• E-mail
• Print
• Comment
• Font Size
• Digg
• del.icio.us
• Discuss article

Update on Cyclooxygenase Inhibitors: Has a Third COX Isoform Entered the Fray?*

Posted on: Tuesday, 11 October 2005, 03:00 CDT

By Hersh, Elliot V; Lally, Edward T; Moore, Paul A

Key words: Acetaminophen - Aspirin - COX-2 inhibitors - COX-3 - NSAIDs - Prostaglandins

ABSTRACT

It has been more than 30 years since Sir John Vane first reported that the pharmacological actions of aspirin-like drugs could be explained by their ability to inhibit cyclooxygenase (COX). Since then, a second isoform of COX, named COX-2, has been discovered and highly selective inhibitors of this isoform have been marketed. Most recently, a splice variant of COX-1 mRNA, retaining intron 1, and given the names COX-3, COX-1b or COX-1v, has been described.

Non-selective NSAIDs such as ibuprofen and naproxen, which inhibit both COX-1 and COX-2, have proven highly effective and safe in the short-term management of acute pain. Highly selective COX-2 inhibitors including celecoxib, rofecoxib, valdecoxib, lumiracoxib, and etoricoxib were developed with the hope of significantly reducing the serious gastrointestinal toxicities associated with chronic high-dose NSAID use. While long-term studies demonstrated that rofecoxib and lumiracoxib reduced the incidence of Gl perforations, ulcerations and bleeds by approximately 60% compared to non-selective NSAIDs, recent reports also demonstrated that the chronic use of rofecoxib and celecoxib in arthritis and colorectal polyp patients, and the short-term use of parecoxib and valdecoxib in patients who had undergone coronary artery bypass surgery, resulted in a significant increase in serious cardiovascular events, including myocardial infarction and stroke compared to naproxen or placebo.

COX-3 mRNA has been isolated in many tissues including canine and human cerebral cortex, human aorta, and rodent cerebral endothelium, heart, kidney and neuronal tissues. In transfected insect cells, canine COX-3 protein is expressed and was selectively inhibited by acetaminophen. However, in humans and rodents an acetaminophen sensitive COX-3 protein is not expressed because the retention of intron-1 adds 94 and 98 nucleotides to the COX-3 mRNA structure respectively. Since the genetic code is a triplicate code (3 nucleotides to form one amino acid), the retention of the intron in both species results in a frame shift in the RNA message and the production of a truncated protein with a completely different amino acid sequence than COX-1 or COX-2 lacking acetaminophen sensitivity.

Advances made through a combination of basic molecular biological and pharmacological techniques, and well designed randomized controlled clinical trials have demonstrated that the apparent gastrointestinal advantage of selective COX-2 inhibitors appears to be outweighed by their potential for cardiovascular toxicity and that acetaminophen's analgesic and antipyretic effects do not involve the inhibition of the COX-1 splice variant protein, putative COX-3.

Introduction: a single COX model

More than 30 years ago, in a discovery that eventually lead to the awarding of a Nobel prize, Sir John Vane reported that the principal therapeutic and toxic effects of aspirin and related drugs, namely analgesia, antiinflammatory effects, fever reduction and the potential for gastrointestinal (GI) ulceration could be explained by a common mechanism: the ability of these compounds to inhibit prostaglandin synthesis'. His thought at the time was that a ubiquitous single cyclooxygenase (COX) enzyme existed and its ability to convert arachidonic acid to various prostaglandin analogues was responsible for a variety of pathological (pain, inflammation, fever) and physiological (GI cytoprotection, platelet aggregation) effects1,2 (Figure 1). His seminal work has influenced thousands of basic and clinical researchers who have studied the effects of prostaglandins and their inhibition by non-steroidal anti- inflammatory drugs (NSAIDs). His recent passing in the fall of 2004, should give us all pause to reflect on his scientific accomplishments.

While therapeutic doses of a number of the more recently developed non-steroidal anti-inflammatory drugs, including ibuprofen 400 mg3,4, ketoprofen 25 mg5,6, naproxen sodium 550mg7 and ketorolac lOmg8 have exhibited analgesic efficacy significantly greater (p < 0.05) than therapeutic doses of aspirin (650 mg) in post-surgical pain models, all of these drugs share aspirin's pharmacological profile of analgesic, antiinflammatory, antipyretic and antiplatelet effects, and the potential for causing gastrointestinal ulcerations, perforations and bleeds, especially when taken on a chronic basis2,9- 11. The fact that the antiplatelet effect of aspirin is more prolonged than other NSAIDs can be explained by aspirin's ability to irreversibly acetylate COX in the platelet, whereas the later drugs reversibly bind and block the enzyme2,12. While this single COX model explained many of the similarities between the actions of aspirin and NSAIDs, it did not account for the differences in the potential for GI toxicity between various drugs; nor did it address the lack of anti-inflammatory, antiplatelet and adverse GI effects characteristic of acetaminophen (paracetamol), another putative inhibitor of prostaglandin synthesis13,14.

The dual COX model

In the early 1990s various groups reported that the overall activity of COX was due to the actions of two distinct COX isoforms, given the names of COX-1 and COX-2(15-18). COX-1, the so called 'constitutive' isoform was reported to be present in numerous tissues including brain, kidney, platelets and the GI mucosa and appeared to be mainly involved in normal physiologic or 'housekeeping' functions such as platelet aggregation, maintaining kidney function, and GI cytoprotection, whereas COX-2 was inducible; its synthesis increased in response to tissue injury and was involved in the pathological processes of pain, inflammation and fever13,14,19,20. Conventional NSAIDs such as ibuprofen and naproxen block both COX-1 and COX-2. Thus while they possessed analgesic, anti-inflammatory and antipyretic activity by blocking COX-2, their COX-1 blocking activity could lead to GI ulcerations, perforations and bleeds, and untoward renal effects manifested by water and sodium retention in sensitive individuals especially with chronic dosing19,20.

It was hoped that highly selective COX-2 inhibitors by virtue of their COX-1 sparing effects would have a much lower incidence of these unwanted side effects19.

Large-scale GI safety studies have been published on three of the six selective COX-2 inhibitors shown in Figure 2. Rofecoxib 50mg once daily in the Vioxx Gastrointestinal Outcome Results (VIGOR) trial and lumiracoxib 400 mg once daily in the Therapeutic Arthritis Research and Gastrointestinal Events Trial (TARGET), both regimens representing supratherapeutic doses of 2-4-fold, significantly lowered the incidence of serious and potentially fatal GI events (perforations, ulcerations and bleeds) in the entire study population compared to standard doses of comparator NSAIDs21,22. In the VIGOR trial, all patients presented with rheumatoid arthritis, aspirin users were excluded, the comparator NSAID was naproxen 500 mg twice per day, the trial duration was 10.5 months, and the risk reduction was approximately 60% (p = 0.005)21. In TARGET, all patients had osteoarthritis, roughly 25% of the population were low- dose aspirin users, the comparator NSAIDs were naproxen 500 mg twice per day or ibuprofen 800 mg three times per day, the trial duration was 12 months and the risk reduction was about 65% (p < 0.0001)22. The third published large scale gastrointestinal outcomes study was the Celecoxib Long-Term Arthritis Study (CLASS)23. Like TARGET, CLASS also included low-dose aspirin users (about 21% of the patient population). Seventy-three per cent of the population had osteoarthritis and the remaining 27% had rheumatoid arthritis, a 2- 4-fold supratherapeutic dose of celecoxib 400 mg twice per day was employed, and the comparator NSAIDs were diclofenac 75 mg twice per day and ibuprofen 800 mg three times per day. While the trial lasted 13 months24, the published results only included the first 6 months of the study. In the entire study population (aspirin and non- aspirin users), a non-significant trend (p = 0.09) was observed representing a risk reduction in serious GI events of about 50%(23). In the subgroup of exclusively nonaspirin users, this difference was statistically significant (p = 0.04). However, it was later revealed that the apparent advantages of celecoxib over non-selective NSAIDs had all but disappeared by 13 months24. Largescale GI outcome studies have yet to be published for the other selective COX-2 inhibitors.

Figure 1. The single COX model as originally proposed by Vane

Figure 2. Chemical structures of highly selective COX-2 inhibitors. Rofecoxib was voluntarily removed from the worldwide marketplace on September 30, 2004. Valdecoxib was removed from the marketplace on April 7, 2005. Etoricoxib, lumiracoxib and parecoxib have not been approved in the United States. Note that parecoxib is an injectable prodrug (inactive) that is metabolized to its active metabolite valdecoxib by the cytochrome P 450 system

Figure 3. The dual COX model. Note the opposing cardiovascular effects of COX-1 and COX-2 products

To further complicate matters, the results of several studies suggest that COX-2 (in addition to COX-1) may also play a role in maintaining GI mucosal integrity, especiall\y in ulcer healing2-8. So, the use of these drugs in a patient with a recently diagnosed ulcer might not be prudent.

The hope that selective COX-2 inhibitors would possess less untoward renal effects than non-selective NSAIDs also did not come to fruition because COX-2 appears to play an important constitutive role in renal homeostasis13,29,30. Some of the proposed physiological functions of COX-1 and COX-2 are illustrated in Figure 3.

It should be stressed that selective COX-2 inhibitors generally possess analgesic and anti-inflammatory effects equal to (not superior to) that of conventional NSAIDs20-23,31-35. While their analgesic activity in acute pain models has been reported to equal or exceed that of optimal doses of peripheral narcotic combination drugs, including acetaminophen 600 mg plus codeine 60 mg or acetaminophen 1000mg plus oxycodone 10 mg36-39, this has also been reported for several conventional NSAIDs4,7,8,34,40.

The dual COX model (Figure 3) explains the differences between highly selective COX-2 inhibitors and conventional NSAIDs in their ability to inhibit platelet aggregation. For example, the cardioprotective effect of aspirin is largely explained by its ability to irreversibly block the action of COX-1 in the platelet2,41. Platelets lack a nucleus and are devoid of the COX-2 isoform. While aspirin also irreversibly acetylates COX-2 in endothelial cells, aspirin at low doses is up to 100-fold more selective for COX-1 and the endothelial cells being nucleated are able to produce a new COX enzyme, unlike the enucleated platelet41. Selective COX-2 inhibitors on the other hand, while possessing analgesic and anti-inflammatory properties do not inhibit platelet aggregation because the COX-1 enzyme is spared.

Unfortunately two recent placebo-controlled trials suggest that when these drugs are taken on a chronic basis (18 months or more), which occurred in two colorectal cancer prevention trials, highly selective COX-2 inhibitors appear to increase the risk of myocardial infarction and occlusive stroke, a potentially serious effect in patients with pre-existing cardiovascular disease42,43. Merck's voluntary removal of rofecoxib from the market place in September 2004(44), due to a doubling in the rate of myocardial infarctions and strokes compared to placebo in subjects who had taken 25 mg once per day for at least 18 months (1.5% for rofecoxib vs. 0.75% for placebo) in the APPROVe (Adenamatous Polyposis Prevention) trial42, and the 2.5-3.4-fold increase in cardiovasuclar events compared to placebo in subjects taking 200mg or 400 mg of celecoxib twice per day in the APC (Adenoma Prevention with Celecoxib) trial43, greatly dampened the enthusiasm initially expressed with the entire COX-2 selective class of drugs. The most recent PDA-request and subsequent removal of valdecoxib from the marketplace in April of 2005 due to a lack of adequate cardiovascular data on the long-term use of the drug, an increased cardiovascular risk in short-term coronary artery bypass grafting (CABG) trials with it and its intravenous prodrug parecoxib and an unusually high incidence of serious and potentially life-threatening skin reactions45-47, leaves only one of these drugs (celecoxib) available in the United States marketplace.

However, not all chronic studies demonstrate an increased cardiovascular risk with the use of highly selective COX-2 inhibitors. A 1-year placebo-controlled study of rofecoxib 25mg/day in 692 patients with Alzheimer's disease revealed no increase in cardiovascular risk with rofecoxib; a 1.1% incidence with rofecoxib and a 3.1% incidence with placebo of confirmed serious cardiovascular events48. Merck's 3-year Alzheimer's database, also showed no increase in cardiovascular risk compared to placebo24. In contrast a meta-analysis of previous rofecoxib trials of between 4 weeks and 56 weeks in duration reported an approximate doubling of myocardial infarction risk compared to placebo or NSAID controls49. While one might hypothesize that a simple solution to the potential cardiovascular problems with selective COX-2 inhibitors would be to place all at-risk patients (or possibly all COX-2 patients) on concomitant low dose aspirin therapy, results from 3 separate studies revealed that co-administration of low dose aspirin appears to greatly diminish, if not abolish, the GI protective effects of selective COX-2 inhibitors compared to non-selective NSAIDs22,23,50.

The VIGOR trial, which was primarily interested in the GI safety of rofecoxib compared to naproxen was actually the first to demonstrate the potential for increased cardiovascular risk of COX- 2 inhibitors. Low dose aspirin therapy was not allowed in this trial. Subjects taking 1000mg of naproxen a day had a 0.1% incidence of myocardial infarction while those taking rofecoxib 50 mg per day demonstrated a 0.4% incidence of the event21. Considering the patient population averaged almost 60 years of age and all had rheumatoid arthritis, a known risk factor for myocardial infarction51, the overall incidence of the event in both groups over an approximate 1 year period was not that high at all. In addition, the comparator drug, naproxen 500 mg taken twice per day, possesses profound but reversible antiplatelet activity52,53, possibly explaining some of the difference in myocardial infarction rate in the VIGOR trial. In fact a meta analysis comparing cardiovascular events following the chronic use of rofecoxib or the nonnaproxen NSAIDs ibuprofen, diclofenac or nabumetone revealed a slightly lower incidence of cardiovascular thrombotic events (relative risk = 0.79) in the rofecoxib group54.

While the CLASS trial did not show an increase in cardiovascular risk in patients taking celecoxib 400 mg twice per day compared to ibuprofen 800 mg three times per day and diclofenac 75 mg twice per day23, 20% of the patient population was consuming low dose aspirin, possibly reducing the number of events in the celecoxib group. In addition, unlike the VIGOR trial where the patients were all diagnosed with rheumatoid arthritis, the subjects in the CLASS trial overwhelmingly had osteoarthritis and thus may have possessed less cardiovascular risk. The published evaluation period was only 6 months and the comparator drugs were less COX-1 selective and thus possessed less antiplatelet activity than naproxen. In fact diclofenac while classified as a non-selective NSAID has reported COX-2 selectivity very close to that of celecoxib55-57. Table 1 summarizes the COX-2 selectivity of various NSAID-type drugs55-57.

It should be stressed that the absolute COX-2 selectivity values of NSAIDs can differ by more than 10-fold depending on the publication; although the rank order of selectivity appears to be fairly consistent. However, whole blood assays, which are the basis for the COX-2 selectivity rankings are all done in vitro. In the clinic, factors such as dose, the duration of dosing, the presence of cardiovascular disease and pharmacokinetic differences between individual molecules also play an important role in the cardiovascular safety of the entire NSAID class58,59. In the APC trial approximately 40% of the subjects had pre-existing hypertension43. As highlighted in the CABG trial, the most 'fragile patients' appear to be especially sensitive to the untoward cardiovascular effects of even a short-term exposure to at least two of the COX-2 selective drugs46. The vast majority of these patients had pre-existing angina, coronary artery atherosclerosis, hypertension and hyperlipidemia. In these patients, a 3 day exposure to intravenous parecoxib (40 mg at the completion of surgery then 20 mg every 12 h) followed by 7 days of oral valdecoxib (20 mg every 12h), or 3 days worth of intravenous placebo followed by oral valdecoxib (20mg every 12h for 7 days), increased the incidence of significant cardiovascular events including myocardial infarction, cardiac arrest, stroke and pulmonary embolism by 4-fold compared to subjects receiving placebo for 10 days.

Table 1. Relative COX-2 versus COX-1 selectivity of various NSAIDs based on respective IC^sub 50^ values in human whole blood assays. Value for lumiracoxib is an estimate based on the value of 515 published by Esser et al.57; and values for other COX-2 selective and non-selective NSAIDs published in this article and previous articles (Riendeau et al.56 and Warner et al.55)

As illustrated in Figure 3, our current understanding of the dual COX model does predict the potential for adverse cardiovascular outcomes among the COX-2 inhibitor class. The ability of highly selective COX-2 inhibitors to only block the COX-2 isoform of cyclooxygenase, would limit the production of the anti-aggregatory, vasodilatory prostacyclin molecule (prostaglandin I^sub 2^), while leaving COX-1 and its subsequent production of thromboxane A^sub 2^ functioning unopposed, leading to a pro-aggregatory, vasoconstrictive state41. The assumption that would follow is that the more highly selective an NSAID is for blocking COX-2 over COX- 1, the greater the risk of untoward thrombotic events. In addition, this model also explains the development or enhancement of hypertension in some patients on selective COX-2 inhibitors42-60, which could further exacerbate cardiovascular risk. Conversely, the lower the selectivity for COX-2, the more ulcerogenic an NSAID might be (and possibly the less cardiotoxic). Ketorolac (Toradol, Roche), which is approximately 400-fold more selective in blocking COX-1, is highly ulcerogenic.

At least one large case-control study supports the hypothesis of enhanced cardiotoxicity with greater COX-2 selectivity among the 'coxibs', with the risk of a cardiovascular event reported to be significantly higher for rofecoxib than celecoxib61. However, a comparison with historical rates of myocardial infarction in subjects administered placebo, demonstrated that both rofecoxib 50 mg and celecoxib 40\1mg twice per day resulted in significantly higher annualized myocardial infarction rates51. In addition, the myocardial infarction rate of lumiracoxib in TARGET which is reported to be the most COX-2 selective agent among the whole class, was no different than the combined incidence of the comparator NSAIDs naproxen and ibuprofen, although the incidence was almost double that of naproxen62. Yet to be published data with etoricoxib whose COX-2 selectivity is second to lumiracoxib reveals a similar trend; no apparent cardiovascular risk when compared to ibuprofen or diclofenac and an apparent cardiovasuclar risk when compared to naproxen63. These results taken together with those of VIGOR, TARGET and CLASS suggest that the COX-2 selectivity of the comparator NSAIDs may also be extremely important as to whether a particular coxib demonstrates cardiotoxicity.

It should be noted that others have proposed nonCOX enzymatic mechanisms for differences in reported cardiotoxicity of individual COX-2 selective and nonselective NSAIDs64. Additional research is needed to further explore these diverging results regarding the cardiovascular risk of the various COX-2 selective inhibitors, and in fact the entire NSAID class59. The recent suspension of the Alzheimer's Disease AntiInflammatory Disease Trial (ADAPT) partly due to an unexpected increased incidence of cardiovascular events in subjects consuming naproxen sodium 220 mg twice per day compared to placebo supports this statement24.

Now there are three?

While the dual COX model resolved many of the issues concerning differences between non-selective NSAIDs and highly selective COX-2 inhibitors, it still could not fully explain the pharmacologie actions of acetaminophen. Many of acetaminophen's actions resemble COX-2 selective inhibitors (analgesic effects, antipyretic effects and a relative lack of GI toxicity)65,66. However it lacks, or at very best possesses weak, antiinflammatory action; an important characteristic of both non-selective NSAIDs and the COX-2 selective drugs13,20. In addition, more than 50 years of clinical experience with this drug has revealed no appreciable anti-aggregatory or pro- aggregatory effects on platelets2,13,14,65,66.

As far back as 1972, Flower and Vane reported that acetaminophen was far more active in inhibiting COX activity in dog brain homogenates than in homogenates from the spleen67. These results led them to be the first to postulate the existence of more than one COX isoform. Others have since proposed a predominantly central mechanism of action for acetaminophen involving either central COX- 2 inhibition, the inhibition of a yet to be isolated COX variant termed COX-3 or the activation of descending serotonergic pathways in the brain and spinal cord65,66,68.

In 2002, Chandrasekharan and colleagues reported the isolation of a splice-variant of COX-1 mRNA found in highest concentrations in the cerebral cortex and heart of the dog, which they reported to be the everelusive COX-369. The messenger RNA that produced the COX-3 protein was derived from the same gene that coded for COX-1, except in COX-3 RNA, an intron made up of 90 nucleotides at or near the 5 prime end of the molecule was retained. The retention of this intron (which is normally cleaved prior to the final synthesis of the RNA) introduces the insertion of an additional 30 amino acids into the dog COX-3 molecule. It was postulated that these extra amino acids would alter the folding and subsequent enzymatic properties of this newly discovered COX type. In experiments performed by this group, it was demonstrated that in transfected insect cells, canine COX-3 protein was expressed and selectively, but weakly inhibited by acetaminophen, whereas transfected murine COX-1 or COX-2 was not acetaminophen sensitive69. In addition, other analgesic/antipyretic drugs that lacked significant anti-inflammatory activity such as phenacetin (which is metabolized to acetaminophen) and dipyrone, and classical NSAIDs with potent anti-inflammatory activity such as ibuprofen and diclofenac also displayed more potent inhibition of COX-3 than the other COX isoforms69 (see Table 2). Two other splice variants termed partial COX-1a and partial COX-1b, with amino acids 119-337 deleted were also isolated by this group. Enthusiasm mounted that the long sought after mechanism of acetaminophen's analgesic and antipyretic action had been discovered, in addition to a central mechanism for NSAID analgesics13,14,69.

Table 2. Dosages (g) of various analgesic/antipyretic drugs (IC^sub 50^ values) needed to reduce the activity of the various COX isoforms by 50%. Caffeine serves as a negative control. Lower IC^sub 50^ values reflect greater potency. Data from Chandrasekharan et al.69

Additional reports of COX-3 mRNA isolation in the central nervous systems of other animal species including human cerebral cortex and hippocampus subsequently appeared1'9'74. It should be noted that a number of scientists prefer calling this newly described protein COX- 1b or COX-1 variant (COX-1v), rather than COX-3, because the mRNA is encoded by the COX-1 gene, and other than the retained intron, the mRNA is indistinguishable from COX-1(75,76).

Both COX-1 and COX-3, but not COX-2 mRNA are expressed in the dorsal root ganglion in mice exhibiting experimental inflammatory pain, supporting that COX-1 and its splice-variant, like COX-2, may be inducible in certain types of pain and that at least in mice, COX- 1 and its variant might be the predominant species76. The hypothermie action of acetaminophen in mice correlates with a reduction of prostaglandin E^sub 2^ levels in their brains, and this effect is greatly attenuated in COX-1 knockout mice supporting a possible role of constitutive COX-3 in temperature regulation77. However, others have reported that it is a blockade of COX-2 or a COX-2 variant that explains the antipyretic action of acetaminophen78. In primary cultures of rat cerebral endothelial cells, a challenge with the known fever inducer, bacterial lipopolysaccharide (LPS), only increased the expression of COX-2 mRNA (not COX-1 or COX-3). In addition, both acetaminophen and the selective COX-2 inhibitor NS398, significantly reduced prostaglandin E^sub 2^ production in LPS-stimulated cultures78. A previous study also demonstrated that the deletion of the COX-2 gene and not the deletion of the COX-1 gene blocked the febrile response of LPS in mice79. An acetaminophen-sensitive COX-2 variant in fact has been reported to play a major role in the apoptosis of J774.2 cells induced by prolonged contact with diclofenac80. Others have demonstrated that the activity of acetaminophen is most pronounced in intact cell preparations where COX-2 in the presence of low concentrations of arachidonic acid or peroxide is the dominant isoenzyme65.

The molecular biology of human and murine COX-3 mRNA also throws 'a monkey wrench' into the clinical significance of COX-3. While canine COX-3 mRNA contains 90 additional nucleotides in the retained intron, the human COX-3 mRNA contains 94 nucleotides81,82. Since the genetic code is a triplicate code (3 nucleotides to form one amino acid), the retention of an intron containing 94 nucleotides without the removal of a single nucleotide prior to protein synthesis, would result in a frame shift in the reading of the downstream message, theoretically producing a protein that has no similarities to COX-1 or COX-2(72-81,82). Figure 4 illustrates a hypothetical frame shift with a retained intron of 94 nucleotides.

COX-1 and COX-2 are very similar in humans, both being approximately 600 amino acids in size and sharing approximately 65% amino acid identity83. In addition, COX-1 and COX-2 orthologs share approximately 70%-95% amino acid identity across vertebrate species83. The predicted COX-3 protein that would be synthesized in humans would possess a completely different protein sequence and would only be about 79 amino acids long due to a stop codon (UGA) on the messenger RNA81. The COX-1 splice variant retaining the 94- nucleotide intron was recently isolated as the predominant COX-3 messenger RNA species from human tissue84. Interestingly two minor subtypes of this variant were also isolated, that contained only 93 nucleotides in the retained intron. However, the protein they produced while able to catalyze the production of prostaglandins from arachidonic acid, did not exhibit differential sensitivity to acetaminophen84.

Studies on rat COX-3, whose message is also out of frame because the retained intron is 98 nucleotides, possesses an amino acid sequence bearing no structural relationship to rodent COX-1, and is only 122 amino acids long because the frame shift induces a UGA stop codon in the downstream message82. Somehow, this RNA avoids the nonsense decay pathway, which targets messenger RNAs with premature stop codons for degradation, and translates the truncated protein. This protein, however, is not acetaminophen sensitive and accordingly has been given the name cyclooxygenase variant protein (COVAP)82. Another study also showed that a functional COX-3 protein is not expressed in the rat85. While some have speculated on the potential significance of COX-3(86,87), it appears that at least in the human and the rodent, while putative COX-3 mRNA is present in a number of tissues, it does not express a functional, acetaminophen- sensitive COX protein88. In addition, the IC^sub 50^ concentration of acetaminophen needed to block COX-3 in canine in vitro preparations is very high (Table T), and is unlikely to be attained in the human hypothalamus where temperature regulation occurs, following therapeutic doses of acetaminophen14,89. It is still puzzling however, why a functional acetaminophen sensitive COX-3 protein would be expressed in the dog and not in the rat or the human.

Figure 4. Hypothetical sections of COX mRNA with and without retained introns

Top of the sc\hematic depicts a hypothetical section of RNA at its 5 prime end and the corresponding amino acid sequence produced. The middle of the schematic illustrates the same piece of messenger RNA with a retained intron of 90 nucleotides as in the dog. Only nucleotides 85 through 90 are shown. Note that other than the extra amino acids translated on the 5 prime end, the downstream message following the intron remains unchanged. However as shown in the bottom of this figure, a retained intron of 94 nucleotides as exists in the human, produces a frame shift in the RNA message down stream from the intron, resulting in a completely different amino acid sequence and a stop codon which terminates the amino acid sequence. Stop codons are UAA, UAG or UGA.

Conclusions

Significant scientific advances in our understanding of cyclooxygenase biology and the drugs that block these enzymes have occurred since Sir John Vane's initial discoveries. However, with these advances, comes set backs, as evidenced by the apparent cardiovascular toxicity of COX-2 inhibitors and a growing body of evidence refuting the existence in humans of an acetaminophen- sensitive splice variant of COX-1, given the name of COX-3.

Acknowledgement

Declaration of interest: Supported in part by a grant from the National Institute of Health DE09517.

The authors would like to thank Bridget Gallagher for her valuable assistance with the figures.

* Presented in part at the 83rd General Session of the International Association of Dental Research, Baltimore, MD, 9-12 March 2005

References

1. Vane JR. Inhibition of prostaglandin synthesis as a mechanism of action for aspirin like drugs. Nature 1971;231:232-5

2. Hersh EV, Moore PA, Ross GL. Over-the-counter analgesics and antipyretics: a critical assessment. Clin Ther 2000;22:500-48

3. Cooper SA, Needle SE, Kruger GO. Comparative analgesic potency of aspirin and ibuprofen. J Oral Maxillofac Surg 1977;35:898-903

4. Cooper SA, Engel J, Ladov M, et al. Analgesic efficacy of an ibuprofen-codeine combination. Pharmacotherapy 1982;2: 162-7

5. Cooper SA, Gelb SB, Cavalere MB, et al. An analgesic relative potency assay comparing ketoprofen and aspirin in postoperative dental pain. Adv Ther 1984; 1:410-8

6. Sunshine A, Olson NZ. Analgesic efficacy of ketoprofen in postpartum, general surgery, and chronic cancer pain. J Clin Pharmacol. 1988;28(Suppl 12):S47-S54

7. Forbes JA, Keller CK, Smith JW, et al. Analgesic effect of naproxen sodium, codeine, a naproxen-codeine combination and aspirin on the postoperative pain of oral surgery. Pharmacotherapy 1986;6:211-8

8. Forbes JA, Butterworth GA, Burchfield WH, Beaver WT. Evaluation of ketorolac, aspirin, and an acetaminophen-codeine combination in postoperative oral surgery pain. Pharmacotherapy 1990;10(Suppl):77S-93S

9. Ferriera SH, Vane JR. New aspects of the mode of action of nonsteroid anti-inflammatory drugs. Annu Rev Pharmacol 1974;14:57- 72

10. Brooks PM, Day RO. Nonsteroidal antiinflammatory drugs differences and similarities. New Engl J Med 1991;324:1716-25

11. Langman MJS, Weil J, Wainright P, et al. Risk of bleeding peptic ulcer associated with individual nonsteroidal antiinflammatory drugs. Lancet 1994;343:1075-8

12. Penderson AK, Fitzgerald GA. Dose-related kinetics of aspirin, presystemic acetylation of platelet cyclooxygenase. New Engl J Medl984;311:1206-ll

13. Warner TD, Mitchell JA. Cyclooxygenase-3 (COX-3): filling in the gaps toward a COX continuum? Proc Natl Acad Sci USA 2002;99:13371-3

14. Schwab JM, Schluesener HJ, Laufer S. COX-3: just another COX or the solitary elusive target of paracetamol. Lancet 2003;361:981- 2

15. Hla T, Neilson K. Human cyclooxygenase-2 DNA. Proc Natl Acad Sci USA 1992;89:7384-8

16. Fletcher BS, Kujubu DA, Perin DM, Herschman HR. Structure of the mitogen-inducible T1S10 gene and demonstration that the TISlO- encoded protein is a functional prostaglandin G/H synthase. J Biol Chem 1992;267:4338-44

17. O'Banion MK, Sadowski HB, Winn V, Young DA. A serum and glucocorticoid regulated 4-kilobase mRNA encodes a cyclooxygenase related protein. J Biol Chem 1991;266:23261-7

18. Xie W, Chipman JG, Robertson DL, et al. Expression of a mitogen-responsive gene encoding prostaglandin synthase is regulated by mRNA splicing. Proc Natl Acad Sci USA 1991;88:2692-6

19. Crofford IJ. COX-1 and COX-2 tissue expression: implications and predictions. J Rheumatol 1997;24(Suppl 149}:15-9

20. Moore PA, Hersh EV. Celecoxib and rofecoxib: the role of COX- 2 inhibitors in dental practice. J Am Dent Assoc 2001;132:451-6

21. Bombadier C, Laine L, Reicin A, et al. Comparison of upper gastrointestinal toxicity of rofecoxib and naproxen in patients with rheumatoid arthritis. New Engl J Med 2000;343:1520-8

22. Schitnizer 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: randomized controlled trial. Lancet 2004;364:665-74

23. Silverstein FE, Faich G, Goldstein JL, et al. Gastrointestinal toxicity with celecoxib vs. nonsteroidal antiinflammatory drugs for osteoarthritis and rheumatoid arthritis : the CLASS study - a randomized controlled trial. J Am Med Assoc 2000;284:1247-55

24. Food and drug administration joint meeting with the Arthritis Advisory Committee and the Drug Safety and Risk Management Advisory Committee [online]. Available from http://www.fda. gov/ohrms/ dockets/ac/05/slides/2005-4090sl.htm [accessed 1 May 2005]

25. Ehrlich K, Sicking C, Respondek M, Peskar BM. Interaction of cyclooxygenase isoenzymes, nitric oxide, and afferent neurons in gastric mucosal defense in rats. J Pharmacol Exp Ther 2004;308:277- 83

26. Wallace JL, McKnight W, Reuter BK, Vergnolle N. NSAIDinduced gastric damage in rats: requirement for inhibition of both cyclooxygenase 1 and 2. Gastroenterology 2000;119:706-14

27. Gretzer B, Maricic N, Respondek M, et al. Effects of specific inhibition of cyclo-oxygenase-1 and cyclo-oxygenase-2 in the rat stomach with normal mucosa and after acid challenge. Br J Pharmacol 2001;132:1565-73

28. Tanaka A, Araki H, Komoike Y, et al. Inhibition of both COX- 1 and COX-2 is required for development of gastric damage in response to nonsteroidal antiinflammatory drugs. J Physiol (Paris) 2001;95:21-7

29. Brater CD, Harris C, Redfern JS, Gertz BJ. Renal effects of COX-2-selective inhibitors. Am J Nephrol 2001;21:1-15

30. Curtis SP, Ng J, Yu Q, et al. Renal effects of etoricoxib and comparator nonsteroidal anti-inflammatory drugs in controlled clinical trials. Clin Ther 2004;26:70-83

31. Morrison BW, Christensen S, Yuan W, et al. Analgesic efficacy of the cyclooxygenase inhibitor rofecoxib in post-dental surgery pain: a randomized, controlled trial. Clin Ther 1999)21:943-53

32. Malmstrom K, Fricke JR, Kotey P, et al. A comparison of rofecoxib versus celecoxib after dental surgery: a single-center, randomized, double-blind, placebo- and active-comparatorcontrolled, parallel-group, single-dose study using the dental impaction pain model. Clin Ther 2002;24:1549-60

33. Daniels SE, Grossman EH, Kuss ME, et al. A double-blind, randomized comparison of intramuscularly and intravenously administered parecoxib sodium versus ketorolac and placebo in a post- oral surgery pain model. Clin Ther 2001,23:1018-31

34. Malmstom K, Kotey P, Coughlin H, Desjardins PJ. A randomized, double-blind, parallel-group study comparing the analgesic effect of etoricoxib to placebo, naproxen sodium, and acetaminophen with codeine using the dental impaction pain model. Clin J Pain 2004;20:147-55

35. Schnitzer TJ, Fricke Jr JR, Gitton X, et al. Lumiracoxib in the treatment of osteoarthritis, rheumatoid arthritis and acute postoperative dental pain: results of three dose-response studies. Curr Med Res Opin 2005;21:151-61

36. Chang DJ, Fricke JR, Bird SR, et al. Rofecoxib versus codeine/ acetaminophen in postoperative dental pain: a double-blind, randomized, placebo- and active comparator-controlled trial. ClinTher 2001;23:1446-55

37. Chang DJ, Desjardins PJ, Bird SR, et al. Comparison of rofecoxib and a multidose oxycodone/acetaminophen regimen for the treatment of acute pain following oral surgery: a randomized controlled trial. Curr Med Res Opin 2004;20:939-49

38. Daniels SE, Desjardins PJ, Talwalker S, et al. The analgesic efficacy of valdecoxib vs. oxycodone/acetaminophen after oral surgery. J Am Dent Assoc 2002;133:611-21

39. Malmstrom K, Ang J, Fricke JR, et al. The analgesic effect of etoricoxib relative to that of two opioid-acetaminophen containing analgesics: a randomized, double-blind, placebo and active comparator controlled single-dose study in acute dental impaction pain. Curr Med Res Opin 2005;21:141-9

40. Cooper SA. Five studies on ibuprofen for postsurgical dental pain. Am J Med 1984;77:70-7

41. Patrono C, Coller B, FitzGerald GA, et al. Platelet-active drugs: the relationships among dose, effectiveness, and side effects: the Seventh ACCP Conference on Antithrombotic and Thrombolytic Therapy. Chest 2004;126(Suppl 3):234S-64S

42. Bresalier RS, Sandier RS, Quan H, et al. Cardiovascular events associated with rofecoxib in colorectal adenoma chemoprevention trial. New Engl J Med 2005;352:1092-102

43. Solomon SD, McMurray JJV, Pfeffer MA, et al. Cardiovascular risk associated with celecoxib in a clinical trial for colorectal adenoma prevention. New Engl J Med 352)2005:1071-80

44. Merck announces voluntary worldwide withdrawal of VIOXX [online]. Available from htpp://www.vioxx.com/rofecoxib/ vioxx/ consumer/index.jsp [accessed 1 October 2004]

45. 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; 125:1481-92

46. Nussmeier NA, Whelton AA, Brown MT, et al. Complications of the COX-2 inhibitors parecoxib and valdecoxib after cardiac surgery. New Engl J Med 3\12;2005:1081-91

47. FDA announces important changes and additional warnings for COX-2 selective and non-selective non-steroidal antiinflammatory drugs (NSAIDs) [online]. Available from http:// www.fda.gov/cder/ drug/advisory/COX2.htm [accessed 7 May 2005]

48. Reines SA, Block GA, Morris JC, et al. Rofecoxib: no effect on Alzheimer's disease in a 1-year, randomized double-blind, controlled study. Neurology 2004;62:66-71

49. Juni P, Nartey L, Reichenbach S, et al. Risk of cardiovascular events and rofecoxib:cumulative meta-analysis. Lancet 2004;364:2021-9

50. Eaine L, Mailer ES, Yu C, et al. Ulcer formation with low- dose enteric-coated aspirin and the effect of COX-2 selective inhibition: a double-blind trial. Gastroenterology 2004; 127: 395- 402

51. Mukherjee D, Nissen SE, Topol EJ. Risk of cardiovascular events associated with selective COX-2 inhibitors. J Am Med Assoc 2001;286:954-9

52. Knijff-Dutmer EAJ, Kalsbeek-Batenburg EM, Koerts J, van de Laar FJ. Platelet function is inhibited by non-selective nonsteroidal anti-inflammatory drugs but not by cyclo-oxygenase2- selective inhibitors in patients with rheumatoid arthritis. Rheumatology 2002;41:458-61

53. Capone ML, Tacconelli S, Sciulli MG, et al. Clinical pharmacology of platelet, monocyte, and vascular cyclooxygenase inhibition by naproxen and low-dose aspirin in healthy subjects. Circulation 2004; 109:1468-71

54. Konstam MA, Weir MR, Reicin A, et al. Cardiovascular thrombotic events in controlled, clinical trials of rofecoxib. Circulation 2001; 104:2280-8

55. Warner TD, Giuliano F, Voinovic 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 USA 1999;96:7563-8

56. Riendeau D, Percival D, Brideau C, et al. Etoricoxib (MK0663): predinical profile and comparison with other agents that selectively inhibit cyclooxygenase-2. J Pharmacol Exp Ther 2001;296:558-66

57. Esser R, Berry C, Du Z, et al. Preclinical pharmacology of lumiracoxib: a novel selective inhibitor of cyclooxygenase-2. Br J Pharmacol 2005; 144:538-50

58. Wong D, Wang M, Cheng Y, Fitzgerald GA. Cardiovascular hazard and non-steroidal anti-inflammatory drugs. Curr Opin Pharmacol 2005;5:204-10

59. Davies NM, Jamali F. COX-2 selective inhibitors cardiac toxicity: getting to the heart of the matter. J Pharm Pharmaceut Sci 2004;7:332-6

60. Fitzgerald GA. Coxibs and cardiovascular disease. New Engl J Med2004;351:1709-11

61. Kimmel SE, Berlin JA, Reilly M, et al. Patients exposed to rofecoxib and celecoxib have different odds of nonfatal myocardial infarction. Ann Intern Med 2005; 142:157-64

62. Farkoush ME, Kirshner H, Harrington RA, et al. Comparison of lumiracoxib with naproxen and ibuprofen in the Therapeutic Arthritis Research and Gastrointestinal Event Trial (TARGET), cardiovascular outcomes: randomized controlled trial. Lancet 2004;364:675-84

63. Ramey DR, Watson DJ, Yu C, et al. The incidence of upper gastrointestinal adverse events in clinical trials of etoricoxib vs. non-selective NSAIDs: an updated combined analysis. Curr Med Res Opin 2005;21:715-22

64. Walter MF, Jacob RF, Day CA, et al. Sulfone COX-2 inhibitors increase susceptibility of human LDL and plasma to oxidative modification: comparison to sulfonamide COX-2 inhibitors and NSAIDs. Atherosclerosis 2004; 177:235-43

65. Graham GG, Scott KF. Mechanisms of action of paracetamol and related analgesics. Inflammopharmacology 2003;! 1:401-13

66. Bolting RH. Mechanisms of action of acetaminophen: is there a cyclooxygenase-3. Clin Infect Dis 2000;31:5202-10

67. Flower RJ, Vane JR. Inhibition of prostaglandin synthetase in brain explains the anti-pyretic activity of paracetamol (4acetamidophenol). Nature 1972;246:410-1

68. Willoughby DA, Moore AR, Collville-Nash PR. COX-1, COX-2, COX- 3 and the future treatment of chronic inflammatory disease. Lancet 2000;355:646-8

69. Chandrasekharan NV, Dai H, Ruos LT, et al. COX-3, a cyclooxygenase-1 variant inhibited by acetaminophen and other analgesic/antipyretic drugs: cloning, structure and expression. Proc Natl Acad Sci USA 2002;99:13926-31

70. Shaftel SS, Olschowka JA, Hurley SD, et al. COX-3: a splice variant of cyclooxygenase-1 in mouse neural tissue and cells. Mol Brain Res 2003; 119:213-5

71. Kis B. Snipes A. Bari F. Busija DW. Regional distribution of cyclooxygenase-3 mRNA in the rat central nervous system. Mol Brain Res 2004; 126:78-80

72. Dinchuk JE, Liu RQ, Trzaskos JM. COX-3: in the wrong frame in mind. Immunol Lett 2003;86:121

73. Kis B, Snipes JA, Isse T, et al. Putative cyclooxygenase-3 expression in rat brain cells. J Cereb Blood Flow Metab 2003;23:1287- 92

74. Cui JG, Kuroda H, Chandrasekharan NV, et al. Cyclooxygenase- 3 gene expression in Alzheimer hippocampus and in stressed human neural cells. Neurochem Res 2004;29:1731-7

75. Davies NM, Good RL, Roupe KA, Yanez JA. Cyclooxygenase-3: axiom, dogma, anomaly, enigma or splice error? - not as easy as 1, 2, 3. J Pharm Pharmaceut Sci 2004;7:217-26

76. Dou W, Jiao Y, Goorha S, et al. Nociception and the differential expression of cyclooxygenase-1 (COX-1), the COX-1 variant retaining intron-1 (COX-Iv), and COX-2 in mouse dorsal root ganglia (DRG). Prostaglandins Other Lipid Mdiat 2004; 74:29-43

77. Ayoub SS, Botting RM, Goorha S, et al. Acetaminopheninduced hypothermia in mice is mediated by a prostaglandin endoperoxide synthase 1 gene-derived protein. Proc Natl Acad Sci USA 2004;101:11165-9

78. Kis B, Snipes JA, Simandle SA, Busija DW. Acetaminophensensitive prostaglandin production in rat cerebral endothelial cells. Am J Physiol Regul Integr Comp Physiol 2005;288:R897902

79. Li S, Wang Y, Matsumura K, et al. The febrile response to lipopolysaccharide is blocked in cyclooxygenase-2(-/-) but not in cyclooxygenase-l(-/-) mice. Brain Res 1999;825:86-94

80. Simmons DL, Bolting RM, Robertson PM, et al. Induction of an acetaminophen-sensitive cyclooxygenase with reduced sensitivity to nonsteroid antiinflammatory drugs. Proc Natl Acad Sci USA 1999;96:3275-80

81. Schwab JM, Beiter T, Linder JU, et al. COX-3 - a virtual pain target in humans? FASEB J 2003;17:2174-5

82. Snipes JA, Kis B, Shelness GS, et al. Cloning and characterization of cyclooxygenase-Ib (putative COX-3) in rat. J Pharmacol Exp Ther2005;313:668-76

83. Chandrasekharan NV, Simmons DL. The cyclooxygenases. Genome Biology 2004;5:241

84. Qin N, Zhang S, Reitz T, et al. Human COX-3? Cloning and characterization of a splicing variant of COX-1. Program No. 287.13. Abstract viewer/itinerary planner. Washington (DC): Society for Neuroscience, 2004. Available from: http://sfn. scholarone.com [accessed 3 June 2005]

85. Warner TD, Vojnovic I, Giuliano F, et al. Cyclooxygenases 1, 2, 3 and the production of prostaglandin I^sub 2^: investigating the activities of acetaminophen and cyclooxygenase-2 selective inhibitors in rat tissues. J Pharmacol Exp Ther 2004;310:642-7

86. Simmons DL. Variants of cyclooxygenase-1 and their roles in medicine. Thromb Res 2003;110:265-8

87. Dotting R. COX-1 and COX-3 inhibitors. Thromb Res 2003;110:269-72

88. Kis B, Snipes JA, Busija DW. Acetaminophen and the COX-3 puzzle: sorting out facts, fictions and uncertainties. J Pharmacol Exp Ther 2005;313 [doi:10.1124/jpet.105.085431]

89. Berenbaum F. COX-3: fact or fancy? Joint Bone Spine 2004;71:451-3

CrossRef links are available in the online published version of this paper: http://www.cmrojournal.com

Paper CMRO-3018_5, Accepted for publication: 08 June 2005

Published Online: 05 July 2005

doi:10.1185/030079905X56367

Elliot V. Hersh(a), Edward T. Lally(b) and Paul A. Moore(c)

a Professor of Pharmacology, University of Pennsylvania School of Dental Medicine, Philadelphia, PA, USA

b Professor of Pathology, University of Pennslyvania School of Dental Medicine, Philadelphia, PA, USA

c Professor of Pharmacology and Dental Public Health, University of Pittsburgh School of Dental Medicine, Pittsburgh, PA, USA

Address for correspondence: Dr Hersh DMD, MS, PhD, Professor of Pharmacology, Department of Oral Surgery and Pharmacology, University of Pennslyvania School of Dental Medicine, 240 South 40th Street, Philadelphia, PA 19104-6030, USA. Tel.: +1 215 898 9686; Fax: +1 215 746 8891; email evhersh@pobox.upenn.edu

Copyright Librapharm Aug 2005

Source: Current Medical Research and Opinion

More News in this Category