The rate-limiting enzyme for prostanoid synthesis cyclooxygenase-2 (COX-2) has been implicated in the basic mechanisms of several brain diseases, including stroke, multiple sclerosis and neurodegenerative diseases.1 The approval by the Food and Drug Administration (FDA) of highly selective COX-2 inhibitors for the treatment of pain and rheumatoid arthritis (RA) raised the possibility that these agents could also be used in the treatment of neurological diseases including stroke. However, the occurrence of serious cardiovascular complications in patients receiving COX-2 inhibitors has led to the recent withdrawal from the market of a popular COX-2 inhibitor and has called for a re-evaluation of the therapeutic potential of these drugs. In this article, we briefly review the role of COX-2 in ischemic brain injury and re-examine the validity of the COX-2 pathway as a therapeutic target for ischemic stroke.
From Arachidonic Acid to Prostanoids: COX, Isomerases, and Prostanoid Receptors
COX enzymes catalyze the conversion of arachidonic acid into prostaglandin H2 (PGH2; Figure).2 Arachidonic acid, produced by the breakdown of membrane phospholipids, is metabolized by COX into PGH2 in a 2-step reaction in which the free radical superoxide is also produced.2 Three isoforms of COX have been described. COX-1 is present in most cells and is involved in normal cellular physiology, such as gastric secretion and platelet function.2 COX-2 is expressed constitutively in some organs, such as brain, but is markedly upregulated by a wide variety of stimuli, most notably inflammatory mediators.2 COX-3, a splice variant of COX-1, is highly sensitive to inhibition by acetaminophen and is most abundant in heart and brain.3 The COX reaction product PGH2 is converted into 5 different prostanoids by a corresponding number of isomerases (Figure), the distribution of which is cell and organ specific.2 Thus, PGH2 can give rise to different prostanoids depending on the isomerases available to metabolize it. Adding further to the complexity of the COX system is the fact that each prostanoid acts on multiple receptors with diverse signaling profiles and often opposing biological actions (Figure).
Cell stressors, such as ischemia, activate phospholipase A2, which cleaves arachidonic acid from membrane phospholipids. Arachidonic acid is metabolized by COX-1 and COX-2 into PGH2, which is metabolized further by distinct isomerases into 5 prostanoids. A third isoform of COX (COX-3) has also been described but is not shown here. Each prostanoid acts on ≥1 specific G-protein–coupled receptors. Because of the cellular and subcellular compartmentation of COX isoforms and isomerases, the prostanoids linked to the enzymatic activity of COX-1 or COX-2 vary from cell type to cell type. The deleterious effects of COX-2 in ischemic brain injury are mediated by PGE2, possibly via activation of EP1 receptors. On the other hand, it has been hypothesized that the cardiovascular complications of COX-2 inhibitors derive from the fact that COX-2 inhibition attenuates endothelial PGI2 production, leaving COX-1–dependent TXA2 synthesis unopposed. The predominance of TXA2-dependent effects leads to vasoconstriction, platelet aggregation, and smooth muscle proliferation, factors that favor atherogenesis and thrombosis. PGD2 indicates prostagladin D2; PGF2α, prostagladin F2α; IP, prostacyclin receptor; TPα, thromboxane receptor α; TPβ, thromboxane receptor β; DP1, prostagladin receptor 1; DP2, prostagladin receptor 2; FPα, prostagladin receptor α; FPβ, prostagladin receptor β; PGJ2, prostagladin J2.
COX-2 and the Brain: Roles in Models of Ischemic Injury
In the normal brain, COX-2 is expressed predominantly in dendritic profiles of glutamatergic neurons,4 a localization consistent with its role in synaptic function and neurovascular regulation.5,6 In rodents as in humans, cerebral ischemia upregulates COX-2 expression in neurons, glia, vascular cells, and in inflammatory cells invading the ischemic brain.7–10 Inhibition of COX-2 attenuates ischemic injury after middle cerebral artery occlusion with a relatively wide therapeutic window (6 to 24 hours).7,11 In addition, ischemic injury is attenuated in COX-2–deficient mice and is exacerbated in transgenic mice overexpressing COX-2.12,13 The mechanisms of the deleterious effects of COX-2 are multifactorial. On the one hand, COX-2 reaction products contribute to glutamate excitotoxicity,12,14 a major factor in the Ca2+ dysregulation initiating the ischemic cascade.15 On the other hand, COX-2 contributes to the deleterious effects of the inflammatory reaction involving the ischemic brain.16 Therefore, COX-2 is an attractive target for stroke therapy.
COX-2 Mediated Neurotoxicity: Quest for Downstream Effectors
Recent studies have focused on the specific mediators of the deleterious effects of COX-2 in ischemic brain injury. COX-2 produces prostanoids and free radicals, both of which could mediate tissue damage. Data in excitoxicity models suggest that prostanoids, mainly prostaglandin E2 (PGE2), and not free radicals are the pathogenic factors mediating brain injury.17 PGE2 activates 4 receptors termed EP1 through EP4 (Figure). Although limited data are available on the role of EP receptors in cerebral ischemia, evidence suggests that EP2 receptors are protective.18 Therefore, EP2 receptors are unlikely to mediate the neurotoxicity of COX-2. Preliminary evidence suggests that activation of EP1 receptors is deleterious in models of excitotoxicity and oxygen–glucose deprivation.19,20 Therefore, activation of EP1 receptors by PGE2 may be a factor responsible for the toxicity exerted by COX-2, but further studies are required to establish this point more firmly.
From Bench to Bedside: COX-2 Inhibitors in Clinical Practice
In the 1970s, Vane suggested that the biological effects of nonsteroidal anti-inflammatory drugs (NSAIDS) were mediated by COX.21 After the discovery of the different isoforms of COX, studies revealed that the analgesic and anti-inflammatory effects of these drugs were mediated by their ability to block COX-2.22 However, nonselective NSAIDS might be harmful because they also block COX-1, leading to altered gastrointestinal function, mucosal ulceration, pain, and bleeding.23 The development of selective COX-2 inhibitors promised to relieve pain and inflammation without the adverse events associated with COX-1 inhibition.24 In the United States, celecoxib and rofecoxib were the first COX-2 inhibitors approved for use by the FDA.24 Celecoxib was labeled for treatment of RA and rofecoxib for treatment of acute pain and menstrual pain. COX-2 inhibitors were studied rigorously to determine whether they provided anti-inflammatory and analgesic effects without gastrointestinal complications. In a study of patients with RA, Celecoxib (100 to 400 mg BID) was effective, with lower incidence of endoscopic ulcers compared with naproxen.25 Another study in patients with RA and osteoarthritis examined whether celecoxib was associated with a lower incidence of significant gastrointestinal complications and other adverse effects compared with conventional NSAIDS.26 Celecoxib at high doses (400 mg BID) was associated with a lower incidence of gastrointestinal side effects and other complications compared with the NSAIDS ibuprofen and diclofenac.26 Similarly, rofecoxib was shown to have a favorable gastrointestinal profile when compared with naproxen.27,28 The successes of the COX-2 inhibitors led to the popularity of these drugs and billions of dollars in sales per year.29 Was there an ominous adverse event signal that had been ignored?
Trouble in Paradise: Cardiovascular Complications of COX-2 Inhibitors
Coincident with the FDA approval of refecoxib and celecoxib, FitzGerald et al reported that these drugs suppress the formation of prostacyclin (PGI2), leaving the production of thromboxane A2 (TXA2) unaltered.30 PGI2 is a key COX product in the endothelium that inhibits platelet aggregation, causes vasodilatation, and prevents proliferation of vascular smooth muscle cells.31 Evidence was also provided that PGI2 production in vivo was COX-2 dependent, possibly through COX-2 induction in endothelial cells by shear stress.31 In contrast to PGI2, the COX-1–derived prostanoid TXA2, causes platelet aggregation, vasoconstriction, and vascular proliferation.31 FitzGerald et al speculated that suppression of COX-2–dependent formation of PGI2 by the COX-2 inhibitors left TXA2 generation unopposed, promoting vasoconstriction, thrombosis, and atherogenesis.32 A number of publications began to surface suggesting that the COX-2 inhibitors might be associated with an increased risk of cardiovascular events. For example, a critical review of the Vioxx Gastrointestinal Outcomes Research Study (VIGOR), the Celecoxib Long-term Arthritis Safety Study (CLASS), and 2 smaller trials showed that the relative risk of a thrombotic cardiovascular event, including myocardial infarction, ischemic stroke, and transient ischemic attack, with rofecoxib treatment was ≈2.4× compared with naproxen (P=0.002).33 In the Tennessee Medicaid program (TennCare) study, high-dose rofecoxib users were ≈1.7× more likely than nonusers to have coronary heart disease. There was no evidence of increased risk at doses of ≤25 mg of rofecoxib.34 Other analyses had raised similar concerns about rofecoxib,35 and one study suggested a higher risk of admission for congestive heart failure in rofecoxib users and nonselective NSAID users, but not with celecoxib, relative to non-NSAID controls.36 The safety of parecoxib and valdecoxib in relation to serious adverse events has also been challenged,37 whereas a large trial comparing lumiracoxib with naproxen and ibuprofen suggested that lumiracoxib might be safe from a cardiovascular standpoint.38,39 On September 30, 2004, Merck, the manufacturer of rofecoxib, withdrew the drug from the market because of an excess risk of myocardial infarctions and strokes.32,40 This action occurred after the results of the Adenomatous Polyp Prevention on Vioxx (APPROVe) study, a study to determine the effect of rofecoxib on benign sporadic colon adenomas. In APPROVe, there was a significant 3.9-fold increase in the incidence of serious thromboembolic adverse events in the group receiving 25 mg of rofecoxib compared with placebo, and the incidence of myocardial infarction and thrombotic stroke diverged progressively after ≥1 year of treatment.32 Is there a unifying explanation for this troublesome cardiovascular event profile with administration of at least certain COX-2 inhibitors? One possibility is that the depression of PGI2 formation caused by the inhibitors led to elevation of blood pressure, accelerated atherogenesis, and exaggerated thrombotic response to atherosclerotic plaque rupture.32
Conclusions
The COX-2 pathway is a valuable therapeutic target for ischemic brain injury. However, the available clinical and experimental evidence suggests that, although COX-2 inhibitors are able to attenuate injury in stroke models, they also produce an unbalance in prostanoid synthesis that promotes deleterious vascular effects. Indeed, clinical trials have demonstrated that chronic inhibition of COX-2 increases the incidence of serious thromboembolic complications that would be devastating in patients with stroke. Therefore, inhibition of the COX-2 pathway with most of the drugs available today does not seem a viable option for stroke treatment. COX-2 inhibitors with a safer cardiovascular profile, such as lumiracoxib, and third-generation inhibitors41 might prove to be more suitable. In addition, new therapeutic strategies targeting the factors mediating the damage downstream of COX-2 offer great promise. These approaches provide the opportunity to block the specific receptors mediating COX-2–dependent neurotoxicity without altering the homeostatic balance between COX-2–derived prostanoids. These novel clinical and experimental approaches, if successful, may offer stroke patients powerful new tools to ameliorate brain damage and improve their functional outcome.
Acknowledgments
This work was supported by National Institutes of Health (NIH) grants HL18974 (C.I.), NS38252 (C.I.), NS33430 (P.B.G.), and AG17934 (P.B.G.). C.I. is the recipient of a Javits award from NIH/National Institute of Neurological Disorders and Stroke.
- Received November 29, 2004.
- Accepted December 1, 2004.
References
Jump to
- Article
- From Arachidonic Acid to Prostanoids: COX, Isomerases, and Prostanoid Receptors
- COX-2 and the Brain: Roles in Models of Ischemic Injury
- COX-2 Mediated Neurotoxicity: Quest for Downstream Effectors
- From Bench to Bedside: COX-2 Inhibitors in Clinical Practice
- Trouble in Paradise: Cardiovascular Complications of COX-2 Inhibitors
- Conclusions
- Acknowledgments
- References
- Figures & Tables
- Info & Metrics
- eLetters
This Issue
Article Tools
Share this Article
- The Janus Face of Cyclooxygenase-2 in Ischemic Stroke
