Journal of Investigative Medicine:
Variability in the Response to Cyclooxygenase Inhibitors: Toward the Individualization of Nonsteroidal Anti-Inflammatory Drug Therapy
Grosser, Tilo MD
From the Institute for Translational Medicine and Therapeutics, University of Pennsylvania, Philadelphia, PA.
Received March 30, 2009, and in revised form May 27, 2009.
Accepted for publication May 27, 2009.
Reprints: Tilo Grosser, MD, Institute for Translational Medicine and Therapeutics, University of Pennsylvania, 421 Curie Blvd, Philadelphia, PA 19104. E-mail: firstname.lastname@example.org.
Supported by a National Scientist Development Grant from the American Heart Association (0430148N). This symposium was supported in part by a grant from the National Center for Research Resources (R13 RR023236).
No potential conflicts of interest to disclose.
Nonsteroidal anti-inflammatory drugs (NSAIDs) relieve pain, inflammation, and fever by inhibiting cyclooxygenases (COXs). Nonsteroidal anti-inflammatory drugs selective for COX-2 were developed to inhibit the major enzymatic source of the prostaglandins that mediate pain and inflammation while sparing COX-1-derived prostaglandins that contribute dominantly to gastric cytoprotection. Indeed, such purpose-designed COX-2 inhibitors reduced the incidence of serious gastrointestinal adverse effects when compared with traditional NSAIDs; however, they confer a small but absolute cardiovascular hazard. The hazard might also extend to traditional NSAIDs, which are relatively selective for COX-2, such as diclofenac, meloxicam, and etodolac. The occurrence of complications and the therapeutic responses to individual NSAIDs may vary substantially from patient to patient. Exploitation of detectable variability in the biochemical response to NSAIDs may offer an approach to the personalization of the management of risk and benefit.
Nonsteroidal anti-inflammatory drugs (NSAIDs), which inhibit the cyclooxygenases (COXs), are among the most frequently used prescription and over-the-counter drugs. They alleviate symptoms in chronic inflammatory conditions such as arthritides, which cause intermittent or chronic pain in almost 50 million US adults annually and are the nation's leading cause of disability. Despite their efficacy in the relief of pain and inflammation, some patients on NSAIDs experience serious adverse events-primarily gastrointestinal and cardiovascular complications. Cyclooxygenase 2 (COX-2)-selective NSAIDs have been developed to reduce the incidence of serious gastrointestinal events but are more likely to cause serious cardiovascular events than nonselective drugs.1 It seems probable that the timely identification of the small minority of patients at risk for emerging complications would increase the overall safety of NSAID treatment. Similarly, marked interindividual variability in the therapeutic response even to structurally similar NSAIDs has long been recognized,2 although potential mechanisms have not been thoroughly studied. Thus, any NSAID can be expected to fail relieving symptoms in some patients, and again, it seems probable that early identification of these individuals would allow sparing them from unnecessary drug exposure and enhance the overall therapeutic efficacy of NSAID treatment. However, identifying individuals who would be expected to attain the highest benefit from a compound within the class and those at risk for serious adverse events, such as myocardial infarction and stroke, is a major challenge.
Currently, the selection of an NSAID is typically guided by pharmacological characteristics of a compound such as onset and duration of action (eg, a long-acting NSAID for long-term treatment) and whether risk factors for gastrointestinal and cardiovascular complications can be identified (eg, a COX-2-selective NSAID for a patient at risk for gastroduodenal ulceration but not for a patient with preexisting cardiovascular disease). Sometimes, but probably not often enough, the initial dose is adjusted, for example, for older patients. And finally, if the selected drug fails to relief symptoms within a few weeks, the dose is increased or an alternative compound is prescribed. We will have to gain a much better understanding of the factors that condition the therapeutic and adverse responses to NSAIDs if we want to move toward more systematic evidence-based paradigms of personalized therapy. This will require the prediction of a patient's response to the drug-likely based on biomarkers of drug response and biomarkers of risk for adverse events.3
The general feasibility of a therapeutic approach based on the predicted drug response has been demonstrated for chemotherapy regimen with substances such as imatinib (Gleevec)4 and trastuzumab (Herceptin).5 Genotype-guided dosing of warfarin might emerge as the first example of the application of genetic information to prescribing decisions of long-term consumed medications.6 However, the personalization of NSAID therapy can be expected to be an even more complex endeavor1 given the softer outcome measures of therapeutic efficacy, the low incidence of complications, and the large number of compounds within the class.
Two major questions need to be addressed before such NSAID personalization paradigms can be explored in controlled trials: (i) Are there mechanism-based differences in efficacy and hazard between distinct types of NSAIDs; for example, are there mechanism-based differences between compounds that inhibit both COX enzymes and COX-2-selective drugs? Large clinical trials comparing purpose-designed (pd) COX-2 inhibitors with traditional (t) NSAIDs were typically designed to demonstrate noninferiority rather than potential differences in the analgesic response; thus, it is currently unknown whether pdCOX-2 inhibitors or tNSAIDs or none of both offer a therapeutic advantage. By contrast, such comparative trials have indeed detected significant differences in complication rates.7-11 (ii) Are there reproducible differences between individuals in their pharmacological response to an NSAID and do these differences relate to tractable genetic or nongenetic host factors? These may include genetic variation in the pharmacokinetics or pharmacodynamics of a compound or preexisting conditions and risk factors such as hypertension, which would inform selection of a compound and a dose. Interestingly, the emergence of the cardiovascular hazard on COX-2-selective NSAIDs has sparked unprecedented basic, translational, and clinical research into the mechanisms of an adverse drug effect, cardiovascular complications,1 and has en-passant revealed determinants of drug action, which might be usefully applied to a more informed individualization of NSAID therapy. This review will summarize some of these discoveries and their potential relevance for a personalization of NSAID therapy.
Nonsteroidal anti-inflammatory drugs relieve pain, inflammation, and fever by inhibiting the formation of prostanoid mediators. The 2 evolutionary conserved12 COX isozymes, COX-1 and COX-2, are key enzymes in the formation of the prostanoids, namely, prostacyclin (PGI2), thromboxane (TxA2) prostaglandin (PG) E2, PGD2, and PGF2α, from arachidonic acid.13 Involving a 2-step reaction, the enzymes incorporate 2 oxygen molecules into arachidonic acid (the COX reaction) and reduce the product to the corresponding alcohol (the hydroperoxidase reaction) to catalyze the synthesis of an unstable endoperoxide precursor, PGH2. PGH2 is subject to further metabolism by PG-synthases that catalyze the formation of the active prostanoids. At least 9 such terminal synthases form the 5 biologically active arachidonic acid products,14 which act through 1 or more G-protein-coupled receptors. All components of the arachidonic acid biosynthetic response pathway underlie tissue-specific regulation, so that a diverse array of signaling events is affected by the inhibition of the COX enzymes.
After the discovery of the second COX isoform in the 1990s, COX-2 emerged as a drug target because of the limited gastrointestinal tolerability of long-term NSAID use. Cyclooxygenase 1 was proposed to be the sole isoform involved in protection of the gastroduodenal mucosa (its inhibition was thought to be responsible for gastrointestinal adverse events on tNSAIDs), whereas COX-2 was proposed to be exclusively responsible for inflammation and pain. Screens of combinatorial libraries allowed the identification of compounds with higher affinity for COX-2 than COX-1, namely, the pdCOX-2 inhibitors. Subsequent crystallography revealed a difference in the tertiary structures, that is, a deeper hydrophobic side pocket in the substrate binding channel of COX-2 than that of COX-1, which explained the differences in selectivity retrospectively.15,16 Thus, COX-2-selective NSAIDs turned out to be molecules with side chains that fit within this hydrophobic pocket but are too large to block COX-1 with equally high affinity. Whereas most tNSAIDs have 1-ring or 2-ring structures, pdCOX-2 inhibitors are typically derived from a 3-ring pharmacophore with a central heterocyclic. Several such compounds have been adelecoxib (Celebrex; Pfizer) is marketed in the United States. Rofecoxib (Vioxx; Merck) and valdecoxib (Bextra; Pfizer) were withdrawn worldwide when their cardiovascular risk was detected in randomized controlled trials.7-9,17 Etoricoxib (Arcoxia; Merck), lumiracoxib (Prexige; Novartis), and an injectable prodrug of valdecoxib, parecoxib (Dynastat; Pfizer), are approved in some countries but failed to attain Food and Drug Administration approval so far.
Celecoxib, rofecoxib, and valdecoxib were approved based on a surrogate gastrointestinal outcome, that is, reduced rates of endoscopically visualized ulcerations, in comparison to equiefficacious doses of a tNSAID, despite the possibility that COX-2 might play a role in ulcer healing. This possibility emerged when low levels of COX-2 in healthy mucosa were detected in addition to COX-1 expression18 and when up-regulation of COX-2 was observed during acute stages of gastric erosion and ulceration.19-21 Subsequently, 4 year-long7,10,22 and 1 short-term23 outcome studies have studied whether the pdCOX-2 inhibitors reduce the incidence of serious gastrointestinal complications, namely, bleeding, perforation, and obstruction. The Vioxx Gastrointestinal Outcomes Research study and the Therapeutic Arthritis Research and Gastrointestinal Event Trial showed a reduction of serious gastrointestinal adverse events on rofecoxib (Vioxx; Merck) and lumiracoxib (Prexige; Novartis) in comparison to non-isoform-selective tNSAIDs naproxen and ibuprofen.7,10 By contrast, the long-term trial studying the only COX-2 inhibitor remaining on the market, the Celecoxib Long-term Arthritis Safety Study,24 found no differences between this compound and the comparator tNSAIDs ibuprofen and diclofenac. Only the interim data of Celecoxib Long-term Arthritis Safety Study at 6 months of treatment22 and a more recent 3-month trial23 supported a favorable gastrointestinal toxicity profile of celecoxib. The largest study, a 34,000-patient trial of etoricoxib, failed to detect a difference in the primary end point, that is, complicated gastrointestinal events, in comparison to diclofenac; only the incidence of a secondary end point, that is, symptomatic ulcers, was reduced in the etoricoxib group.11 Thus, pdCOX-2 inhibitors reduce the risk of gastrointestinal complications but bare a clinically significant residual risk. Ironically, the development of the pdCOX-2 inhibitors was pursued at a time when the incidence of gastrointestinal complications on tNSAIDs had been falling for several years, probably attributable to the prescription of lower doses and the emergence of concomitant gastroprotective therapy.25 Thus, the overall gain in gastrointestinal safety by pdCOX-2 inhibitors was markedly less than expected and offset by the increase in cardiovascular risk.
Despite their improved gastrointestinal safety, pdCOX-2 inhibitors increase the risks of myocardial infarction, stroke, systemic and pulmonary hypertension, thrombosis, congestive heart failure, and sudden cardiac death compared with nonselective tNSAIDs.1 Seven randomized controlled trials, involving 3 structurally distinct compounds, namely, rofecoxib, celecoxib, and valdecoxib, provide convincing evidence for the cardiovascular hazard of COX-2 inhibition.1 Suppression of cardioprotective COX-2-derived prostanoids, primarily PGI2 and perhaps PGE2, affords a plausible mechanism for the increase in cardiovascular events on these drugs.1,26 Prostacyclin I2 acts as a general restraint on platelet activation and limits vascular proliferation, remodeling, atherogenesis, and hypertension in rodent models.27-30 Perturbation of these effects of PGI2 in the vasculature would be expected to augment the cardiovascular risk for humans. Indeed, a nonsynonymous single nuclear polymorphism in the fifth intracellular loop of the PGI2 receptor (IP) which disrupts signaling, was associated with increased cardiovascular risk in a recent study.26 Interestingly, mice deficient in PGI2 signaling were not prone to spontaneous thrombosis31; the thrombotic process had to be induced by endothelial damage but, once initiated, proceeded more vigorously than in mice with intact PGI2 function.28 This suggests that, primarily, patients with preexisting risk factors for thrombotic events, such as atherosclerotic vessel wall lesions or vascular inflammation, would be at risk for cardiovascular complications by COX-2 inhibition, a hypothesis that is supported by a recent epidemiological study.32
The effects of PGE2 on the cardiovascular system are more complex because, in contrast to PGI2, which signals through just 1 receptor (the IP), PGE2 has 4 specific receptors coupled to various downstream signaling events. It is also synthesized not just by 1 terminal isomerase, like PGI2 (PGI2 synthase), but is formed by distinct PGE synthase (PGES) isoforms, the cytosolic (c) and the microsomal (m) PGES isoforms. Microsomal PGES-1 is colocalized in some settings with COX-2, and both enzymes are subject to regulation during development33 and in inflammation. Deletion or inhibition of this enzyme markedly reduces inflammatory responses in several mouse models. Deletion of mPGES-1 retards atherogenesis in fat-fed hyperlipidemic mice34 but does not accelerate the response to a thrombogenic stimulus in vivo, in contrast to either selective inhibition of COX-2 or deletion of the IP.28 It remains to be seen whether mPGES-1 inhibition can reverse established atherosclerosis in addition to delaying its progression. Microsomal PGES-1 deletion also markedly attenuates the formation of abdominal aortic aneurysms that develop in hyperlipidemic mice infused with angiotensin II, an impact that coincides with a significant reduction in the attendant oxidative stress.35 Interestingly, in addition to the expected depression of PGE2 production, deletion of mPGES-1 in mice augmented biosynthesis of PGI2, presumably through rediversion of the intermediate COX product PGH2 to PGI2. It is currently unclear, however, whether elevated PGI2 contributes to the cardiovascular profile of mPGES-1 inhibition or whether this is solely the result of reduced PGE2 formation or a combination of both. Although the relative roles of PGI2 and PGE2 (and perhaps other COX products) in the development of cardiovascular complications remain to be elucidated, the combined observations in murine models of disease, randomized controlled trials, and the genetic association study suggest that all COX-2-selective compounds augment cardiovascular risk through the unifying mechanism,1 that is, depression of COX-2-derived protective prostanoids.
DIFFERENCES BETWEEN NSAIDs
Are there mechanism-based differences between NSAIDs that may affect their risk-benefit profile? Nonsteroidal anti-inflammatory drugs are a chemically heterogeneous group of compounds; thus, it is unsurprising that individual compounds are distinguished by multiple characteristics with potential impact on their risk-benefit profile of which 2 duration of action and selectivity for COX-2, will be discussed here. The duration of action of any drug is conditioned by its concentration kinetics at the site of action (eg, in an inflamed joint) and by pharmacodynamic properties such as the chemistry of enzyme inhibition. For example, aspirin is the only irreversible COX inhibitor, and thus, its duration of action is largely determined by the formation rate of new enzyme. Aspirin acetylates serine529 in COX-1 or serine516 in COX-2 covalently to inactivate the enzyme. Its plasma half-life is in the range of 15 to 20 minutes and has very little influence on its duration of action. Some tissues recover their pool of COX protein relatively quickly after acetylation, such as the endothelium, whereas others, such as the anucleate platelets, have very limited capacity, if any, to recover enzyme function by de novo protein synthesis. New platelets have been formed to restore function.36 Thus, aspirin's risk-benefit ratio is, at least in part, determined by this tissue-specific heterogeneity in drug action. For example, low doses of aspirin have been shown to be less gastrotoxic than high doses while retaining their full cardioprotective effect, presumably because expression of functional COX protein in the gastrointestinal mucosa, but not in platelets, recovers within the dosing interval. Indeed, aspirin is the only COX inhibitor with proven cardioprotective activity.36
Nonsteroidal anti-inflammatory drugs act both peripherally at the site of inflammation and centrally in the spinal chord. Many compounds are organic acids with relatively low pKa values, which may facilitate penetration of active drug into inflamed tissue, where the pH is lower. The same mechanism, however, may limit penetration into the central compartment. The combination of drug activities at both sites is thought to result in the net efficacy profile of a compound. An extreme example is acetaminophen, which has a very weak anti-inflammatory activity but is a useful analgesic. Acetaminophen is an isoform-nonspecific COX inhibitor that inactivates the COXs through a mechanism distinct from other NSAIDs. Thus, its ability to inhibit the COX enzymes is thought to be conditioned by the peroxide tone of the immediate environment.37 Because sites of inflammation usually contain increased concentrations of leukocyte-generated peroxides, COX inhibition by acetaminophen in inflamed tissue is limited. However, it has been suggested that COX inhibition might be disproportionately pronounced in the central nervous system, explaining its good analgesic (and antipyretic) effect.37
The plasma half-life of a compound is an important determinant of drug action because this compartment precedes distribution into the synovial fluid and into the central nervous system. Plasma half-life times vary considerably among diverse NSAIDs. For example, ibuprofen, indomethacin, acetaminophen, and diclofenac are characterized by relatively rapid elimination (t1/2, 1-4 hours), whereas, on the other extreme, piroxicam has a t1/2 of 46 to 58 hours at a steady state that can increase to up to 75 hours in the elderly.38 Interestingly, concentrations of NSAIDs, which are short lived in plasma, are often sustained in synovial fluid.39 This may contribute to their prolonged clinical effect, especially when administered at intervals longer than their plasma half-life time. Lumiracoxib affords an example for the potential relevance of the distribution into synovial fluid. As mentioned above, lumiracoxib has never been approved in the United States, but marketing authorization was granted in several countries worldwide. Lumiracoxib is a unique pdCOX-2 inhibitor because it is derived from the phenyl acetic acid structure of diclofenac rather than from the 3-ring heterocyclic pharmacophore that is the basis of the other compounds within this group.40 It differs from diclofenac only by an additional methyl group and 1 fluorine substitution for chlorine and is thus, like diclofenac, a weak acid. Its plasma half-life is short, that is, approximately 6 hours. The acidic nature allows it to penetrate well into areas of inflammation. Thus, the half-life in synovial fluid is considerably longer than in plasma,41 and the concentration of lumiracoxib in synovial fluid 24 hours after dosing would be expected to result in substantial COX-2 inhibition. This may explain why once-daily dosing may afford sufficient symptom relief for some patients with arthritis despite its short plasma half-life. Despite this potentially advantageous feature, lumiracoxib doses were selected for approval, which result in plasma concentrations exceeding by far those required for maximal inhibition of COX-2 throughout the dosing interval.42 Regulatory authorities in several countries have since withdrawn lumiracoxib because of its association with sporadic reports of serious hepatic toxicity, including lethal liver failure. In the Therapeutic Arthritis Research and Gastrointestinal Event Trial, the frequency of hepatic transaminase elevation (>3-fold) was 2.6% for lumiracoxib versus 0.6% for the comparator tNSAIDs.10 There is some indication that risk for liver toxicity relates to dose, although this has not been conclusively addressed and the molecular mechanism remains to be elucidated.42 However, if toxicity indeed related to dose, lower dosing might have bypassed that risk while retaining the clinical efficacy in most patients because of sufficiently high synovial concentrations throughout the dosing interval.42 Indeed, the selection of the lowest efficacious dose would likely also reduce the risk for cardiovascular complications.43
The interaction of all NSAIDs (including both tNSAIDs and pdCOX-2 inhibitors) with COX-1 and COX-2 is defined by their molecular structure, and there is no absolute selectivity for one or the other isoform. For example, a pdCOX-2-selective compound will also inhibit COX-1 at high-enough concentrations. Ideally, such high concentrations should not be achieved in vivo, but we have observed that a single therapeutic dose of 200 mg of celecoxib inhibits COX-1 to a small degree, on average, approximately 6%, in healthy individuals.3 Thus, selectivity, a compounds' relative affinity to COX-1 versus COX-2, is a discreet variable, and all NSAIDs can be ranked on the selectivity scale.43 In human studies, the degree of COX-2 selectivity is commonly determined by whole-blood assays that measure COX-1 and COX-2 inhibition ex vivo.44 Significant differences exist in the relative degree of selectivity within the group of pdCOX-2 inhibitors, which can be approximately described as lumiracoxib = etoricoxib > valdecoxib = rofecoxib ≫ celecoxib.45 It is less well known that some of the older tNSAIDs, including diclofenac, meloxicam, and etodolac, are surprisingly similar in their degree of COX-2 selectivity to celecoxib.45,46 By contrast, substances such as naproxen and ibuprofen are slightly more potent inhibitors of COX-1 than of COX-2.45 Indeed, observational studies and a comparison of diclofenac, the most commonly prescribed NSAID in Europe, with etoricoxib, a pdCOX-2 inhibitor, in a randomized controlled trial indicate that diclofenac behaves somewhat like a selective COX-2 inhibitor in vivo.11,47 Etoricoxib is approved in some countries but not in the United States. The Multinational Etoricoxib and Diclofenac Arthritis Long-term program, by far the largest randomized controlled trial program to assess safety of a pdCOX-2 inhibitor, found comparable rates of myocardial infarction and stroke between etoricoxib and diclofenac and comparable rates of ulcer complications.11 Differences between the 2 drugs were only detected in the weaker, more frequently occurring exploratory end points, such as discontinuation due to hypertension and uncomplicated ulcers. The rate of this gastrointestinal end point was lower in the etoricoxib group, whereas hypertension occurred more frequently on etoricoxib. These observations are consistent with the notion that diclofenac is relatively selective for COX-2, yet, on average, not as selective as etoricoxib.
Interestingly, half-life and relative degree of COX-2 selectivity may interact with each other. Again, diclofenac affords an example. As mentioned, its degree of COX-2 selectivity is similar to that of celecoxib.45 However, it has a very short half-life (1-2 hours)38 and is dosed highly, like lumiracoxib, to produce the drug concentrations thought necessary for effective analgesia throughout the entire dosing interval. Early in the dosing interval, diclofenac plasma concentrations exceed those necessary to inhibit COX-2 greatly and inhibit COX-1 coincidently; diclofenac is one of the most potent COX inhibitors. Because plasma concentrations fall, the preference for COX-2 becomes apparent,48 and diclofenac continues to inhibit COX-2 completely, whereas its effect on COX-1 subsides (Fig. 1). Such discordant offset rates of COX isoform inhibition by diclofenac in vivo might result in a "window of cardiovascular hazard."1 The degree of residual COX-1 inhibition would be expected to be relevant because there is increasing evidence that inhibition of COX-1 products may reduce the likelihood of hypotension and atherothrombosis.1,13 The effect on platelets, such as, depression of COX-1-derived TxA2 formation, is unlikely to contribute significantly to such a potential protective effect of temporary COX-1 inhibition by diclofenac because only complete inhibition of platelet COX-1 (>95%) results in the inhibition of platelet function.49 However, experiments in mice indicate that reduction of COX-1 product formation protects from blood pressure increases by COX-2 inhibition28; clearly, elevation of blood pressure by COX-2 inhibition may accelerate atherogenesis and increase cardiovascular risk. These opposing effects of COX-1 and COX-2 inhibition on blood pressure are supported by a meta-analysis of approximately 45,000 patients in 19 clinical trials, which suggests that selective inhibition of COX-2 elevates blood pressure more than nonselective inhibition does.50 As mentioned, the rate of hypertension in the Multinational Etoricoxib and Diclofenac Arthritis Long-term trial was augmented in the group receiving the highly selective COX-2 inhibitor etoricoxib in comparison to the less COX-2-selective comparator diclofenac.11 A second protective effect of the inhibition of COX-1 products may relate to the progression of atherosclerosis; suppression of COX-1-derived TxA2 activity retards atherogenesis in mice.29,51,52 Thus, tNSAIDs, which inhibit COX-1 and COX-2 nonselectively, such as naproxen or ibuprofen, can be expected to bare a markedly lower cardiovascular risk, if any, than COX-2-selective NSAIDs. Unlike diclofenac, naproxen, a non-isoform-selective tNSAID with a long biological half-life, provides platelet inhibition throughout the dosing interval in some individuals, which may explain its moderate cardioprotective effect.53
Given the variable half lives and the variable degrees of COX-2 selectivity across distinct NSAIDs (Fig. 2), one would expect a marked heterogeneity in the risk-benefit profiles within the class of NSAIDs. Understanding this heterogeneity better seems to be an important first step in the personalization of NSAID therapy.
VARIATION BETWEEN INDIVIDUALS
Are there reproducible differences between individuals in their pharmacological response to an NSAID and do these differences relate to tractable genetic or nongenetic host factors? Nonsteroidal anti-inflammatory drug treatment paradigms, such as those of most drug therapies, have been configured on large average signals of efficacy and risk observed in clinical studies.54 However, it has also long been recognized that the therapeutic response to any NSAIDs may vary substantially from patient to patient, which typically results in a trial-and-error approach to the selection of the compound providing the best pain relief for individual patients. Interestingly, the variability in the analgesic response to NSAIDs has not been systematically studied. Observational data indicate that rofecoxib was frequently more efficacious in patients with arthritis than a tNSAID,55,56 and numerous case reports surfaced after rofecoxib and valdecoxib were removed from the market, suggesting that these drugs may have provided a particular quality of relief from symptoms for some patients. However, whether this is actually true has never been addressed systematically.
The occurrence of complications may also vary substantially from patient to patient. Although several risk factors for gastrointestinal complications of NSAIDs have been identified in epidemiological studies (eg, age, previous ulcers, Helicobacter infection, comedications) and are used to personalize certain aspects of NSAID treatment such as the prescription of gastroprotective therapy or the selection of a COX-2-selective NSAID,54 much less is known about risk factors for fluid retention and hypertension on NSAIDs. These tend to occur in patients who have predisposing conditions, such as preexisting hypertension, heart disease, and diabetes, or who have effective circulating volume depletion, such as severe dehydration. However, the weakness of the association between these conditions and renal complications on NSAIDs allows for not much more than an increased vigilance for the detection of such complications. Interestingly, the adverse effect discovered last is the best studied one: the risk of myocardial infarction and stroke on COX-2-selective NSAIDs.1 However, cardiovascular adverse events occurred only in a minority of patients, probably in ∼1% to 2%, who were apparently at low initial risk of cardiovascular disease, suggesting that selective inhibition of COX-2 caused a transformation of risk during extended treatment.1 Thus, one would indeed expect that genetic, pharmacokinetic, biochemical and/or physiological responses can be detected that identify patients at emerging risk.
In addition to genetic variability, more or less stable environmental factors or host factors, and their interactions with genetic factors would be expected to contribute to an individual's response to a compound. For example, small variations in plasma protein concentration or binding capacity may significantly increase free concentrations of highly protein-bound NSAIDs. Thus, age-dependent differences in plasma protein binding capacity and oxidative liver metabolism are relevant to NSAID action. Diurnal rhythms modulate the disposition of many NSAIDs. Some produce higher plasma levels, some with an increase in half-life, when taken in the early morning compared with other times of the day.57 The targeted pathological processes, such as inflammation and pain,58 and tissues involved in adverse reactions, such as the vasculature,59 are also subject to circadian variation. Thus, the likelihood of favorable and unfavorable drug responses may vary over time, and such sources of variability can mask clinically relevant associations.
Therefore, it is necessary to assess the relative importance of stable interindividual sources of variability in NSAID response, such as genetic variants, compared with less predictable environmental sources of variability. We have recently examined the variability, both within and between healthy subjects, in response to equipotent doses of celecoxib and rofecoxib in a placebo-controlled crossover study.3 We quantified plasma drug concentrations and pharmacodynamic responses as measured by both ex vivo and in vivo indices of COX-1 and COX-2 inhibition. Significant intraindividual variability was evident in all parameters and may have, in part, derived from circadian variability. Circadian variation in COX activity has not been described, but as mentioned, diurnal variation in the pharmacokinetics of NSAIDs,57 including celecoxib,60 is well recognized. Despite substantial intraindividual variability in response, this was exceeded by the variation in the subject population as a whole. Approximately a third of the variability was attributable to differences between individuals, suggesting the contribution of stable sources of variance including genetic variability.3 Interestingly, almost half of the study population had genetic variants in the target enzymes, some with allelic frequencies of 5% or more. Although the study was not sized to afford a comprehensive analysis of genetic variants, 2 of the single-nucleotide polymorphisms tested showed evidence of allelic association with elements of drug response. CYP2C9*2 is a variant associated with reduced activity of a major metabolizing enzyme of celecoxib61 and is associated with elevated plasma concentrations of the drug. A proline17 leucine substitution in COX-1 seemed to be associated with a failure of inhibition of thromboxane formation with both drugs, and interindividual variability of inhibition of COX-1 may be attributable in part to this variant.3 Indeed, this may represent a factor that influences the actual selectivity for inhibition of COX-2 attained within an individual. Thus, selectivity for COX-2 achieved in humans is not a purely structural property of a compound but may also be influenced by pharmacokinetic (eg, plasma concentration) and pharmacodynamic factors (eg, genetic variations of the target enzymes). It seems quite possible that these sources of variability might be exploited for the development of approaches to the personalization of NSAID therapy.
Several NSAIDs with selectivity for COX-2 have been withdrawn or were unapprovable because of their cardiovascular risk profile, and the pharmacoepidemiology of tNSAIDs points to a heterogeneous risk profile within this group. This has generated uncertainty as to the most appropriate use of NSAIDs by individual patients. Enzymes and receptors downstream of the COXs are currently being explored as potential drug targets, which might afford a safer and more effective anti-inflammatory pain therapy.62 Exploitation of detectable variability in response to existing drugs may offer an alternative approach and may also facilitate development of new drugs within the pathway. However, the personalization of NSAID therapy will require the precise understanding of differences between NSAIDs in pharmacokinetics and pharmacodynamics and of the genetic and nongenetic factors that contribute to a patient's response to a given drug.
1. 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(1):4-15.
2. Day RO, Graham GG, Williams KM, et al. Variability in response to NSAIDs. Fact or fiction? Drugs. 1988;36(6):643-651.
3. Fries S, Grosser T, Price TS, et al. Marked interindividual variability in the response to selective inhibitors of cyclooxygenase-2. Gastroenterology. 2006;130(1):55-64.
4. Gorre ME, Mohammed M, Ellwood K, et al. Clinical resistance to STI-571 cancer therapy caused by BCR-ABL gene mutation or amplification. Science. 2001;293(5531):876-880.
5. Berns K, Horlings HM, Hennessy BT, et al. A functional genetic approach identifies the PI3K pathway as a major determinant of trastuzumab resistance in breast cancer. Cancer Cell. 2007;12(4):395-402.
6. Schwarz UI, Ritchie MD, Bradford Y, et al. Genetic determinants of response to warfarin during initial anticoagulation. N Engl J Med. 2008;358(10):999-1008.
7. Bombardier C, Laine L, Reicin A, et al. Comparison of upper gastrointestinal toxicity of rofecoxib and naproxen in patients with rheumatoid arthritis. VIGOR study group. N Engl J Med. 2000;343(21):1520-1528, 1522 p following 1528.
8. Ott E, Nussmeier NA, Duke PC, et al. Efficacy and safety of the cyclooxygenase 2 inhibitors parecoxib and valdecoxib in patients undergoing coronary artery bypass surgery. J Thorac Cardiovasc Surg. 2003;125(6):1481-1492.
9. Nussmeier NA, Whelton AA, Brown MT, et al. Complications of the COX-2 inhibitors parecoxib and valdecoxib after cardiac surgery. N Engl J Med. 2005;352(11):1081-1091.
10. Schnitzer TJ, Burmester GR, Mysler E, et al. Comparison of lumiracoxib with naproxen and ibuprofen in the Therapeutic Arthritis Research and Gastrointestinal Event Trial (TARGET), reduction in ulcer complications: randomised controlled trial. Lancet. 2004;364(9435):665-674.
11. Cannon CP, Curtis SP, FitzGerald GA, et al. Cardiovascular outcomes with etoricoxib and diclofenac in patients with osteoarthritis and rheumatoid arthritis in the Multinational Etoricoxib and Diclofenac Arthritis Long-term (MEDAL) programme: a randomised comparison. Lancet. 2006;368(9549):1771-1781.
12. Grosser T, Yusuff S, Cheskis E, et al. Developmental expression of functional cyclooxygenases in zebrafish. Proc Natl Acad Sci U S A. 2002;99(12):8418-8423.
13. Smyth EM, Grosser T, Wang M, et al. Prostanoids in health and disease. J Lipid Res. 2009;50Suppl:S423-S428.
14. Narumiya S, FitzGerald GA. Genetic and pharmacological analysis of prostanoid receptor function. J Clin Invest. 2001;108(1):25-30.
15. Garavito RM, DeWitt DL. The cyclooxygenase isoforms: structural insights into the conversion of arachidonic acid to prostaglandins. Biochim Biophys Acta. 1999;1441(2-3):278-287.
16. Picot D, Loll PJ, Garavito RM. The x-ray crystal structure of the membrane protein prostaglandin H2 synthase-1. Nature. 1994;367(6460):243-249.
17. Bresalier RS, Sandler RS, Quan H, et al. Cardiovascular events associated with rofecoxib in a colorectal adenoma chemoprevention trial. N Engl J Med. 2005;352(11):1092-1102.
18. Zimmermann KC, Sarbia M, Schror K, et al. Constitutive cyclooxygenase-2 expression in healthy human and rabbit gastric mucosa. Mol Pharmacol. 1998;54(3):536-540.
19. Mizuno H, Sakamoto C, Matsuda K, et al. Induction of cyclooxygenase-2 in gastric mucosal lesions and its inhibition by the specific antagonist delays healing in mice. Gastroenterology. 1997;112(2):387-397.
20. Shigeta J, Takahashi S, Okabe S. Role of cyclooxygenase-2 in the healing of gastric ulcers in rats. J Pharmacol Exp Ther. 1998;286(3):1383-1390.
21. Lipsky PE, Brooks P, Crofford LJ, et al. Unresolved issues in the role of cyclooxygenase-2 in normal physiologic processes and disease. Arch Intern Med. 2000;160(7):913-920.
22. Silverstein FE, Faich G, Goldstein JL, et al. Gastrointestinal toxicity with celecoxib vs nonsteroidal anti- inflammatory drugs for osteoarthritis and rheumatoid arthritis: the CLASS study: a randomized controlled trial. Celecoxib Long-term Arthritis Safety Study. JAMA. 2000;284(10):1247-1255.
23. Singh G, Fort JG, Goldstein JL, et al. Celecoxib versus naproxen and diclofenac in osteoarthritis patients: SUCCESS-I study. Am J Med. 2006;119(3):255-266.
24. Witter J. Celebrex Capsules (Celecoxib). NDA 20-998/S-009. Medical Officer Review. Silver Spring, MD: US Department of Health and Human Services Food and Drug Administration; 2000.
25. Fries JF, Murtagh KN, Bennett M, et al. The rise and decline of nonsteroidal antiinflammatory drug-associated gastropathy in rheumatoid arthritis. Arthritis Rheum. 2004;50(8):2433-2440.
26. Arehart E, Stitham J, Asselbergs FW, et al. Acceleration of cardiovascular disease by a dysfunctional prostacyclin receptor mutation: potential implications for cyclooxygenase-2 inhibition. Circ Res. 2008;102(8):986-993.
27. Cheng Y, Austin SC, Rocca B, et al. Role of prostacyclin in the cardiovascular response to thromboxane A2. Science. 2002;296(5567):539-541.
28. Cheng Y, Wang M, Yu Y, et al. Cyclooxygenases, microsomal prostaglandin E synthase-1, and cardiovascular function. J Clin Invest. 2006;116(5):1391-1399.
29. Kobayashi T, Tahara Y, Matsumoto M, et al. Roles of thromboxane A2 and prostacyclin in the development of atherosclerosis in apoE-deficient mice. J Clin Invest. 2004;114(6):784-794.
30. Egan KM, Lawson JA, Fries S, et al. COX-2-derived prostacyclin confers atheroprotection on female mice. Science. 2004;306(5703):1954-1957.
31. Murata T, Ushikubi F, Matsuoka T, et al. Altered pain perception and inflammatory response in mice lacking prostacyclin receptor. Nature. 1997;388(6643):678-682.
32. Solomon DH, Glynn RJ, Rothman KJ, et al. Subgroup analyses to determine cardiovascular risk associated with nonsteroidal antiinflammatory drugs and coxibs in specific patient groups. Arthritis Rheum. 2008;59(8):1097-1104.
33. Pini B, Grosser T, Lawson JA, et al. Prostaglandin E synthases in zebrafish. Arterioscler Thromb Vasc Biol. 2005;25(2):315-320.
34. Wang M, Zukas AM, Hui Y, et al. Deletion of microsomal prostaglandin E synthase-1 augments prostacyclin and retards atherogenesis. Proc Natl Acad Sci U S A. 2006;103(39):14507-14512.
35. Wang M, Lee E, Song W, et al. Microsomal prostaglandin E synthase-1 deletion suppresses oxidative stress and angiotensin II-induced abdominal aortic aneurysm formation. Circulation. 2008;117(10):1302-1309.
36. Patrono C, Garcia Rodriguez LA, Landolfi R, et al. Low-dose aspirin for the prevention of atherothrombosis. N Engl J Med. 2005;353(22):2373-2383.
37. Boutaud O, Aronoff DM, Richardson JH, et al. Determinants of the cellular specificity of acetaminophen as an inhibitor of prostaglandin H2 synthases. Proc Natl Acad Sci U S A. 2002;99(10):7130-7135.
38. Verbeeck RK, Blackburn JL, Loewen GR. Clinical pharmacokinetics of non-steroidal anti-inflammatory drugs. Clin Pharmacokinet. 1983;8(4):297-331.
39. Day RO, McLachlan AJ, Graham GG, et al. Pharmacokinetics of nonsteroidal anti-inflammatory drugs in synovial fluid. Clin Pharmacokinet. 1999;36(3):191-210.
40. Patrignani P. Lumiracoxib: a viewpoint by Paola Patrignani. Drugs. 2004;64(19):2247-2248.
41. Scott G, Rordorf C, Reynolds C, et al. Pharmacokinetics of lumiracoxib in plasma and synovial fluid. Clin Pharmacokinet. 2004;43(7):467-478.
42. Hinz B, Renner B, Cheremina O, et al. Lumiracoxib inhibits cyclo-oxygenase 2 completely at the 50 mg dose: is liver toxicity avoidable by adequate dosing? Ann Rheum Dis. 2009;68(2):289-291.
43. Grosser T. The pharmacology of selective inhibition of COX-2. Thromb Haemost. 2006;96(4):393-400.
44. Patrignani P, Panara MR, Greco A, et al. Biochemical and pharmacological characterization of the cyclooxygenase activity of human blood prostaglandin endoperoxide synthases. J Pharmacol Exp Ther. 1994;271(3):1705-1712.
45. FitzGerald GA, Patrono C. The coxibs, selective inhibitors of cyclooxygenase-2. N Engl J Med. 2001;345(6):433-442.
46. Warner TD, Giuliano F, Vojnovic I, et al. Nonsteroid drug selectivities for cyclo-oxygenase-1 rather than cyclo-oxygenase-2 are associated with human gastrointestinal toxicity: a full in vitro analysis. Proc Natl Acad Sci U S A. 1999;96(13):7563-7568.
47. Garcia Rodriguez LA, Gonzàlez-Pérez A. Long-term use of non-steroidal anti-inflammatory drugs and the risk of myocardial infarction in the general population. BMC. 2005;3:17.
48. Schwartz JI, Dallob AL, Larson PJ, et al. Comparative inhibitory activity of etoricoxib, celecoxib, and diclofenac on COX-2 versus COX-1 in healthy subjects. J Clin Pharmacol. 2008;48(6):745-754.
49. Reilly IA, FitzGerald GA. Inhibition of thromboxane formation in vivo and ex vivo: implications for therapy with platelet inhibitory drugs. Blood. 1987;69(1):180-186.
50. Aw TJ, Haas SJ, Liew D, et al. Meta-analysis of cyclooxygenase-2 inhibitors and their effects on blood pressure. Arch Intern Med. 2005;165(5):490-496.
51. Burleigh ME, Babaev VR, Oates JA, et al. Cyclooxygenase-2 promotes early atherosclerotic lesion formation in LDL receptor-deficient mice. Circulation. 2002;105(15):1816-1823.
52. Egan KM, Wang M, Lucitt MB, et al. Cyclooxygenases, thromboxane, and atherosclerosis: plaque destabilization by cyclooxygenase-2 inhibition combined with thromboxane receptor antagonism. Circulation. 2005;111(3):334-342.
53. Capone ML, Sciulli MG, Tacconelli S, et al. Pharmacodynamic interaction of naproxen with low-dose aspirin in healthy subjects. J Am Coll Cardiol. 2005;45(8):1295-1301.
54. Wilcox CM, Allison J, Benzuly K, et al. Consensus development conference on the use of nonsteroidal anti-inflammatory agents, including cyclooxygenase-2 enzyme inhibitors and aspirin. Clin Gastroenterol Hepatol. 2006;4(9):1082-1089.
55. Arboleya LR, de la Figuera E, Soledad Garcia M, et al. Experience of rofecoxib in patients with osteoarthritis previously treated with traditional non-steroidal anti-inflammatory drugs in Spain: results of phase 2 of the VICOXX study. Curr Med Res Opin. 2003;19(4):288-297.
56. Zhao SZ, Wentworth C, Burke TA, et al. Drug switching patterns among patients with rheumatoid arthritis and osteoarthritis using COX-2 specific inhibitors and non-specific NSAIDs. Pharmacoepidemiol Drug Saf. 2004;13(5):277-287.
57. Markiewicz A, Semenowicz K. Time dependent changes in the pharmacokinetics of aspirin. Int J Clin Pharmacol Biopharm. 1979;17(10):409-411.
58. Labrecque G, Dore F, Belanger PM. Circadian variation of carrageenan-paw edema in the rat. Life Sci. 1981;28(12):1337-1343.
59. Rudic RD, McNamara P, Reilly D, et al. Bioinformatic analysis of circadian gene oscillation in mouse aorta. Circulation. 2005;112(17):2716-2724.
60. FDA. New Drug Application 20-998. Clinical Pharmacology and Biopharmaceutics Review Celecoxib. Bethesda, MD: Food and Drug Administration; 1998.
61. Tang C, Shou M, Mei Q, et al. Major role of human liver microsomal cytochrome P450 2C9 (CYP2C9) in the oxidative metabolism of celecoxib, a novel cyclooxygenase-II inhibitor. J Pharmacol Exp Ther. 2000;293(2):453-459.
62. Wang M, Song WL, Cheng Y, et al. Microsomal prostaglandin E synthase-1 inhibition in cardiovascular inflammatory disease. J Intern Med. 2008;263(5):500-505.
This article has been cited 2 time(s).
Acta OncologicaNon-steroidal anti-inflammatory treatment in cancer cachexia: A systematic literature reviewActa Oncologica
Restorative Neurology and NeuroscienceEffects of a novel herbal formulation JSK on acute spinal cord injury in ratsRestorative Neurology and Neuroscience
NSAIDs; COX-2; personalized medicine
© 2009 American Federation for Medical Research
Highlight selected keywords in the article text.