COX-2

6 912 Circulation February 25, 2014 Figure 5. Effects of naproxen and nonnaproxen nonsteroidal anti-inflammatory drugs (NSAIDs) on major vascular events. Rate ratios (RRs) for all trials are indicated by squares; their 99% confidence intervals (CIs), by horizontal lines. Subtotals and their 95% CIs are represented by diamonds. Squares or diamonds to the left of the solid line indicate benefit. RRs are for comparisons of a traditional (t)nsaid vs placebo, calculated indirectly from ratio of RRs for a coxib vs placebo and RRs for a coxib vs tnsaid, each of which is shown in the vertical columns (see References 7 and 44 for statistical methods). in sustaining the physiological production of vasodilator and natriuretic prostanoids in the kidney. 50 Several prostanoids contribute to blood pressure homeostasis, and practically all COX-2 inhibitors have been associated with increased blood pressure or impaired response to antihypertensive drugs. 1,4 Furthermore, vascular disruption of COX-2 in mice depressed the expression of endothelial nitric oxide synthase and the consequent release and function of nitric oxide. 25 Effects on Congestive Heart Failure Use of traditional NSAIDs and coxibs can lead to the development of congestive heart failure in susceptible individuals, 3 and in the CNT meta-analysis, there was a doubling of this risk, with no apparent differences in this relative increase associated with different classes of COX-2 inhibitors, consistent with this being a COX-2 dependent hazard unrelated to variable platelet inhibition. 7 This is also consistent with the COX-2 dependence of hemodynamic stability and renal function in at-risk individuals and with the role of cardiomyocyte COX-2 in cardiac rhythm and function. 51 The finding that coxibs and traditional NSAIDs caused a similar increased risk of heart failure indirectly suggests that the high-dose regimens of these agents used in the coxib trials produced comparable levels of vascular COX-2 inhibition. Effects on Gastroduodenal Outcomes Several coxibs have been shown to be associated with a lower incidence of endoscopic gastroduodenal ulcers than doses of traditional NSAIDs with similar analgesic efficacy, which is consistent with the COX-1 dependence of mucosal cytoprotection. 1,2 Similarly, in at least 2 independent, large-scale trials, the highly selective COX-2 inhibitors rofecoxib and lumiracoxib were associated with a statistically significant 50% to 60% reduction in ulcer complications (mostly bleeding) compared with ibuprofen and naproxen, 52,53 consistent with the COX-1 dependence of the platelet contribution to primary hemostasis. 1,2 In the CNT meta-analysis, all NSAID regimens increased upper gastrointestinal complications (mostly bleeding) by 2- to 4-fold. 7 It should be emphasized that any COX-2 selective inhibitor may eventually inhibit COX-1 in some patients because of interindividual variability in pharmacokinetics or pharmacodynamics, and there may be untoward COX-2 dependent effects (eg, impaired ulcer healing) that are shared by all COX-2 inhibitors regardless of selectivity. 1,2 However, coxibs reduced the risk of symptomatic ulcers or upper gastrointestinal complications compared with traditional NSAIDs. 7 The risk of these complications was twice as high for naproxen as for coxibs or diclofenac, consistent with more profound and longer-lasting inhibition of COX-1 in both the gastrointestinal mucosa and platelets associated with the former 37,38 compared with the latter. 39 In conclusion, with the notable exception of gastrointestinal toxicity, neither the clinical efficacy nor the major cardiorenal complications of COX-2 inhibitors appear to be influenced by the variable COX isozyme selectivity of these agents to any clinically detectable extent. The extent (depending on dose)

7 Patrono and Baigent COX-2 Inhibitors and the Heart 913 and duration (depending on half-life and dosing interval) of concomitant inhibition of platelet COX-1 may attenuate or amplify the risk of platelet-dependent thrombotic or hemorrhagic complications, respectively, associated with NSAIDs. 33 Balance of Benefits and Risks To illustrate the absolute effects of allocation to different NSAID regimens at particular levels of cardiovascular or gastrointestinal risk, the CNT Collaboration calculated hypothetical excess risks for major vascular events in patients at high (2% per year) or low (0.5% per year) risk of major vascular events (Figure 6, left) and for upper gastrointestinal complications in patients at moderate (0.5% per year) or low (0.2% per year) risk of such complications (Figure 6, right). 7 Figure 6 illustrates several general points. Among those at low risk of vascular disease, the predicted risks of major vascular events were small regardless of the particular regimen chosen. For high-risk individuals ( 40% of whom were taking aspirin), for every 1000 patients allocated to 1 year of treatment with a coxib regimen or high-dose diclofenac regimen, 7 or 8 more would have a major vascular event, of which 2 would be fatal. Whether naproxen represents a suitable alternative regimen depends on the absolute level of gastrointestinal risk and patient attitudes to these risks. Current Guidelines and Recommendations for Individualized NSAID Therapy The American College of Rheumatology (ACR) 2012 recommendations for the use of nonpharmacological and pharmacological therapies in osteoarthritis of the hand, hip, and knee strongly recommend that, if the patient does not have a satisfactory clinical response to full-dose (ie, up to 4000 mg/d) acetaminophen, then oral NSAIDs (or other pharmacological agents) should be used, except for persons 75 years of age, in whom the use of topical rather than oral NSAIDs is recommended. 54 If the patient has a history of a symptomatic or complicated upper gastrointestinal ulcer but has not had a bleed in the past year, the ACR Technical Expert Panel strongly recommends using either a selective COX-2 inhibitor (celecoxib only in the United States; celecoxib or etoricoxib in other countries, including the European Union) or a nonselective NSAID in combination with a proton-pump inhibitor. 54 If the above patient has had an upper gastrointestinal bleed within the past year and the practitioner still chooses to Figure 6. Annual absolute effects per 1000 of coxibs and traditional nonsteroidal anti-inflammatory drugs (tnsaids) at different baseline risks of major vascular events (MVEs) and upper gastrointestinal (GI) complications. For each category of drug (coxib, diclofenac, ibuprofen, and naproxen), the predicted annual absolute risks of MVEs (±SE) are shown (left) for patients with a predicted risk of 2.0% or 0.5% per annum of an MVE. For comparison, predicted annual absolute risks of upper gastrointestinal complications (UGICs; ±SE) are shown for patients with predicted risks of 0.5% or 0.2% per annum (right). Absolute annual risks for placeboallocated patients are assumed to be those of a hypothetical patient after all appropriate forms of prophylactic treatment (eg, antiplatelet therapy, statin therapy, protonpump inhibitors) have been instituted. Reproduced from Reference 7 with permission from the publisher. Copyright 2013 Elsevier B.V.

8 914 Circulation February 25, 2014 use an oral NSAID, the ACR Technical Expert Panel strongly recommends using a selective COX-2 inhibitor in combination with a proton-pump inhibitor. 54 In the clinical scenario in which the osteoarthritic patient is on low-dose aspirin for cardioprotection and the practitioner chooses to use an oral NSAID, the ACR Technical Expert Panel strongly recommends using a nonselective NSAID other than ibuprofen in combination with a proton-pump inhibitor to avoid a potential pharmacodynamic interaction with low-dose aspirin. 54 Although the same type of interaction with low-dose aspirin is not apparent for either diclofenac 39 or celecoxib, 40 the ACR Technical Expert Panel strongly recommends that a selective COX-2 inhibitor should not be used in the above situation. 54 No specific recommendation was made about the other individual NSAIDs. Although issued by a North American organization, the ACR recommendations are more in line with the European Medicines Agency than the Food and Drug Administration position. 55 In fact, the European Medicines Agency issued the recommendation that coxibs (and more recently diclofenac), but not other traditional NSAIDs, be considered contraindicated in patients with ischemic heart disease or stroke and that they be avoided in patients with risk factors for coronary heart disease. 55 In contrast, the Food and Drug Administration requested that a boxed warning be added to the package insert of all traditional NSAIDs and celecoxib, emphasizing the potential for increased risk of serious cardiovascular thrombotic events. 55 A scientific statement from the American Heart Association on the use of NSAIDs for the management of musculoskeletal symptoms in patients with known cardiovascular disease or risk factors for ischemic heart disease has suggested a stepped-care approach to pharmacological therapy that focuses on agents with the lowest reported risk of cardiovascular events (acetaminophen, aspirin, tramadol, narcotic analgesics, nonacetylated salicylates) and then progresses toward other agents considered to be associated with increasing levels of cardiovascular risk, that is, non COX-2 selective NSAIDs, NSAIDs with some COX-2 activity, and COX- 2 selective NSAIDs. 55 In the absence of further specification of individual agents, this terminology is potentially misleading inasmuch as all NSAIDs are COX-2 inhibitors and, as illustrated in Figure 3, COX-2 selectivity is a continuous variable. 2 Moreover, as reviewed above, there is no convincing evidence that variable COX-2 selectivity influences the cardiovascular risk associated with COX-2 inhibitors. 7 Another limitation of this American Heart Association scientific statement 55 is that the low-risk medications listed above for initial therapy, with the exception of aspirin, have not been studied in sufficiently large, placebo-controlled, randomized trials to demonstrate their alleged superior cardiovascular safety. Naproxen, without any specification of dose, is indicated as the preferred choice 55 on the basis of meta-analyses of randomized trials 41 and observational studies. 42 Although the CNT analyses reviewed above suggest that high-dose naproxen may not be associated with an increased risk of major vascular events in low-risk individuals, this result should be extrapolated with caution to high-risk patients. First, it is currently unknown whether this would be true among patients with known cardiovascular disease treated with low-dose aspirin, in whom naproxen will not result in any additional inhibition of platelet COX-1 and may actually interfere with the antiplatelet effect of low-dose aspirin. 56 Under these circumstances, the longer half-life of naproxen vis-à-vis ibuprofen or diclofenac may result in longer- lasting inhibition of COX- 2 dependent PGI 2 biosynthesis. 37 Second, the cardiovascular effects of lower naproxen doses such as those typically used in over-the-counter preparations (eg, 220 mg twice daily) are uncertain because a lower-dose regimen does not mimic the aspirin-like effect of 500 mg twice daily. 38 Third, the apparent safety advantage of high-dose naproxen regimens over other traditional NSAIDs may not be preserved after long-term use because of substantially similar untoward hemodynamic and renal effects that may increase the risk of heart failure (see above) and possibly ischemic stroke. It is reasonable to expect than an evidence-based debate on the above issues may allow the medical/scientific community to reach a consensus on treatment guidelines for NSAID therapy and eventually lead to regulatory harmonization worldwide. Conclusion High-dose NSAID regimens are associated with serious cardiovascular and gastrointestinal risks, the size of which varies according to the extent and duration of the inhibition of COX-1 and COX-2 by the regimen and, for a given regimen, is proportional to a patient s baseline risk. The choice of the most appropriate analgesic regimen for particular patients with inflammatory conditions should be determined after a careful assessment of their vascular and gastrointestinal risks and should be re-evaluated after proper management of risk factors. Although the considerations of benefit versus risk that derive from the CNT meta-analysis may well apply to many high-income countries, less widespread availability of adjunct therapies to manage modifiable risk factors for both cardiovascular and upper gastrointestinal complications in low- and middle-income countries may require a different strategy, starting with an examination of essential medicines lists. 57 In general, the current advice that NSAIDs should be taken at the lowest effective dose for the shortest possible time to minimize these risks remains reasonable, but the CNT results suggest that strict avoidance of NSAIDs among patients with chronic inflammatory disorders and vascular risk factors may not always be justified when a patient places a high value on the control of symptoms and understands the magnitude of the associated risks. Finally, it is hoped that lessons learned from the coxib failure story may help guide the successful development of a new class of safer NSAIDs, targeting mediators unrelated to arachidonic acid metabolism or drug targets downstream of COX isozymes. 5 Acknowledgments The expert editorial assistance of Daniela Basilico and Patrizia Barbi is gratefully acknowledged. Sources of Funding The authors studies were supported by grants from the European Commission (EICOSANOX Integrated Project ), the UK Medical Research Council, and the British Heart Foundation.

9 Patrono and Baigent COX-2 Inhibitors and the Heart 915 Disclosures Dr Patrono reports having received honoraria and consulting fees from Bayer AG and Merck & Co. The Clinical Trial Service Unit and Epidemiological Studies Unit of the University of Oxford has a policy of staff not accepting fees, honoraria, or paid consultancies. Dr Baigent was chief investigator of the Study of Heart and Renal Protection of ezetimibe/simvastatin funded by research grants from Merck to the University of Oxford. References 1. Grosser T, Smyth E, FitzGerald GA. Anti-inflammatory, antipyretic and analgesic agents; pharmacotherapy of gout. In: Brunton LL, Chabner BA, Knollmann BC, eds. Goodman & Gilman s The Pharmacological Basis of Therapeutics. 12th ed. New York: McGraw-Hill; 2011: FitzGerald GA, Patrono C. The coxibs, selective inhibitors of cyclooxygenase-2. N Engl J Med. 2001;345: García Rodríguez LA, Hernández-Díaz S. Nonsteroidal antiinflammatory drugs as a trigger of clinical heart failure. Epidemiology. 2003;14: Grosser T, Fries S, FitzGerald GA. Biological basis for the cardiovascular consequences of COX-2 inhibition: therapeutic challenges and opportunities. J Clin Invest. 2006;116: Patrono C, Rocca B. Nonsteroidal antiinflammatory drugs: past, present and future. Pharmacol Res. 2009;59: Patrono C, García Rodríguez LA, Landolfi R, Baigent C. Low-dose aspirin for the prevention of atherothrombosis. N Engl J Med. 2005; 353: Coxib and Traditional NSAID Trialists (CNT) Collaboration. Vascular and upper gastrointestinal effects of non-steroidal anti-inflammatory drugs: meta-analyses of individual participant data from randomised trials. Lancet. 2013;382: Rocca B, Secchiero P, Ciabattoni G, Ranelletti FO, Catani L, Guidotti L, Melloni E, Maggiano N, Zauli G, Patrono C. Cyclooxygenase-2 expression is induced during human megakaryopoiesis and characterizes newly formed platelets. Proc Natl Acad Sci U S A. 2002;99: Davì G, Patrono C. Platelet activation and atherothrombosis. N Engl J Med. 2007;357: Patrono C, Ciabattoni G, Pugliese F, Pierucci A, Blair IA, FitzGerald GA. Estimated rate of thromboxane secretion into the circulation of normal humans. J Clin Invest. 1986;77: Reilly IA, FitzGerald GA. Inhibition of thromboxane formation in vivo and ex vivo: implications for therapy with platelet inhibitory drugs. Blood. 1987;69: Santilli F, Rocca B, De Cristofaro R, Lattanzio S, Pietrangelo L, Habib A, Pettinella C, Recchiuti A, Ferrante E, Ciabattoni G, Davì G, Patrono C. Platelet cyclooxygenase inhibition by low-dose aspirin is not reflected consistently by platelet function assays: implications for aspirin resistance. J Am Coll Cardiol. 2009;53: Patrignani P, Filabozzi P, Patrono C. Selective cumulative inhibition of platelet thromboxane production by low-dose aspirin in healthy subjects. J Clin Invest. 1982;69: Jaffe EA, Weksler BB. Recovery of endothelial cell prostacyclin production after inhibition by low doses of aspirin. J Clin Invest. 1979;63: Moncada S, Vane JR. Arachidonic acid metabolites and the interactions between platelets and blood-vessel walls. N Engl J Med. 1979;300: FitzGerald GA, Brash AR, Falardeau P, Oates JA. Estimated rate of prostacyclin secretion into the circulation of normal man. J Clin Invest. 1981;68: FitzGerald GA, Smith B, Pedersen AK, Brash AR. Increased prostacyclin biosynthesis in patients with severe atherosclerosis and platelet activation. N Engl J Med. 1984;310: Majerus PW. Arachidonate metabolism in vascular disorders. J Clin Invest. 1983;72: Murata T, Ushikubi F, Matsuoka T, Hirata M, Yamasaki A, Sugimoto Y, Ichikawa A, Aze Y, Tanaka T, Yoshida N, Ueno A, Oh-ishi S, Narumiya S. Altered pain perception and inflammatory response in mice lacking prostacyclin receptor. Nature. 1997;388: Cheng Y, Austin SC, Rocca B, Koller BH, Coffman TM, Grosser T, Lawson JA, FitzGerald GA. Role of prostacyclin in the cardiovascular response to thromboxane A2. Science. 2002;296: Cullen L, Kelly L, Connor SO, Fitzgerald DJ. Selective cyclooxygenase-2 inhibition by nimesulide in man. J Pharmacol Exp Ther. 1998; 287: McAdam BF, Catella-Lawson F, Mardini IA, Kapoor S, Lawson JA, FitzGerald GA. Systemic biosynthesis of prostacyclin by cyclooxygenase (COX)-2: the human pharmacology of a selective inhibitor of COX-2. Proc Natl Acad Sci U S A. 1999;96: Catella-Lawson F, McAdam B, Morrison BW, Kapoor S, Kujubu D, Antes L, Lasseter KC, Quan H, Gertz BJ, FitzGerald GA. Effects of specific inhibition of cyclooxygenase-2 on sodium balance, hemodynamics, and vasoactive eicosanoids. J Pharmacol Exp Ther. 1999;289: Topper JN, Cai J, Falb D, Gimbrone MA Jr. Identification of vascular endothelial genes differentially responsive to fluid mechanical stimuli: cyclooxygenase-2, manganese superoxide dismutase, and endothelial cell nitric oxide synthase are selectively up-regulated by steady laminar shear stress. Proc Natl Acad Sci U S A. 1996;93: Yu Y, Ricciotti E, Scalia R, Tang SY, Grant G, Yu Z, Landesberg G, Crichton I, Wu W, Puré E, Funk CD, FitzGerald GA. Vascular COX-2 modulates blood pressure and thrombosis in mice. Sci Transl Med. 2012;4:132ra Ricciotti E, Yu Y, Grosser T, Fitzgerald GA. COX-2, the dominant source of prostacyclin. Proc Natl Acad Sci U S A. 2013;110:E Patrignani P, Panara MR, Greco A, Fusco O, Natoli C, Iacobelli S, Cipollone F, Ganci A, Créminon C, Maclouf J, Patrono C. Biochemical and pharmacological characterization of the cyclooxygenase activity of human blood prostaglandin endoperoxide synthases. J Pharmacol Exp Ther. 1994;271: García Rodríguez LA, Tacconelli S, Patrignani P. Role of dose potency in the prediction of risk of myocardial infarction associated with nonsteroidal anti-inflammatory drugs in the general population. J Am Coll Cardiol. 2008;52: Cheng Y, Wang M, Yu Y, Lawson J, Funk CD, FitzGerald GA. Cyclooxygenases, microsomal prostaglandin E synthase-1, and cardiovascular function. J Clin Invest. 2006;116: Smith WL, Langenbach R. Why there are two cyclooxygenase isozymes. J Clin Invest. 2001;107: Boutaud O, Aronoff DM, Richardson JH, Marnett LJ, Oates JA. Determinants of the cellular specificity of acetaminophen as an inhibitor of prostaglandin H2 synthases. Proc Natl Acad Sci U S A. 2002;99: Hinz B, Cheremina O, Brune K. Acetaminophen (paracetamol) is a selective cyclooxygenase-2 inhibitor in man. FASEB J. 2008;22: Patrono C, Patrignani P, García Rodríguez LA. Cyclooxygenase-selective inhibition of prostanoid formation: transducing biochemical selectivity into clinical read-outs. J Clin Invest. 2001;108: Patrono C, Baigent C, Hirsh J, Roth G; American College of Chest Physicians. Antiplatelet drugs: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines (8 th Edition). Chest. 2008;133(suppl):199S 233S. 35. Pedersen AK, FitzGerald GA. Dose-related kinetics of aspirin: presystemic acetylation of platelet cyclooxygenase. N Engl J Med. 1984;311: Patrono C, Ciabattoni G, Patrignani P, Pugliese F, Filabozzi P, Catella F, Davì G, Forni L. Clinical pharmacology of platelet cyclooxygenase inhibition. Circulation. 1985;72: Capone ML, Tacconelli S, Sciulli MG, Grana M, Ricciotti E, Minuz P, Di Gregorio P, Merciaro G, Patrono C, Patrignani P. Clinical pharmacology of platelet, monocyte, and vascular cyclooxygenase inhibition by naproxen and low-dose aspirin in healthy subjects. Circulation. 2004;109: Capone ML, Tacconelli S, Sciulli MG, Anzellotti P, Di Francesco L, Merciaro G, Di Gregorio P, Patrignani P. Human pharmacology of naproxen sodium. J Pharmacol Exp Ther. 2007;322: Catella-Lawson F, Reilly MP, Kapoor SC, Cucchiara AJ, DeMarco S, Tournier B, Vyas SN, FitzGerald GA. Cyclooxygenase inhibitors and the antiplatelet effects of aspirin. N Engl J Med. 2001;345: Renda G, Tacconelli S, Capone ML, Sacchetta D, Santarelli F, Sciulli MG, Zimarino M, Grana M, D Amelio E, Zurro M, Price TS, Patrono C, De Caterina R, Patrignani P. Celecoxib, ibuprofen, and the antiplatelet effect of aspirin in patients with osteoarthritis and ischemic heart disease. Clin Pharmacol Ther. 2006;80: Kearney PM, Baigent C, Godwin J, Halls H, Emberson JR, Patrono C. Do selective cyclo-oxygenase-2 inhibitors and traditional non-steroidal antiinflammatory drugs increase the risk of atherothrombosis? Meta-analysis of randomised trials. BMJ. 2006;332: McGettigan P, Henry D. Cardiovascular risk with non-steroidal antiinflammatory drugs: systematic review of population-based controlled observational studies. PLoS Med. 2011;8:e